From Casetext: Smarter Legal Research

Valencia v. Franklin County Water District

United States District Court, E.D. California
Jan 5, 2011
761 F. Supp. 2d 1007 (E.D. Cal. 2011)


rejecting defendant's challenge to expert's opinion that relied on data and information as unreliable

Summary of this case from Victorino v. FCA U.S. LLC



January 5, 2011.

Brett Lee Runyon, Heather Sharon Cohen, Michael Edward Lehman, Mardero-sian, Runyon, Cercone, Lehman Armo, Fresno, CA, Ricardo Echeverria, Michael John Bidart, Shernoff Bidart Echeverria, LLP, Claremont, CA, Thomas Vincent Girardi, Stephen G. Larson, Girardi and Keese, Los Angeles, CA, Jack Silver, Law Office of Jack Silver, Santa Rosa, CA, for Plaintiffs.

Joseph A. Salazar, Jr., Yamin Thuzar Maung, Lewis Brisbois Bisgaard and Smith LLP, Carissa Marie Beecham, Best Best Krieger, LLP, Sacramento, CA, Jeffrey Frank Oneal, Thomas S. Brazier, LaMore Brazier Riddle and Giampaoli, Randall C. Creech, Creech Liebow Kraus, San Jose, CA, Robert Harry Greenfield, Greenfield, Hardy, El Dorado Hills, CA, Jan A. Greben, Danielle Lauren De Smeth, Greben Associates, Santa Barbara, CA, Terry L. Allen, Berliner Cohen, Merced, CA, Eugene Tanaka, Melanie Donnelly, Best Best Krieger LLP, David Allan Gifford, Gary T. Drummond, Steven, Drummond Gifford, Walnut Creek, CA, Glenn Barger, Chapman, Glucksman, Dean, Roeb Barger, Los Angeles, CA, Stephen E. Carroll, McCor-mick Barstow Sheppard Wayte and Car-ruth LLP, Scott D. Laird, Timothy Jones, John P. Kinsey, Jones Helsley PC, Fresno, CA, Donald Evan Sobelman, John F. Barg, Kathryn L. Oehlschlager, R. Morgan Gilhuly, Stephen C. Lewis, Barg Coffin Lewis and Trapp LLP, San Francisco, CA, Eric Martin Steinle, for Defendants.

Michael G. Marderosian, Marderosian, Runyon, Cercone, Lehman Armo, Fresno, CA. Article 49: Untitled



This lawsuit arises out of a now-closed cooling tower manufacturing facility (the "BAC site") that pressure treated wood and was operated by entities formerly owned by the BAC Defendants ("BAC"). Plaintiffs, current or former residents of residential neighborhoods ("Beachwood") near the BAC Site, allege that two contaminants from the BAC Site migrated from the treating area via groundwater, surface water, surface soil, private well, and air

The BAC Defendants include four separate entities: (1) Merck Co; (2) Amsted Industries, Inc.; (3) Baltimore Aircoil Company, Inc.; and (4) Track Four, Inc.

The former BAC Site is located at 3058 Beachwood Drive, Merced, California.

pathways to locations where plaintiffs were exposed to them. Also named as defendants are various municipalities, water districts, and developers, including the Franklin County Water District, Merced Irrigation District, the City and County of Merced, and the Meadowbrook Water District.

The BAC Defendants motions to summarily adjudicate the surface water and surface soil pathways are decided by separate Memorandum Decision.

All of these defendants have separately moved for partial summary judgment on grounds that Plaintiffs failed to demonstrate exposure, i.e., satisfy their "Phase 1" burden.

Before the court for decision is BAC Defendants' motion to summarily adjudicate Plaintiffs' tenth, twelfth, thirteenth, and fourteenth causes of action. According to Defendants, Plaintiffs have failed to present any admissible evidence of actual exposure to contaminants from the BAC site, via any pathway, which was required under "Phase 1" of the Court's "Order Modifying Scheduling Conference Order." (Doc. 540.) The first phase of discovery was to focus on "whether contaminants from the former [] BAC Site, Franklin County Water District or the April 2006 Flood have ever reached any location where plaintiffs could have been exposed to them, and if so, when such contaminants arrived, how such contaminants arrived at the location, how long they were present, and at what levels they were present." (Id. at 1:14-1:28.) By this motion, the BAC Defendants assert that Plaintiffs have not met their "Phase 1" or "general exposure" burden, entitling them to partial summary judgment.

Plaintiffs opposed the motion on July 1, 2010, submitting over 3,000 pages of documentation and 168 exhibits. Plaintiffs' opposition is founded on their claim that they have presented "substantial evidence" in the form of expert opinion and analysis to show that contaminants migrated from the BAC facility to Plaintiffs' homes and/or properties. Plaintiffs further argue that the BAC Defendants' motions to exclude certain expert testimony fail because their criticisms go to its weight, not its admissibility.

Oral argument on the BAC Defendants' motions was held on October 6, 7, 13, 14, and 15, 2010, during which the parties presented argument and evidence relevant to the Phase 1 pathway exposure issues. The parties were permitted to introduce expert testimony on key scientific issues, namely the methodologies and assumptions used to model contamination via the groundwater and air pathways. The testifying experts were examined by counsel for Plaintiffs and Defendants, as well as the Court.

The parties submitted final supplemental briefs on the pathway exposure issues October 19th and 20th, 2010. (Docs. 864-65, 867.)

On October 22 and November 1, 2010, the parties were advised the Court intended to appoint independent experts under Fed.R.Evid. 706 to assist in the understanding and resolving the complex scientific disputes over groundwater and air modeling. The analysis included determining concentration levels of hexavalent chromium and/or arsenic in Meadowbrook Well No. 2 and surrounding areas (i.e., monitor networks and private wells), as well as the assumptions/calculations underlying the air model. The Fed.R.Evid. 706 Experts were appointed on October, 26. 2010 (Kenneth D. Schmidt, Ph.D. — groundwater) and November 1, 2010

(Chatten Cowherd Jr., Ph.D. and Richard Countess, Ph.D. — air). The parties and the Court jointly prepared a list of questions for the experts, which were transmitted to the experts in early November. The independent expert reports were submitted to the Court on November 22, 2010.

A formal and comprehensive nominating process, in which the parties fully participated, was followed in each instance.

Drs. Countess and Cowherd prepared a joint expert report. (Doc. 942.)

The Fed.R.Evid. 706 experts were examined by the Court and parties on December 2, 3, and 15, 2010. Rebuttal testimony was permitted on a limited basis.

On December 2, 3, 15, and 28, 2010, Plaintiff's groundwater and air experts, Mr. Douglas Bartlett and Ms. Camille Sears provided rebuttal testimony to the specific criticisms raised in the Fed.R.Evid. 706 expert reports.

The Rule 56 motions were submitted for decision following summation arguments on December 28 and 29, 2010.


The following background facts are taken from the parties' submissions in connection with the motions and other documents on file in this case. The parties have filed various objections to the evidence submitted in support of their adversary's motion for summary judgment and to exclude expert testimony. In deciding the motions, no inadmissible evidence was considered, including improper expert opinion under Rules 26 and 37 of the Federal Rules of Civil Procedure. Evidentiary rulings were made in open court as expert testimony was presented.

The facts underlying this case are summarized in the Court's previous Memorandum Decisions in this case, filed on November 13, 2008, May 18, 2009, and July 15, 2009, in brief: approximately 2,100 Plaintiffs seek damages relating to two occurrences: (1) an April 2006 flood; and (2) alleged long-term contamination released from the former site of a cooling tower manufacturing facility operated by entities formerly owned by the BAC Defendants. The now-closed facility, which is the alleged source of contamination, is located in Merced, California, approximately 1,600 feet southwest from Meadowbrook Well No. 2 ("MWC-2"), which was the primary well supplying domestic water to the Beachwood community, until it was taken out of service in March 2008. Plaintiffs allege that MWC-2, which was operated by the Meadowbrook Water District, was contaminated by hexavelant chromium and total chromium in the well and from the aquifer, which drew on the contaminated plume during times of high water production, and exposed Plaintiffs to contaminants.

See, e.g., Valencia v. Merck Co., 2009 WL 2136384 (E.D. Cal. July 15, 2009); Abarca v. Franklin County Water Dist., 2009 WL 1393511 (E.D. Cal. May 18, 2009); Affholter v. Franklin County Water Dist., 2008 WL 4911406 (E.D. Cal. Nov. 13, 2008).

Meadowbrook operates five wells in Merced, California, two of which are located within a half-mile of the BAC Site, MWC-2 and MWC-4.

It is further alleged that the remaining defendants contributed to Plaintiffs' exposure to carcinogens and/or toxins based on their operation of a nearby drainage system (City and County of Merced), a collection system and ponds (Franklin County Water District), and an irrigation canal (Merced Irrigation District). According to Plaintiffs, these municipalities and water districts were "direct participants" in the contamination based on their location and significance to the local community's water supply.

These entities are referred to as the "Public Entity Defendants."

A. History and Remediation of BAC Site

From 1969 until 1984, the BAC Site housed a cooling tower manufacturing facility and BAC-Pritchard, Inc., the operator/owner, used pressure-treated wood to make cooling tower frames. The wood was treated in an on-site cylindrical vessel (retort), using two different solutions. Specifically, from 1969 until 1980, the wood was treated with a solution of chromium, copper, and arsenic ("CCA"). In 1980, BAC-Pritchard stopped using arsenic and instead used a solution of acid, copper, and chromium ("ACC"). BAC-Pritchard ceased treating wood at the site in May 1991 and the facility was closed in early 1994.

Merck and Amsted were indirect parents of BAC-Pritchard, which was a subsidiary of Baltimore Aircoil Company ("BAC"). From 1970 to 1985, BAC was a subsidiary of Merck. In 1985, Merck sold its interest in BAC to Amsted Industries, Inc.

BAC-Pritchard, Inc. was dissolved in September 1993. However, the parent company, BAC, owned the Site until 2001 when it was transferred to Defendant Track Four, Inc., a subsidiary of BAC. Track Four, Inc. owned the Site from February 2001 until July 2002, when it was sold to Santa Fe Aero Vista, LLC. Santa Fe Aero Vista, LLC is not a party to this lawsuit.

The BAC Defendants acknowledge that elevated levels of hexavelant chromium and arsenic were discovered at the BAC Site as early as 1986. That same year, BAC Defendants hired an environmental consultant to investigate and implement a number of remedial measures associated with contamination at the Site. The retained consultant found elevated levels of hexavelant chromium and arsenic in the soil at the BAC Site. Subsequent consultants found hexavelant chromium and arsenic in groundwater beneath the Site and in a stormwater pond located on the southeastern portion of the Site.

A 16-inch pipe connects the stormwater pond to the adjacent El Capitan irrigation canal, owned by Defendant Merced Irrigation District ("MID"). When the pipe is open, water can flow to or from the pond and the canal. The pipe is the focus of the "surface water" pathway.

In 1991, IT Corporation, a remediation consultant hired by Merck and Amsted, completed a "clean closure" of the storm water pond. The closure involved excavating contaminated soil from the pond and disposing of it at a landfill under the supervision of the California Department of Toxic Substances Control ("DTSC"). In 1993 and 1994, IT Corp., under the supervision of the Regional Water Quality Control Board ("RWQCB"), installed a pump-and-treat groundwater remediation system to control migration of contaminants beneath and off of the Site and to remove hexavelant chromium and arsenic from groundwater. IT Corp. continued to expand and modify the groundwater treatment system through 2005.

In 1994, as part of the cleanup efforts, IT Corp. prepared a health risk assessment for the BAC Site. According to BAC Defendants, the health risk assessment employed "conservative exposure assumptions" to protect public health and was premised on the fact that no further remediation would take place. Based on these assumptions, IT Corp. determined that contamination at the Site created a cancer risk of two in one million (ratio of 2:1,000,000). The BAC Defendants assert that the DTSC employs a ratio of one in one million (1:1,000,000) to determine whether additional remediation should be performed.

BAC Defendants contend that the EPA considers risks in the range of one in ten thousand to one in one million (1:10,000 to 1:1,000,000) as "acceptable."

IT Corp.'s risk assessment did not evaluate the risk posed by the Site in the past (pre-1994), but looked prospectively at the future risk the Site might create if no further remediation was performed. IT Corp.'s risk assessment determined that if soils in the area of the former pressure-treating system were excavated or capped, that would "effectively eliminate the exposure pathways (and subsequent risk)." In 1996, IT Corp. completed excavation of contaminated soil from the area of the former pressure-treating system and installed a four-inch thick asphalt cap over the entire area. The BAC Defendants maintain that the excavation and cap "effectively eliminated the risks identified in the 1994 risk assessment."

Additional remediation at the BAC Site was performed in the mid to late-2000's by Arcadis, Inc., an environmental consulting firm to the BAC Defendants, which was paid $17 million to complete the cleanup of the BAC Site. This included excavation of the contaminated soil beneath the asphalt cap installed in 1996. Once the soil was removed, Arcadis re-paved the exposed soil. Arcadis also treated the groundwater with methanol to convert the hexavelant chromium to the more benign trivalent chromium. The groundwater treatment is expected to be completed within a few years, at which point the BAC Site will be monitored by the RWQCB until its final closure.

BAC Defendants assert that they have spent approximately $39 million remediating the BAC Site.

B. RWQCB Involvement — 2007

In 2007, the RWQCB announced the availability for public review of a revised cleanup plan for the BAC Site. According to the BAC Defendants, the RWQCB solicited comments from the community and held an informational meeting to discuss the status of the BAC Site remediation. Following the public comment period, in May 2007, the RWQCB published a "fact sheet" regarding remediation of the Site, which purportedly expressed the RWQCB's opinion that contaminants "posed no health risk to residents in the vicinity of the Site." The BAC Defendants assert that the RWQCB's "fact sheet" expressed several additional opinions re: the BAC Site:

The RWCQB's "Fact Sheet," issued in February 2007, provides a "summary of the cleanup plans, site history, the contaminant[s] present, and opportunities for public involvement." (Doc. 787-35.)

1. Groundwater supply wells in the vicinity of the Site have been tested and have not been impacted by the pollution associated with the former BAC facility;
2. Impacted groundwater is not being used for drinking water supply; and
3. The Site is currently not a threat to public health.

(Doc. 677-2 at 7:4-7:7.)

C. 2009 RWQCB Briefing to Senator Dianne Feinstein

On January 15, 2009, the RWQCB provided a "briefing paper" on the BAC Site in response to an inquiry by Senator Diane Feinstein. (Doc. 725-4.) The brief "provide[d] a description of the environmental conditions at the site, as well as a summary of the regulatory actions and cleanup responses taken at the site to address hexavalent chromium contamination." (Id.) In the brief, the RWQCB summarized its regulatory involvement, which commenced in 1987, and concluded that "[b]ased upon a thorough review of all of the data currently in the Central Valley Water Board's files, Board staff believe that the public is not being exposed to harmful levels of hazardous substances originating from the BAC site." Key details of the RWQCB's briefing are summarized as follows:

• In 1989, the Board determined that sediments in a storm water pond at the BAC site were contaminated with hexavalent chromium. The Board required the BAC Defendants to excavate the pond and further remediate the pond area. Storm water monitoring indicates that the BAC Defendants' remedial actions were effective in reducing total chromium levels in the storm water that was discharged from the site to below California's MCL for total chromium in drinking water of 50 ppb.
• In 1992, the U.S. EPA conducted an investigation to determine if the site warranted actions pursuant to the federal Superfund law ("CERCLA"). Based on its evaluation, the EPA recommended that no further remedial action under CERCLA was required because it found no hazardous substances were detected in drinking water wells or private wells and no designated water intakes or fisheries within 15 miles. RWQCB notified residents of the BAC Defendants' cleanup activities through public notices and a fact sheet.
• In 1993, under the regulatory oversight of the RWQCB, a groundwater monitoring extraction system was installed. The system commenced operations in 1994 and was regulated pursuant to a permit issued by the Central Valley Water Board. The system was active in January 2009 and has extracted and treated over 2,200 gallons of water. It has removed over 5,400 pounds of hexavelant chromium from that water.
• Merck installed a network of 65 monitoring wells that are used to measure the extent of hexavelant chromium pollution in the groundwater. A thorough review, conducted by Board staff, of the sampling and analyses results form the monitoring well network has determined that pollution from the BAC site has not impacted drinking water supply wells in the vicinity of the BAC site. Testing of the water supply wells indicate that hexavelant chromium levels are consistent with background levels for the area.
• In 2000, RWQCB staff reviewed and concurred with the BAC Defendants' reports, which indicated that no domestic supply wells were impacted by pollution originating from the BAC site.
• RWQCB staff believes storm water discharges from the site since the cleanup of the pond occurred in 1991 do not present a significant risk for exposure to the public of hazardous contaminants because: (1) wood treating operations ceased in 1991; (2) all significant sources of hexavelant chromium have been eliminated [and] are not exposed to storm water drainage from the site; (3) soils and sediment in the storm water pond that were contaminated with hexavelant chromium have been excavated and disposed off site; (4) total chromium concentrations have consistently been below the drinking water standardPGPage 12 for total chromium.

The brief noted that "hexavelant chromium is naturally occurring in many geologic formulations in the Central Valley." (Doc. 725-4.)

(Doc. 725-4 at pgs. 2-3.)

The RWQCB's briefing contained a detailed review of the groundwater pollution and cleanup at the BAC Site:

Groundwater at the site is located about 40 feet below the ground surface. Hexavalent chromium has seeped into the groundwater, which has migrated away from the area of release with the flow of the groundwater, which is generally to the north. Seasonal migration of contamination to south, beneath a small portion of the Beachwood neighborhood has also occurred [. . .]
The contamination has remained confined to the upper regions of the groundwater due to the occurrence of a clay layer at about 90 to 100 feet below the ground surface. The clay layer has helped to prevent the migration of contaminants to deeper levels that serve as the drinking water supply for the local community.
The two nearby drinking water supply wells owned by the Meadowbrook Water Company, which provided water to the Beachwood neighborhood, take water from deeper groundwater zones. The closest well to the BAC site, MWC-4, draws water from a depth 210 feet below the ground surface and is below the clay layer. To ensure shallow groundwater is not pumped into the drinking water supply, the well is sealed with cement grout to 200 feet below ground surface. A second supply well, MWC-2, takes it water from more than 140 feet below the ground, but the well is older and the depth of the seal against shallow groundwater entering the well is unknown. MWC-2 is no longer in service. Monitoring wells emplaced between the hexavalent chromium plume and the two supply wells indicate that the plume has not reached the location of the supply wells in groundwater situated above and below the clay layer.
The community's drinking wells, MWC-2 and MWC-4, have been tested a number of time and hexavalent chromium levels in the wells are within the expected background concentration range for hexavalent chromium.

(Id. at pgs. 7-8.)

The RWQCB attached a table summary of water supply well data for hexavalent chromium from public water supply systems near the Beachwood neighborhood, specifically, Meadowbrook Water Company, the City of Atwater and Winton Water District. All of the Meadowbrook wells tested within background levels, including MWC-2 and MWC-4. The RWQCB noted: "For comparison purposes, a number of supply wells for the City of Los Banos, located approximately 25 miles southwest of the site, contain higher levels of hexavalent chromium [than the three closest public water supply systems]."

The RQWCB obtained the public well data from the Department of Public Health. (Id. at pg. 10.)

In its "closing observations," the RWQCB stated that "the public is not currently being exposed to harmful substances originating from the BAC Site," and that the cleanup actions "appear reasonable under the circumstances and the cleanup is proceeding in a satisfactory manner." (Id. at pg. 12.)

D. Plaintiffs' Response to BAC Site History/Remediation

Plaintiffs do not specifically dispute the BAC Defendants' recitation of the BAC Site history, including the remediation and RWQCB involvement, rather they take issue with the testing data used to support the BAC Defendants' motions. In particular, Plaintiffs argue that the data from the monitoring wells and MWC-2 Well is unreliable and/or inaccurate for the following reasons, among others: (1) the monitoring wells are too shallow; (2) the use of "selective" and/or "interval sampling"; and (3) the "honor system" employed by the Meadowbrook Water District lacks the necessary formalities to assure accurate records. Based on these factors, Plaintiffs' groundwater modeler, Mr. Douglas Bartlett, allegedly excluded portions of the sampling data from his modeling and analyses. The exclusion, if any, of the MWC-2 and monitoring well testing data is heavily disputed among the parties and forms the basis for BAC Defendants' motion to exclude Bartlett's expert testimony and model.

Plaintiffs also allege that several public entities, including the RWCQB and Defendants Meadowbrook Water District, Merced Irrigation District, and Merced County conspired with the BAC Defendants to: avoid collecting data from the MWC-2 Well, hide documents from third parties, and conceal the true danger/risk of the contaminants at or near the BAC Site. In particular, Plaintiffs contend that the BAC Defendants "conceal[ed] from the Regional Board [RWCQB] and plaintiffs [] the Arcadis groundwater modeling findings concerning the impact of MWC-2 on the contaminated plume." Plaintiffs further allege: "The draft report addressing [Arcadis'] model which was to be submitted to the Regional Board in December of 2007, was altered by Merck by deleting references to the findings of the model and by deleting a graphic illustration of contaminants flowing from the BAC directly into MWC-2."


On March 8, 2007, Plaintiffs commenced this civil action against the current public entity defendants, alleging property damage caused by the April 2006 flood. (Doc. 1.) On September 13, 2007, in the second amended complaint, Plaintiffs named Merck Co., Inc., Amsted Industries, Inc., Baltimore Aircoil Company, and Track Four, Inc. as Defendants in this action. (Doc. 35.) The operative complaint, the eighth amended complaint, was filed by Plaintiffs on March 26, 2010. (Doc. 633.) The eighth amended complaint alleges ten claims against the BAC Defendants: (1) violation of 42 U.S.C. 6972(a)(1) [RCRA]; (2) violation of 42 U.S.C. 6972(a)(1)(b) [RCRA]; violation of 33 U.S.C. 1311(a) [CWA]; (4) violation of 33 U.S.C. 1342(a) and (b) [CWA]; (5) negligence; (6) trespass; (7) nuisance; (8) wrongful death; (9) fraud and deceit; and (10) civil conspiracy.

On March 23, 2009, BAC Defendants filed a "Motion for Case Management Order Re: Exposure" to "compel plaintiffs to make a prima facie showing of exposure." (Doc. 355.) The motion was denied on July 6, 2009, however, on August 12, 2009, the Court established a multi-phase trial plan in which case-wide exposure issues were to be tried first ("Phase 1"), before general medical causation ("Phase 2") and plaintiff-specific exposure and causation ("Phase 3"). The August 12, 2009 "Order Modifying Scheduling Conference Order" provides, in relevant part:

Discovery and expert disclosures shall be conducted in phases. Phase 1 shall focus on the issue of general exposure; that is, whether contaminants from the former [] BAC Site, Franklin County Water District or the April 2006 Flood have ever reached any location where plaintiffs could have been exposed to them, and if so, when such contaminants arrived, how such contaminants arrived at the location, how long they were present, and at what levels they were present.

(Doc. 540 at 1:14-1:28.)

On June 1, 2010, BAC Defendants moved for partial summary judgment on Plaintiffs' state law tort claims for personal injury and property damages, i.e., Plaintiffs' claims for negligence (Claim X), trespass (Claim XII), nuisance (Claim XIII), and wrongful death (Claim XIV). (Doc. 677.) According to Defendants, Plaintiffs have failed to produce evidence sufficient to show that any plaintiff was exposed to contaminants allegedly released from the former BAC Site located in Merced, California. BAC Defendants argue that since there is no contamination evidence, there is no genuine issue of material fact on the threshold issue of causation.

Defendants Merced Irrigation District and Merced County joined BAC Defendants' motion for partial summary judgment on June 29 and August 13, 2010. (Docs. 772 813.)

In support of their motion, BAC Defendants submit: (1) a Memorandum of Points and Authorities ("Memorandum"); (2) a Statement of Undisputed Facts in Support of its Motion; (3) the declarations of attorney R. Morgan Gilhuly and experts Scott Fendorf, Daniel B. Stephens, John L. Wilson, and Paolo Zannetti; and (4) separate motions to exclude the testimony of Plaintiffs' groundwater expert, Douglas Bartlett, and air expert, Camille Sears.

Plaintiffs opposed the motion on July 1, 2010, submitting over 3,000 pages of documentation and 168 exhibits. Plaintiffs oppose summary judgment on grounds that they have "substantial evidence" supporting their contention that "contaminants from the BAC facility have historically migrated from the [BAC] facility via groundwater, surface water and air pathways to locations were exposed to them and at levels which could cause harm." In particular, Plaintiffs argue that Sears and Bartlett's expert testimony is admissible and creates "competing expert opinions as to the migration of the contaminants from the BAC facility." According to Plaintiffs, Sears and Bartlett's expert opinions, by themselves, are sufficient to withstand a Rule 56 challenge.

Attached to Plaintiffs' opposition are the "rebuttal" declarations of experts Frank Agardy and Patrick Sullivan. 8787 Dr. Agardy, a purported soil and groundwater expert, opines that Camille Sears incorporated several of Agardy's opinions in her calculations of historical air borne contaminants. The substance of Dr. Agardy's declaration is to undermine Dr. Fendorf, BAC Defendants' air borne expert, who opined that "any chromium in the [wood treating solution — CCA or ACC] solutions that had reacted with the wood would not be in the hexavelant form but would instead be in the trivalent form." In his rebuttal declaration, Dr. Argady explains that "treated wood can contain [hexavelant chromium] after treatment" and, more importantly, Dr. Fendorf's conclusions are mistaken because "his analysis purposefully ignores the liquid released from the treated lumber that have not `reacted with wood.'" According to Dr. Argady, these releases are a source of hexavelant chromium at the BAC Site.

In addition to a global opposition memorandum, Plaintiffs separately opposed the BAC Defendants' motion to exclude the testimony of Dr. Bartlett, (Doc. 776), and Camille Sears, (Doc. 781).

BAC Defendants filed their reply briefs on August 13, 2010. (Docs. 834 835.)

Following oral argument on October 6, 7, 13, 14, and 15, 2010, the parties were requested to provide supplemental briefing on the pathway exposure issues. The parties submitted final supplemental briefs on October 19 and 20, 2010. (Docs. 864-65, 867.)


Summary judgment/adjudication is appropriate when "the pleadings, the discovery and disclosure materials on file, and any affidavits show that there is no genuine issue as to any material fact and that the movant is entitled to judgment as a matter of law." Fed.R.Civ.P. 56(c). The movant "always bears the initial responsibility of informing the district court of the basis for its motion, and identifying those portions of the pleadings, depositions, answers to interrogatories, and admissions on file, together with the affidavits, if any, which it believes demonstrate the absence of a genuine issue of material fact." Celotex Corp. v. Catrett, 477 U.S. 317, 323, 106 S.ct. 2548, 91 L. Ed. 2d 265 (1986) (internal quotation marks omitted). Where the movant will have the burden of proof on an issue at trial, it must "affirmatively demonstrate that no reasonable trier of fact could find other than for the moving party." Soremekun v. Thrifty Payless, Inc., 509 F.3d 978, 984 (9th Cir. 2007). With respect to an issue as to which the non-moving party will have the burden of proof, the movant "can prevail merely by pointing out that there is an absence of evidence to support the nonmoving party's case." Soremekun, 509 F.3d at 984.

When a motion for summary judgment is properly made and supported, the non-movant cannot defeat the motion by resting upon the allegations or denials of its own pleading, rather the "non-moving party must set forth, by affidavit or as otherwise provided in Rule 56, `specific facts showing that there is a genuine issue for trial.'" Soremekun, 509 F.3d at 984. (quoting Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 250, 106 S.ct. 2505, 91 L. Ed. 2d 202 (1986)). "A non-movant's bald assertions or a mere scintilla of evidence in his favor are both insufficient to withstand summary judgment." FTC v. Stefanchik, 559 F.3d 924, 929 (9th Cir. 2009). "[A] non-movant must show a genuine issue of material fact by presenting affirmative evidence from which a jury could find in his favor." Id. (emphasis in original). "[S]ummary judgment will not lie if [a] dispute about a material fact is `genuine,' that is, if the evidence is such that a reasonable jury could return a verdict for the nonmoving party." Anderson, 477 U.S. at 248. In determining whether a genuine dispute exists, a district court does not make credibility determinations; rather, the "evidence of the non-movant is to be believed, and all justifiable inferences are to be drawn in his favor." Id. at 255.


A. Introduction

Defendants move for summary judgment on Plaintiffs' negligence, trespass, nuisance, and wrongful death causes of action. Defendants' claim there is no evidence of exposure via any pathway, i.e., groundwater, private domestic wells, surface water, air, or soil. Plaintiffs oppose each facet of Defendants' motion, arguing that their expert evidence clearly shows that "contaminants from the BAC have historically migrated from the facility via groundwater, surface water, and air pathways to locations were plaintiffs were exposed to them and at levels which could cause harm." (Doc. 792 at 1:6-1:9.)

The substance of the parties' briefing, more than 5,000 pages of argument and expert/scientific reports, focuses on two pathways, groundwater and air. As to these pathways, Plaintiffs support their Phase 1 exposure burden with the testimony of Douglas Bartlett, groundwater hydrologist (groundwater modeler), and Camille Sears, meteorologist (air modeler). Defendants move to exclude these expert opinions and testimony pursuant to Federal Rule of Evidence 702 and two United States Supreme Court cases, Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579 and Kumho Tire Co. v. Carmichael, 526 U.S. 137 (1999). In particular, Defendants challenge Bartlett and Sears' expert testimony on grounds that it cannot pass Daubert's "gatekeeping" requirement.

B. Daubert Legal Standard

Rule 702 governs the admissibility of expert testimony. Pursuant to Rule 702, a witness qualified as an expert in "scientific . . . knowledge" may testify thereto if: "(1) the testimony is based upon sufficient facts or data; (2) the testimony is the product of reliable principles and methods; and (3) the witness has applied the principles and methods to the facts of the case." Fed.R.Evid. 702.

The trial court acts as a gatekeeper to the admission of expert scientific testimony under Rule 702. Daubert, 509 U.S. at 579-580. The court must conduct a preliminary assessment to "ensure that any and all scientific testimony or evidence admitted is not only relevant but reliable." Id. at 589. This two-step assessment requires consideration of whether (1) the reasoning or methodology underlying the testimony is scientifically valid (the reliability prong); and (2) whether the reasoning or methodology properly can be applied to the facts in issue (the relevancy prong). Id. at 592-93; Kennedy v. Collagen Corp., 161 F.3d 1226, 1228 (9th Cir. 1998).

Reliable testimony must be grounded in the methods and procedures of science and signify something beyond "subjective belief or unsupported speculation." Daubert, 509 U.S. at 590. The inferences or assertions drawn by the expert must be derived by the scientific method. Id. In essence, the court must determine whether the expert's work product amounts to "`good science.'" Daubert v. Merrell Dow Pharms., Inc., 43 F.3d 1311, 1315 (9th Cir. 1995) ("Daubert II") (quoting Daubert, 509 U.S. at 593). In Daubert, the Supreme Court outlined factors relevant to the reliability prong, including: (1) whether the theory can be and has been tested; (2) whether it has been subjected to peer review; (3) the known or potential rate of error; and (4) whether the theory or methodology employed is generally accepted in the relevant scientific community. Daubert, 509 U.S. at 593-94. The Supreme Court emphasized the "flexible" nature of this inquiry. Id. at 594. As later confirmed in Kumho Tire Co. v. Carmichael, 526 U.S. 137 (1999): "Daubert's list of specific factors neither necessarily nor exclusively applies to all experts or in every case. Rather the law grants a district court the same broad latitude when it decides how to determine reliability as [the court] enjoys in respect to its ultimate reliability determination." Id. at 141-42.

The relevancy, or "fit," prong requires that the testimony be "relevant to the task at hand, . . . i.e., that it logically advances a material aspect of the proposing party's case." Daubert II, 43 F.3d at 1315 (quoting Daubert, 509 U.S. at 597). Relevancy requires opinions that would assist the trier of fact in reaching a conclusion necessary to the case. See Kennedy, 161 F.3d at 1230.

The Daubert analysis focuses on the principles and methodology underlying an expert's testimony, not on the expert's conclusions. Daubert, 509 U.S. at 595. However, the Supreme Court has cautioned that "conclusions and methodology are not entirely distinct from one another." General Elec. v. Joiner, 522 U.S. 136, 146 (1997).

As such, "[a] court may conclude that there is simply too great an analytical gap between the data and the opinion proffered." Id. Nothing in either Daubert or the Federal Rules of Evidence requires the admission of opinion evidence connected to existing data "only by the ipse dixit of the expert." Id.

C. Pathway # 1 — Air

Plaintiffs allege that soil contaminated with hexavelant chromium and/or arsenic was transported to their homes or properties via wind and other airborne pathways. To satisfy their Phase 1 burden on the air pathway, Plaintiffs submit: (1) the expert testimony of air modeler Camille Sears (air exposure for the years 1969-1993); and (2) a 1994 Risk Assessment completed by IT Corp (air exposure post-1993 years).

Defendants contend that summary judgment is appropriate with respect to any air pathway because Plaintiffs' expert's testimony is inadmissible under Daubert. Specifically, Defendants move to exclude Ms. Sears' testimony on grounds that: (1) she is not qualified to perform soil calculations; (2) her entire emission scenario is flawed and lacks scientific reliability; and (3) her input parameters are scientifically unsound. As to the Risk Assessment, Defendants assert that it has no evidentiary value because "it was based on a hypothetical scenario that never occurred, and it does not purport to accurately calculate actual exposure or risks." (Doc. 677-2 at 23:18-23:19.) To support their Daubert motion, Defendants rely on the declaration and testimony of Dr. Paulo Zanetti, an air modeler, and Dr. Scott Fendorf, a chemist. Both these experts opine that Ms. Sears' model is scientifically unreliable.

Plaintiffs rejoin that Ms. Sears has performed similar soil calculations in modeling air dispersion, her emission scenario is well-accepted in the air modeling community and the dispute over her input parameters bear on the weight of the opinion rather than on its admissibility.

1. Fed.R.Evid. 706

Based on the extent of the dispute between the parties' experts and the complexity of the scientific issues presented by air modeling, the Court retained two independent experts, Chatten Cowherd Jr., Ph.D. and Richard Countess, Ph.D., under Fed.R.Evid. 706, to assist in the understanding and analyzing the air pathway dispute. Drs. Countess and Cowherd prepared a joint expert report, submitted on November 22, 2010. (Doc. 942.) The joint independent expert report and the opinions contained therein are discussed in the context of Defendants' criticisms of the air model.

Under Ninth Circuit law, "district courts enjoy wide latitude to make Rule 706 expert appointments." Monolithic Power Systems, Inc. v. O2 Micro Intern. Ltd., 558 F.3d 1341, 1348 (9th Cir. 2009). In Monolithic Power, the Ninth Circuit stated that it "perceive[d] no abuse of discretion [in appointing a 706 expert] in this case where the district court was confronted by what it viewed as an unusually complex case and what appeared to be starkly conflicting expert testimony." Id. citing Walker v. Am. Home Shield Long Term Disability Plan, 180 F.3d 1065, 1071 (9th Cir. 1999) (finding no abuse of discretion in Rule 706 appointment where the scientific evidence was "confusing and conflicting" and the appointment "assist [ed] the court in evaluating contradictory evidence about an elusive disease of unknown cause").

2. Supplemental Briefing Fed.R.Evid. 706 Experts

The parties filed supplement briefing concerning the air expert reports on November 29, 2010. The supplemental arguments mirror those advanced in the parties' earlier briefing, however, Plaintiffs reiterate that "the FRE 706 reports serve only to confirm the grave concerns plaintiffs have repeatedly expressed to this Court related to these Daubert proceedings." Plaintiffs contend that the Court "may be tempted to act as a trier of fact by weighing the relative strengths and weaknesses of factual and scientific assumptions made by the experts." They further assert that the air model disagreements "go to the weight of their opinions, and not their admissibility."

Plaintiffs' concerns are unsound. There is no "temptation" to weigh facts, rather the Supreme Court requires that the trial court act as a "gatekeeper" to the admission of expert scientific testimony under Rule 702. Daubert, 509 U.S. at 579-580. Daubert commands that district courts conduct a preliminary assessment to "ensure that any and all scientific testimony or evidence admitted is not only relevant but reliable." Id. at 589.

Defendants rejoin that the 706 expert reports "confirm[] that Plaintiffs' groundwater and air models contain serious methodological flaws, including faulty assumptions unsupported by science or facts, and are not reliable." Responding to Plaintiffs' "weight" arguments, Defendants assert that if Ms. Sears' expert report is either unreliable or irrelevant, it is inadmissible. An inadmissible report cannot create a disputed issue of fact. BAC Defendants also include a string citation to a number of federal cases holding that, under Daubert, an expert opinion cannot be "based on assumptions of fact without evidentiary support, or on speculative or conjectural factors."

BAC Defendants correctly distinguish Elam v. Alcolac, 765 S.W.2d 42 (Mo. App. 1988), the primary case relied on by Plaintiffs concerning the lack of data to establish exposure, i.e., why modeling is needed in this case. Elam, however, is factually distinguishable as the defendant in that case was required to conduct monitoring but refused to do so. Moreover, an unpublished twenty-two year old decision from a state appellate court is not binding on any federal court.

3. Testimony/Model of Camille Sears (1969-1993)

Plaintiffs designated Sears as an air pathways expert who will opine on whether Plaintiffs' homes were exposed to significantly elevated air concentrations of hexavalent chromium and arsenic. Sears' declaration specifies she has worked in the air quality field since January 1982, following her graduation from the University of California at Davis (M.S. and B.S. degrees in atmospheric science). Prior to forming her own consulting firm in 1992, Sears worked as a private air consultant/scientist (Dames Moore, URS Consultants) and Air Toxics Program Coordinator (Santa Barbara Air Pollution Control District). Sears has been a testifying expert for twenty years, since 1990, and provided air modeling testimony in California Dept. of Toxic Substances Control v. Interstate Non-Ferrous Corp., No. 97-CV-5016-OWW-LJO, an environmental contamination case involving, among other things, the alleged air dispersion of dioxin-containing ash released from incendiary operations at a smelting site in Mojave, California.

In her expert report, Ms. Sears states that she was "asked to analyze and characterize the air dispersion of particulates from fugitive dust emissions sources at the former BAC-Pritchard site, located in Merced, California." (Doc. 785-2, Expert Report of C. Sears, at pg. 4.)

According to her expert reports, declarations, and testimony, Sears' followed a three-step process. First, she calculated air concentrations and surface deposition of hexavelant chromium and arsenic resulting from the BAC site based on "widely-accepted air dispersion modeling techniques." Second, she input these calculations into an air flow model endorsed by the U.S. Environmental Protection Agency called AERMOD to determine if Beachwood residents were exposed to particulates from fugitive dust emissions sources at the former BAC site. Third, she provided opinions re: whether, from 1969 to 1993, areas surrounding the BAC facility were exposed to elevated air concentrations of arsenic and hexavelant chromium.

Sears built her model by extrapolating data from chromium at the BAC site, in a treating solution that was released to the ground at the retort area where it entered the soil, and that this exposed solution in and on the soil was later disturbed by fork-lifts driving on the site. She offered the following opinions of exposures caused by the BAC facility wood storage area emissions:

• During the period 1969 through 1993, areas surrounding the BAC facility were exposed to significantly elevated air concentrations of hexavalent chromium;
• During 1969, and the period 1991 through 1993, areas surrounding the BAC facility were exposed to elevated air concentrations of hexavalent chromium, but not as high as 1970 through 1990;
• During the period 1969 through 1983, areas surrounding the BAC facility were exposed to significantly elevated air concentrations of arsenic;
• During 1969, and the period 1981 through 1983, areas surrounding the BAC facility were exposed to elevated air concentrations of arsenic, but not as high as 1970 through 1980;
• During the period 1969 through 1983, areas surrounding the BAC facility were exposed to increased arsenic surface deposition levels;
• During the period 1969 through 1993, areas surrounding the BAC facility were exposed to increased hexavalent chromium surface deposition levels.

(Doc. 781 at 5:21-6:9.)

Defendants move to exclude these opinions because her testimony rests on a number of unsupported assumptions about the chemical and physical properties of soils at the Site during the relevant time period (1969-1993). According to Defendants, Sears is unqualified to make these assumptions because she "is an atmospheric scientist and meteorologist, not an expert in chemistry, geology, geochemistry, or soil science." In addition, as part of her calculation of fugitive emissions, Sears allegedly employed an improper EPA emission technique, namely the paved road emission scenario (AP-42, § 13.2.1). The flawed assumptions, enumerated below, relate solely to Sears' "emission rate" calculations:

1. Valence — Sears assumes that all of the chromium that dropped from treated wood at the BAC site was hexavalent chromium;
2. Particle Size — She assumes the contaminated dust was comprised of particles that were less than 10 microns in diameter, making them small enough to blow off the site; and
3. Accumulation — She assumes that one year's worth of drippage accumulated and remained on the surface of the site, in the form of dust that is 90 percent pure (900,000 parts per million) chromium and arsenic.

Defendants contend that these assumptions are demonstrably false and Sears does not have the expertise to make them. To support their Daubert motion, Defendants rely on the declaration and testimony of Dr. Paulo Zanetti, an air modeler, and Dr. Scott Fendorf, a chemist. Dr. Zanetti disputes Sears' use of the EPA's "paved road" emission scenario and, critically, the assumption that drippage would accumulate and remain on the surface of the wood storage area in the form of dust at a concentration of 900,000 parts per million. Dr. Fendorf opines that chromium does not drip off only in hexavalent state and, assuming it did, the chromium particles would be at least 100, not 10, microns in size.

Sears responds first to Defendants' claims that she is unqualified to calculate fugitive dust emissions from a wood processing facility:

While this is the first time I have calculated fugitive dust emissions from a wood treating facility, the principles involved are no different than any other paved surface fugitive dust source. Furthermore, I have calculated air emissions of hexavalent chromium dozens of times — from cooling towers, annealing operations, chrome plating and anodizing, surface coatings (including primers and pigmented — paints), welding operations and cold fire combustion. I also have experience calculating air emissions of arsenic — from combustion sources, fugitive dust and mining activities.

(Doc. 785 at ¶ 15.)

On October 7, 2010, the second day of Daubert hearings, Sears testified that while she lacks formal "soil science" training, she has calculated air emissions in a number of environmental contamination cases and teaches a class in chemistry:

First of all, I heard [defense counsel] say that I had no education in chemistry or geology or soil science. That's — that's never anything that I've said. I have taken a number of chemistry courses and I've been involved in a number of soil analyses over the years. I've taught chemistry to air toxic students through the UCSB extension as part of my air toxics class. I never said I had no education in chemistry. And I don't believe it's correct for [defense counsel] to say so.
Over the last 30 years, which is the time that I've been doing these types of analyses, I've been involved in the use of chemistry on a weekly basis. It's part of the job experience. You don't need to be a chemist or a geologist, soil scientist or geochemist to calculate air emissions from basically soil releases or surface releases of dust.
On the contrary, Dr. Fendorf, who they offer as their expert on this issue, I don't believe has ever calculated air emissions, or at least I did not see any of his qualifications stating that he has ever calculated air emissions.

(RT, Oct. 7, 2010, 257:7-257:25.)

Ms. Sears detailed her experience calculating fugitive air emissions during counsel's direct examination on October 13, 2010. Ms. Sears' air emission experience, which is quite extensive and includes review of industrial sites and hexavalent chromium, rules out any characterization of her as "unqualified" in the field of calculating air emissions:

Q. Ms. Sears, are you qualified to calculate air emissions? A. Yes, I believe so. Q. And why do you feel that you're qualified to do that? A. Well, first, I want to correct — or give my viewpoint on the issue of whether a meteorologist is qualified to calculate air emissions. We heard testimony last week that Ms. Sears is a meteorologist and atmospheric scientist and therefore she's not qualified to calculate air emissions. I believe that's untrue. First of all, meteorologists, atmospheric scientists, we need to take all the chemistry, all the physics that basically any engineer would take. And basically the rest of it is real world experience. I started calculating air emissions in 1983. I was hired by the Santa Barbara county air pollution control district as an air pollution engineer. And my job was to calculate emissions, not only for permitting purpose, but also for fee calculations. Our district became a fee based entity, which the fees were generated by the amount of emissions that companies emitted. And during that time, when I was at the air pollution control district, not only did I calculate emissions for a number of different sources, I also had to develop what we called emission estimating techniques, or EETs for many sources for which we never calculated emissions before. Q. Let me ask you just on this subject. What types of emissions did you calculate when you were with the Santa Barbara County Air Pollution Control District? A. I was just about to answer that. Thanks. The types of emissions I would calculate were generally by the — I'll do it by type of pollutant. Originally in the early `80s, the only thing we dealt with would have been called criteria air pollutants. And those are pollutants for which there were ambient air quality standards. For example, nitrogen dioxide, sulphur dioxide, carbon monoxide, lead and particulate matter were the ones where we would calculate emissions. And for ozone, we would calculate emissions of volatile organic compounds as well. Later in the `80s, when air toxics, we also called non-criteria pollutants became the focus of regulation, we had to calculate air emissions of essentially all the toxic chemicals that you can think of from every source in the district. And those would include oil and gas processing facilities, excuse me, medical device manufacturers, diatomaceous earth mining facilities, we had Casmalia class one landfill. We had mining operations. There were no end to the different types of sources that we had to calculate emissions from. There's several hundred different facilities I was responsible for calculating air emissions. Q. And give us an example as to how you did calculate the emissions at that time. A. Well, I — I could think of a lot of them. However, the type of emissions that — the type of sources that we would generally deal with for air toxics, you might think of as an oil and gas processing facility. And we had offshore and on shore components, stationary and mobile sources. So we had to identify all the emission sources for all the different pollutants and then quantify them using emission calculation methods. Sometimes we did source testing. And other times, like I said, we have to use what we call emission factors. And again, for many of these sources, for air toxics, none of these emission factors existed. We had to develop them. And I was on the committee for the State of California, the criterion guidelines regulation committee, which established a number of emission estimating techniques or EETs for air toxics. Q. And how many cases have you testified in either a courtroom like this or in state court or an administrative hearing, wherein you have rendered opinions and testimony on the calculation of air emissions, Ms. Sears? A. About 15 cases. And I outlined them in my opposition declaration. So about 15 cases where I testified and where I've actually calculated emissions that went into my model. Q. And what experience do you have in calculating air emissions of hexavalent chromium? A. Over the years, probably starting in the late `80s, I calculated air emissions of hexavalent chromium dozens of times from cooling towers, where zinc chromate was added as an — vis-à-vis an anti-corrosion compound. Calculated hexavalent chromium emissions from chrome anodizing and chrome plating operations, both hard chrome and soft chrome. And also from sparging, which is air injection into chrome plating operations. Q. Now, in terms of the central valley of California here, what experience do you have in evaluating the impact of hexavalent chromium in the air here in the central valley of California? Which is where the BAC plant is located. A. Okay. Q. Former BAC plant. A. I'll answer that. But I also want to say I've also calculated hex chrome emissions from a number of surface coating operations, such as spray painting and aircraft refurbishing and that type of thing. Certain coatings, zinc chromate, strontium chromate, lead chromate have hexavalent chromium in the formulation. Now, as far as the Central Valley goes, I have calculated emissions from cooling towers. This was from a case where the emissions were in the past. And we calculated — or I calculated hexavalent chrome emissions. And then I modeled these emissions and calculated the air concentrations. And then we were also involved in looking at the background levels of hexavalent chromium in the central valley, basically in the `90s. Because that's when we had data. And so I've had quite a bit of experience looking at the data from the hexavalent chrome in the air from Stockton, Modesto, Fresno, Bakersfield, the site where the Air Resources Board was collecting this data in the 1990s. Q. What experience — there was some mention last week by [defense counsel] about your background or lack of background in regard to the chemistry of air pollutants. Do you remember that testimony? A. I do. Q. What is your experience in regard to the chemistry of air pollutants, Ms. Sears? A. Well, again, I want to say that, again, I'm a meteorologist, and I would never say I'm a chemist, but I would never say I don't have any education in chemistry. I do. And — but I — and again, a lot of it, though, is on the job training in the last 30 years since I graduated with a masters from UC Davis. And essentially what you have to think of is that when you're looking at air toxics, toxic air pollutants, you need to know something about the chemistry of these chemicals. You know, chlorinated alkanes, chlorinated alkines, valence of metals, all these different things come into play as well as the reactivity of volatile organic compounds in ozone formation. All these are issues that we have to deal with consistently in the air pollution field. And so it's a matter of needing to know certain details of the chemistry, but not having to be a chemist to do so. Q. So let me ask this more specifically. What is your education, training and background in regard to the subject of understanding the chemistry of air pollutants? A. Well, I've taken the basic, you know, again, I'm not a chemist, but I've taken the basic freshman, sophomore chemistry classes. And essentially I was an atmospheric physicist in — at Davis, that was my specialty. I wasn't an air modeling student at UC Davis. And so we dealt a lot with the issues of electricity, magnetism, electromagnetic issues, that would deal with individual compounds as well. And again, after I graduated and became working in the air toxics field, I had to refresh my memory and, you know, go back into dealing withal canes, al keens, al kinds, air mat I can hydrocarbons and so forth and how all these compounds relate to each other. So again, I'm not a chemist, but I'm not a completely ignorant chemistry either. Q. In the opinions that you have rendered in other cases and in other courts, have your opinions included the subject of chemistry in regard to air pollutants? A. Yes. I can give you some examples. Q. Why don't you do that. A. One case that I was working on, it was a federal court case. It was Akee versus Dow, it was in Hawaii. It dealt with soil fumigants. And the soil fumigants were applied to pineapple fields on the island of Oahu from 1946 to — 1946 to 2001. I calculated air emissions from volatilization of Dibromochloropropane, ethylene dibromide, 1,3-trichloropropene, 1,2-dichloropropane, and epichlorohydrin from 319 pineapple fields for 56 years. And, of course, the defendants' experts challenged everything I did. They didn't file any motions to exclude any testimony, they just disagreed with me [. . . .] (RT, Oct. 13, 2010, 560:12-567:4.)

Camille Sears is qualified to offer the testimony about how to calculate fugitive air emissions from a former wood-treatment facility. She has substantial experience in the air emission field, having calculated air emission rates for over twenty years in and around Northern and Central California. Sears has calculated fugitive air emissions for public and private employers and has experience with industrial sites and a variety of harmful pollutants, including hexavelant chromium. By education and experience, Ms. Sears is qualified to opine on air emissions, analysis and modeling in this case and her calculations can be challenged through cross-examination and presentation of contrary evidence. See Robinson v. GEICO General Ins. Co., 447 F.3d 1096, 1106 (8th Cir. 2006) ("Gaps in an expert witness's qualifications or knowledge generally go to the weight of the witness's testimony, not its admissibility.").

Defendants take an overly-narrow view of the degree of expertise necessary to calculate fugitive air emissions in the context of air modeling. An air modeler, who is also an experienced atmospheric scientist and meteorologist, does not need to have a "sub-specialty" or advanced degree in chemistry or soil science before that person has sufficient qualifications to perform air emissions. Defendants cite no authority for such a proposition. Rather, Rule 702 only requires that an expert possess "knowledge, skill, experience, training, or education" sufficient to "assist" the trier of fact, which is "satisfied where expert testimony advances the trier of fact's understanding to any degree." See Ms.Lauria v. Nat'l R.R. Passenger Corp., 145 F.3d

593, 598 (3rd Cir. 1998) (holding trial court abused its discretion by excluding testimony simply because the trial court did not deem proposed expert to be the best qualified or because proposed expert did not have the specialization that the trial court considered most appropriate); see also United States v. Newmont USA Ltd., No. CV-05-020-JLQ, 2007 WL 4856859, at 2 (E.D. Wash Nov. 16, 2007) (in a mining enterprise case, holding that although the expert "is a historian, as opposed to an expert in the specific areas of `corporate organization' is not disqualifying in this case, as his background appears to provide sufficient expertise upon which his opinions, as a general matter, are based."). Ms. Sears has engaged in the field of air modeling and analysis as a regulator and expert for over twenty years. She has worked on sites that have toxic dust emissions. She satisfies Rule 702's foundational standards and her testimony will assist the trier of fact.

Cf. Ponca Tribe of Indians of Okla. v. Continental Carbon Co., No. CIV-05-445-C, 2009 WL 4547612, at 5 (W.D. Okla. Jan. 16, 2009) ("The Court agrees that Mr. Hamlin is unqualified to offer opinions as an expert in interpreting aerial photographs. He has had no training in this area and only a minimal amount of experience doing such work. In addition, even if he were qualified, his opinions in this area would be of limited relevance. His testimony would not aid the Court any more than would simply looking at the photographs.").

Defendants next argue that Ms. Sears' entire emission scenario is flawed and lacks scientific reliability. According to Dr. Zanetti, Ms. Sears' flawed methodology begins and ends with her use of the "paved road" or "parking lot" emission scenario, AP-42 § 13.2.1. Dr. Zanetti opines that there is no scientific basis to model the paved portions at the BAC Site because they do not qualify as "roads." Rather, according to Dr. Zanetti, the proper emission scenario in this case is the EPA equation for unpaved surfaces. The selection of AP-42 as an emission scenario was fully developed during defense counsel's direct examination of Dr. Zanetti on October 14, 2010:

A: In Ms. Sears' scenario, she did all the work by herself, my understanding, she developed this new theory, she developed emission rates, she applied the AERMOD model. She applied the AP-42 equations. Certainly she has the capability to select an AP-42 equation and to use it. The problem is is this the correct methodology? It's not a question of her lacking the mathematical skills to use those equations, the question is is the equation or the methodology appropriate [. . .] I tried to summarize in a way that is easy to understand, my reading of Ms. Sears' work in this case. And in particular, the emission scenario. So there are seven points. And I can start giving a brief description. Q. Okay. Before you do that, if you would, please, explain to us all what you mean by the phrase "emission scenario." A. An air pollution model is a computer program that requires basically two main inputs. The emissions and the meteorology. Q. The emission? A. The meteorology, like the wind speed, the wind direction, the temperature. So these are the two main inputs. Emission and meteorology. And the results will be the concentration data in the community. So the meteorological data are generally easy to collect. There are airports all over the United States providing meteorological data, like wind speed, wind direction that we can collect easily. So generally this is not a big task. But the emission calculation is often a major task and that has enormous consequences in air pollution modeling application because if the emissions are incorrect, the concentration will be incorrect in the same proportional amount. Q. Okay. So emission scenario refers in the BAC case to what? A. In Ms. Sears' theory, she is considering emissions from paved surface in the wood storage area. So task number one is my description of her work, that for the period `69 to 1993, she assumes that contaminated liquid, liquid contaminated with heavy metals, arsenic and chromium 6, is dripping from the treated wood in the wood storage area and this liquid will drip over — or this paved surface of the wood storage area. That's step number one. (RT, Oct. 14, 2010, 703:8-707:2.)

Dr. Zanetti argued that there was insufficient "road traffic" at the wood treatment area to support Ms. Sears' selection of the emission scenario for paved roads. He also argued that AP-42 explicitly states that the AP-42 formula should not be applied to "stop and go" traffic:

A: And she is using the AP formulas for road traffic. And this formula has been used many, many times [. . .] so this formula has been used by hundreds of times all over the world. It's a simple formula. Q. But it's used for, you said, road traffic? A. Road traffic. And my comment is that, first of all, I look at the picture of the wood storage area. There is no road. Q. Okay. A. First of all. Second, there is a picture of, in the AP-42 document, which I have in my files, and it shows a road with traffic. And no forklifts in the traffic, of course. This formula is designed for calculating emissions from traffic, not for forklifts. I don't think I ever seen, and I don't think Ms. Sears has cited one literature study where this formula has been applied to forklifts. And in order to apply to forklifts, you need really to stretch enormously the assumption that are inherent in the development of this formula. The most important one is the speed. The formulas was developed for vehicles that travel with an average speed, average, between 10 miles per hour up to 60 miles per hour, if I remember correctly. But 10 is the main one. There is no way you can claim a forklifts have an average speed of 10miles per hour. It's stop and go. It's — Q. Does AP-42 apply to stop and go traffic? A. No. It explicitly say, the formula that Ms. Sears applied, the formula should not be applied for stop and go traffic. So it is very questionable, the use of AP-42 for forklifts. And again, the most important thing, there is no literature support for doing that. I haven't found one single study. (RT, Oct. 14, 2010, 715:15-717:1.)

During counsel's re-direct examination on October 14, 2010, the fourth day of Daubert hearings, Ms. Sears testified that AP-42 is appropriate in these circumstances given the lack of an emission scenario for forklift use at industrial sites:

A: Now, I think it's important to describe that — the process that I used to calculate the air emissions from the fugitive dust, the treated wood storage area. I used an EPA factor. It's from a document called AP 42, which is number 42 in an air pollution document series prepared by US EPA. And it's a compilation of emission factors. Which are techniques used to calculate air emissions from various types of activities. And in this case, Section 13.2.1 deals with fugitive dust from paved roads or parking lots. It's a technique that can be used to any paved road or parking lot. There's nothing unique about fugitive dust coming off a paved surface area at a wood treatment plant or a treated wood storage area paved surface. It's the same technique. And I used that same technique scores and scores and scores of times. I've used it for refineries, I've used it for oil and gas processing facilities, for steel mills, for quarries, or battery recycling plants. The list goes on. I've done it scores and scores of times. And there is a very simple, I think, a very simple determination based on my experience that it would also apply to this facility. Q. You mean the type of surface from which fugitive dust would be sitting upon and then emitted through some type of activity? A. Right. What happens is that you have an amount of surface dust on the surface of the soil. That then can be released into the air due to mechanical disruption by vehicular tires, mainly. And the vehicle doesn't matter, it's the weight of the vehicle that matters. The fact that Ms. Sears used a different algorithm to calculate fugitive air emissions than Dr. Zanetti does not render her unqualified nor does it suggest her emissions scenario is invalid. Rather, based on the current record, her emissions scenario is relevant, admissible, and can be challenged through cross-examination and presentation of contrary evidence. In this Circuit, an expert's decision to use one form of scientific methodology over another goes to the expert's credibility rather than the admissibility of the testimony. See, e.g., United States v. Garcia, 7 F.3d 885, 889-90 (9th Cir. 1993). This is such a case.

In addition, the Fed.R.Evid. 706 air experts, Drs. Countess and Cowherd, agree that there is a scientific basis to use AP-42 as an emission factor equation in this case. The relevant portion of the joint expert report provides, in relevant part:

We believe that Sears has made reasonable scientific judgment in selecting the paved road equation to project emissions from forklift traffic in the TWSA. The errors associated with applicability problems (low speed, "stop and go" motion) are probably small compared with errors and uncertainties associated with other items in her analysis as detailed below. If a separate equation were available for application to stop and go activity, it would be advisable to use it.

(Doc. 942, pg. 12.)

Contrary to Defendants' contentions, the application of the "paved road" emission scenario was within Ms. Sears' realm of expertise as an air modeler and meteorologist. She has selected emission scenarios and calculated fugitive emissions for over twenty years. Ms. Sears has the expertise to use scientific judgment that AP-42 was the most accurate emission scenario for the former-BAC Site. This is especially true given the lack of a scientific algorithm designed explicitly for the types of industrial use and emission estimates required in this case. Dr. Zanetti's contention that only "unpaved surfaces" at the BAC site can be modeled is unreasonable given the 706 experts' contrary opinions. He offers no alternative to his unpaved model, citing only the "lack of scientific literature" to support Ms. Sears' AP-42 emission scenario.

Given the unique facts of this case, Dr. Zanetti's refusal to model over 95% of the alleged source area on the basis of a "complete match" is puzzling.

The review of Sears and Zanetti's expert reports and deposition/hearing testimony, as well as the opinion of the Court's Fed.R.Evid. 706 experts, demonstrate that although the use of AP-42 to calculate fugitive emissions from the wood storage area of the former-BAC Site is sharply disputed, a disagreement over methodology is left to the adversary process and the trier of fact. The methodology here is not comparable to that found unreliable in In re Voluntary Purchasing Groups, Inc. Litigation, 2000 WL 1842779 (N.D. Tex. 2000), a case excluding an air modeler's expert testimony under Rule 702 and Daubert.

Defendants' remaining arguments focus on the total mass calculation and the input parameters of Sears' model, i.e., drippage accumulation, the particle size, valence/purity of the contaminant and silt loading values. However, several of these inputs are functions Dr. Agardy's total mass calculation, which is not challenged and admissible:

Rather than estimate the amount of hexavalent chromium and arsenic at the former BAC wood storage area, I calculated the amount of elemental material contributing to silt loading (in grams per square meter) based on Dr. Agardy's drippage calculations. Dr. Agardy's calculations are not subject of a motion to exclude [. . .] [Dr. Agardy's] drippage calculations were conservative because they were limited to the chromium levels found in treated wood storage area soils. These levels could only have result from drippage of the Cr VI [hexavalent chromium] from treated wood that subsequently entered the soil. My calculation did not take into account the amount of contaminants that never entered the soil.

Experts commonly rely on material prepared by others to develop their findings and ultimate conclusions. See Banks III, 75 Fed.Cl. at 304; see also United States v. Smith, 869 F.2d 348, 355 (7th Cir. 1989) ("[I]t is well settled that expert witnesses may rely on material commonly used by others in the field, even if those materials were prepared by others.").

(Doc. 785, C. Sears Decl., at ¶ 50.)

The majority of Dr. Agardy's opinions are proper as they relate to the use of chemicals and chromium at the BAC site. These figures were a basis for Sears' model and are properly supported by Agardy's qualifications and expert reports. However, in his "new" declarations, (Docs. 782 955), Agardy opines on subject matter outside of his expertise: chemistry. Agardy is not qualified to opine on the conversion rates between Chromium III and Chromium VI. A review of his declaration reveals that Agardy simply quotes from several leading chemistry treatises and hornbooks, without establishing his own expertise to do so. Agardy's opinions on the chromium conversion are questionable. He is, however, qualified to opine on the total chromium mass at the former-BAC Site.

The final dispute concerns the scientific reliability of Sears' input parameters: the drippage accumulation; the silt loading values; the valence/purity of the contaminant; and the particle size (10 microns vs. 100 microns). Sears contends that her input parameters conform with well-recognized scientific principles. She asserts that the input parameter calculations are based on her scientific knowledge, experience and familiarity with air modeling, as well as a review and analysis of historical information at the BAC Site, Dr. Agardy's chromium drippage calculations, USEPA air pollution emission factors for paved roads, environmental and wind data from the site, and a review of pictures of the BAC Site and industrial vehicles used at the site. Defendants rejoin that these four input parameters lack scientific

Sears defines "silt loading" as the "amount of silt, which is on this 200mesh or less than 75 micrometers in diameter. Measured per square meter of surface. And the units are usually in grams per square meter. And it's a standard input to paved road fugitive dust emissions." (Rough RT, Dec. 2, 2010, 168:8-168:12.)

Sears defines "valence" as "for hexavalent chromium, it's plus 6. For chromium trioxide you've got the valence of both chromium and oxygen. You've got three oxygens, those are going to be two minus each. And then you've — so you've got three oxygens that's minus 6. To balance it out, you need a plus 6 from the chromium. So chromium, trioxide, the valence of the chromium is always going to be plus 6." (Rough RT, Dec. 2, 2010, 179:9-179:16.)

reliability and are inadmissible to generate a factual dispute.

Drs. Countess and Cowherd's Fed.R.Evid. 706 expert report provides a starting point to analyze the dispute over the input parameters. The Fed.R.Evid. 706 expert opinions are reproduced below, contrasted with the Daubert testimony of Ms. Sears. The expert report and opinions of Drs. Countess and Cowherd cast considerable doubt on Ms. Sears' silt loading, valence, accumulation and particle size inputs. Under Rule 702 and Daubert, however, the proper analysis is not whether some of the inputs can be questioned, but whether Sears' testimony is relevant and reliable, and whether the methods and principles upon which she has relied in forming her opinion have a sound basis in science. In this case, a comparison of the relevant expert opinions demonstrates a reasonable scientific dispute over input parameters and scientific principles.

First, Drs. Countess and Cowherd's independent expert report characterizes Sears' silt loading value as "seriously flawed":

Sears's assumptions about silt loading are seriously flawed. There is no precedent for assuming that the silt loading is comprised almost entirely of Cr +6 [hexavelant chromium] and arsenic. Sears should have based her estimates of silt loading on chromium trioxide and arsenic pentaoxide as well as (at the very least) deterioration of the asphalt pavement that would produce particles in the silt size range.

(Doc. 942, pg. 15.)

On December 2, 2010, on direct examination, Sears explained that her silt loading values comported with "standard modeling practices":

In her declaration, attached to Plaintiffs' opposition, Sears reinforced her core assumptions and the steady-state silt loading value used in her model, which was the subject of extensive disagreement between the parties:

Defendants attack my steady-state silt loading value. For plant operating years, I have assigned a steady-state silt loading value equal to the yearly drip rate. I did not accumulate the silt loading value after a number of years of wood treating drippage, even though that is a possibility. The steady-state silt values I used for hexavalent chromium and arsenic at the former BAC site are entirely consistent with the steady-state silt loading values I have used in every USEPA paved surface fugitive dust emission calculation that I have performed. I've also reviewed these same emission calculations made by dozens of industrial source consultants, and I cannot think of one instance where steady-state silt loading values were not used [. . .] I believe the steady-state silt loading values I calculated are reliable and represent a better method than simply estimating the otherwise unknowable value that once existed at the former BAC facility.

(Id. at ¶ 52-53.)

Q. With respect to the assumptions and the input assumptions with respect to silt loading, why do you believe that the technique, the method you used and the assumptions you utilized were in keeping with generally accepted standards and principles? A. Again, the silt loading values that I used were based on Dr. Agardy `s drip calculations. Which I believe he performed correctly. And I talked to him about them. There was just no black box that was handed to me the day I ran the model. We had a lot of communication. The generally accepted standard and approach is to use a steady state value. Even when you're modeling short — term air exposures. I've been reviewing dozens, like I said, of these coal fired power plant permit applications, steel plant, pig iron plant applications and they all use the same steady state value. And I'll give another example. I recently was asked by the California Attorney General's office to help them review potential proposition 65 case. The issue was hexavalent chromium and arsenic from fugitive dust at the Lehigh Cement Plant in Cupertino. They wanted to know whether they should intervene because some citizens had filed a 60 day notice of requirements to warn under Proposition 65. So they asked me to look at it. The producer of this report was a AMEC Geomatrix, which has been a consultant to Merck in other matters in this case. And they did the exact same approach that I'm talking about. They — even though there was an existing site, they didn't measure the silt loading. They estimated it. They estimated what it would be based on the average for a cement plant. They could have measured it. But the key point is that they used a steady state value for the life of the exposure assessment to run 70 years. And that's generally accepted scientific standards and practice for silt loading. You don't vary it. To do so would be beyond that, to do so would be holding me to basically unprecedented and unnecessary standard. Unnecessary because we're calculating long-term average emission rates. (Rough RT, Dec. 2, 2010, 173:14-175:1.)

With respect to the "accumulation" factor the Fed.R.Evid. 706 experts provide, in relevant part:

Sears's assumption of a one-year accumulation period is unreasonable because it assumes that no losses such as runoff from the paved surface due to rainfall occur during that period. Because of the solubility of the contaminants in original form, any precipitation would undoubtedly cause the contamination to wash off the paved surface and/or penetrate the thin layer of asphalt into the soil below the paved surface, so that the equilibrium time would be much shorter, especially during the rainy season. Our review of Merced Municipal Airport precipitation records from 1948 to 2006 (Court Exhibit A-17, Folder Q) indicates that the wettest months are November through April with 5-10 days per month with at least 0.01 in. of precipitation. During this six-month period there are 3-6 days per month with at least 0.1 in. of precipitation, and an average of one day per month with at least 0.5 in. of precipitation. The driest months are May through October with an average of one day per month with at least 0.01 in ofPGPage 46 precipitation. Thus, we conclude that a more reasonable accumulation period would range from one month to six months for the six-month period May through October, and a maximum of several weeks for the six-month period November through April.

(Doc. 942, pg. 17.)

Although the one-year accumulation rate is allegedly unreasonable, Sears testified at the Daubert hearings that her figure was an "average" and that "it's going to vary from year to year":

A: Accumulation, I think, is one of the defendants' words that popped up in their motion, Daubert motion. I think they're very related. So if you're going to talk about accumulation, I'll wait until you get do that point [. . .] Accumulation is really another word for silt loading. It's how much of this material is on the surface of the paved parking area, paved driving area of the treated wood storage area. As I acknowledge in my reports, it's going to vary from year to year. It's going to vary from time to time. However, in AP-42, the emission rates are generally almost [steady state], I can't think of even one time where an industrial source emission rate was modeled anything other than steady state. Q. What does that mean? A. It doesn't change. Q. Okay. A. You use an average. And the — what I did in the Daubert hearing and it's described in more detail in my declaration, is that I showed that the silt loading, and in this case chromium that I used in grams per square meter over the life of the project, 22 years, when mixed into the underlying soil matches remarkably close to the measured chromium levels, seen in 1999. And so it shows to me that the input that I used, which is really just trying to explain how the chromium got into the level and the soil down below is entirely reliable. And the issue of it varying from month-to-month or year to year, since we're calculating long term cancer risk exposures, that's the end result of these things. The month-to-month exposures don't matter. Because these are long-term exposures that cause the health effects. (Rough RT, Dec. 2, 2010, 172:9-173:10.)

Drs. Countess and Cowherd also characterize Ms. Sears' particle size values as "seriously flawed":

Sears calculated the silt loading as consisting of 90% contaminants combined with only 10% soil. Even if virtually all of the contaminants on the surface consisted of particles smaller than 75 microns, emissions from the surface would span a wide range of particle size. Only emissions of particles smaller than about 30 microns are normally categorized as suspended particulate matter. The emissions of airborne particles in specified particle size ranges including PM10 can be estimated using the AP42 equation. AP42 indicates that the PM10/PM30 ratio in the emissions generated by vehicle travel on paved surfaces is 0.19. Because the PM10 emissions encompass a particle size range extending from less than 1 micron up to 10 microns, it is problematic to assume PM10 emissions can be represented as 5 micron particles in the dispersion process. In addition, because the settling tendency of particles increase with particle size, the ratio of contaminant lost by deposition from the emission plume to that remaining airborne across the community would be much larger than that estimated by Sears [. . .]
Sears's assumed particle size for airborne hexavalent chromium and arsenic particle is seriously flawed.

(Doc. 942, pg. 15.)

With respect to Drs. Countess and Cowherd's particle size criticisms, Ms. Sears responds:

A. Particle size is an input for deposition, the fallout of particles from the plume as the air travels off site. Typically particles — deposition is not calculated in most anti-air quality impact analysis. But not to include it would tend to overestimate the air impacts. So I included the particle size for dry deposition to basically properly account for the plume depletion that takes place due to material falling out. Q. Did AP-42 mandate that particle size be included as a factor? A. No. It's not an input in AP-42 at all. It's an input to AERMOD, which is the dispersion model. Q. If it wasn't in A P-42, why is it that you made certain assumptions regarding particle size? A. I made the particle size analysis to account for plume depletion in running the air dispersion model. And the — you don't need to know the particle size distributions in AP-42, emission calculations. All you need are the silt loading. You need to know the vehicle weight, the vehicle miles traveled. Now — and the K factor we've been using for PM10 or PM30. And I think this is where I strongly disagree with the 706 air experts. They said that my particle size for dry deposition was seriously flawed. Now, I thought, well, how could they say that? Even Dr. Zannetti, from questions from [defense counsel] said that my analysis was fine. Because he understands, I believe, that it's not a sensitive input. Q. Why isn't it a sensitive input? A. I ran sensitivity analysis in my — and it's discussed in my Rule 26 report. Again, I'm modeling PM10. That's what we're — that's what we're limiting here to, because those are the size particles that cause inhalation or get into the lungs. That cause the problems from the hexavalent chromium exposure. It's the small particles. So I limited the exposure to that size. Generally, we don't have particle size distributions, we don't have measurements. Again, we don't have measurements from this plant at all because it closed 20 years ago and no one ever took the time to measure the data back then. So I said, okay, I've done a lot of this before. I — five microns is — is — I've used it before and it's not been any problem. But just to make sure, I'll look at two and a half, I'll look at seven and a half. I'll look at a combination of 2.5, 5 and 7 and a half microns. And then see what effects that has on the impacts. Trivial. I actually did an analysis where I didn't even include any particle sales for dry deposition at all. It increased the impacts, air impacts by about 15 percent. So if anything, including the particle sizes tends to lower the air concentrations. And so I thought, okay. I've done this sensitivity analysis, it's trivial. I never expected anyone to have any problem with it. And I thought it was interesting that they — that these 706 air experts disagree with both Dr. Zannetti and myself, who I believe have had more air modeling experience. Q. Again, with respect to particle size, based upon everything you testified, you believe that your approach was well within generally accepted standards for that part of the project? A. Yes, I do. And I took it a step beyond and did the sensitivity analysis and the probability distribution and an uncertainty analysis. I covered all the basis. (Rough RT, Dec. 2, 2010, 176:2-179:5.)

Drs. Countess and Cowherd are critical of the assigned valence values:

Agardy states that one should be concerned with the amount of amount of Cr+6 in drippage not whether Cr+6 is converted to Cr+3 in treated wood. Sears assumes that 100% of the Cr in the material that drips on to the ground is Cr+6 based on her discussions with Agardy. This is reasonable for freshly dripped CCA or ACC preservative and drippage that doesn't come into contact with organic material, or iron — or sulfide — containing soil deposited on a paved surface. Furthermore, there is no disagreement between the experts for both parties that Cr+6 in the treated wood can be converted to Cr+3 as the treated wood ages. Thus, if there is any transfer to the ground of sawdust created from cutting treated wood or crystallized preservative formed on the outside of treated wood, the paved surfaces at the plant will contain Cr+3 from these sources as well as Cr+6 from drippage. [Note: The shallow soil samples from the TWSA indicated that most of the chromium in the soil was present as Cr+3; the Cr+6/Cr ratio for these samples before correcting for background was 0.0051.]
Sears's assumption that 100% of the Cr in the preservative that drips on to the ground is Cr+6 is reasonable for fresh deposits but not for preservative that has been in contact for several weeks or longer with soil deposited on a paved surface.

(Doc. 942, pg. 18.)

Ms. Sears rejoined that her valence value conformed to scientific values and, in any event, the value "subject to an overall emission rate calculation uncertainty analysis" to account for variability:

Q. What is your understanding of what criticism the 706 report had of your input assumptions with respect to valence? A. They were concerned that, as I understand it, that the assumption that I made, the input that I made of 100 percent hexavalent chromium was only valid for fresh deposits of chromium trioxide. And over time, there would be some degradation, some reduction to trivalent form. And in the — on the surface of the paved road area of the treated wood storage area. Q. Do you disagree with that? A. I do. Q. Why? A. For a number of reasons. And they're outlined in number 8. Now, I've been intimately involved in developing the risk assessment guidelines for the State of California since 1989, 1990. That time frame. And I prepared hundreds of these risk assessments, many of them, scores of them included hexavalent chromium. The standard process is not to convert hexavalent chromium to trivalent chromium in these risk assessments. And I discussed that in the California Air Resources Board and the California office and environmental health hazard assessment procedures. The only time that you account for a reduction of hexavalent chromium to trivalent is under certain soil conditions, where you have the measured data to predict, to calculate how that conversion will take place. We don't have that here. For two reasons. One, we're talking on the surface of the paved area, of the treated wood storage area. And second, they don't have any other data underground to calculate that conversion process. Not that I know of. And again, underground is not my concern. I'm concerned about the top level of the paved area and the treated wood storage [. . .] But again, as I talked about earlier [] we have the 1994 IT Corporation Data Gaps study measurements of hexavalent and total chromium in the top one foot underneath the treated wood storage area that shows that accounting for background, Dr. Agardy's lower background, the hexavalent chromium to total chromium ratio is still about 80 percent. Three years after they stopped treating. So if it's 80 percent hexavalent chromium in the soil three years after they stopped using it, where, underneath the soil it would be under to leaching — again, hex chrome is more soluble than trivalent, so it would tend to disappear more than the trivalent would and tend to go to lower levels. And in the soil, that's where you have, if any, capability of reduction from hexavalent to trivalent. But after three years, if you're still seeing such a high hex chrome to total chrome ratio, what other conclusion can you draw than it was basically 100 percent during the time that the plant was operating? Another thing to consider too is that this wasn't a one-time dump of chromium trioxide on the soil. This was constantly being replenished. Every day they were treating more wood, every day there was more drippage. So that silt loading is constantly being replenished with fresh chromium trioxide. But the 706 air experts say unquestionably 100 percent hexavalent. So I think my assumption of 100 percent hexavalent, if I did anything other than that, would be inconsistent with California risk assessment guidelines and would be inconsistent with the data [. . .] And again, all of these things are subject to an overall emission rate calculation uncertainty analysis that I did to account for any variability and sensitivity of these parameters. I was very careful. I covered all these basis. And I'm surprised that the 706 air experts didn't look at the same data that I did. (Rough RT, Dec. 2, 2010, 179:17-183:6.)

The comparison between the relevant expert opinions demonstrates that a reasonable scientific disagreement exists among the experts. Ms. Sears input parameters raise a host of serious concerns, but do not strip the model of all scientific reliability. See, e.g., Renaud v. Martin Marietta Corp., 749 F. Supp. 1545, 1552 (D. Colo. 1990) ("Most of these alleged methodological flaws, however, must be considered by a jury when evaluating the weight to be given to the opinions, rather than by the Court in making the preliminary assessment concerning the admissibility of a given opinion.") (citations omitted). Such a dispute goes the credibility and the weight of the opinions, not admissibility.

Sears' allegedly flawed drippage accumulation calculation is illustrative. Defendants argue that Sears has no basis to given an opinion that a year's worth of drippage will accumulate at the surface; such a value contradicts reality as it assumes no disturbance or rainfall at the BAC site. On this point, Defendants submit the testimony of Dr. Fendorf, an expert in chromium and arsenic and a geochemist. Fendorf opines that hexavalent chromium is highly soluble in water and any significant rainfall dissolves hexavalent chromium at the surface and carries it into the soil.

Ms. Sears testified that the figure "is entirely reasonable" and supported by proper science, in addition to being an "average" accumulation rate. In this Circuit, the opposing party to a Rule 56 motion need not establish a material issue of fact conclusively in its favor. It is sufficient that "the claimed factual dispute be shown to require a jury or judge to resolve the parties' differing versions of the truth at trial." First Nat. Bank of Ariz. v. Cities Service Co., 391 U.S. 253, 290; Hopper v. City of Pasco, 248 F.3d 1067, 1087 (9th Cir. 2001). A jury could consider the parties' competing expert testimony and reasonably conclude that chromium did not dissolve into the soil but stayed on the surface, reinforcing Sears' testimony (and her drippage calculation). The determination of the credibility and weight of the expert's conclusions should be reserved for the trier of fact.

The particle size dispute provides an additional example of a scientific factual dispute. Dr. Zanetti, an air modeler retained by Defendants, testified that the "five micron [size] [i]s not a problem," while Dr. Fendorf, another defense expert, stated in his declaration that "the chromium-arsenic salt particles that would be left behind by that evaporative process would not be in the size range that Ms. Sears assumes [. . .] [i]nstead, they would be sand-sized particles greater than 100 microns in diameter." (Doc. 722-1 at ¶ 5-7.) Drs. Cowherd and Countess provide yet another opinion, that it is "problematic to assume PM10 emissions can be represented as 5 micron particles in the dispersion process" because the PM10 emissions encompass a particle size range extending "from less than 1 micron up to 10 microns" and "the settling tendency of particles increase with particle size." In the scientific community, experts reach different conclusions from each other, however, reasonable differences in scientific evaluation are not a basis for exclusion. See Daubert I, 509 U.S. at 596 ("[v]igorous cross-examination, presentation of contrary evidence, and careful instruction on the burden of proof are the traditional and appropriate means of attacking shaky but admissible evidence").

Having observed the experts at the Daubert hearing under intensive cross-examination by counsel, it is clear that the current dispute over Ms. Sears' calculations should be determined by the trier of fact, not by the Court on a motion to exclude. In re Scrap Metal Antitrust Litig., 527 F.3d 517, 529 (6th Cir. 2008) (the "rejection of expert testimony is the exception rather than the rule.") (citation omitted). The decision to view Sears' differing analytical inputs as weaknesses in her analysis are a matter of weight rather than admissibility and thus properly a subject of argument and jury judgment. There is an insufficient showing of unreliability to exclude Ms. Sears' entire air model.

Finally, Defendants support their motion with a citation to air quality samples taken during excavation activities at the BAC site in 2007 and 2008. According to Defendants, Arcadis excavated the contaminated soil from the former retort area and installed four monitoring stations to detect any dust that might blow off the site during the excavation. Arcadis allegedly did not detect any chromium or arsenic-containing dust at the monitoring stations. This evidence, however, does not prove/disprove Sears' testimony concerning the air pathway for the years 1969-1993. Rather, it is one of many facts that may presented to the jury on the Phase 1 issue of whether soil contaminated with hexavalent chromium and/or arsenic was transported to Plaintiffs' homes or properties.

4. Risk Assessment (post-1993 years)

Plaintiffs claim that the 1994 Risk Assessment demonstrates that Plaintiffs in the Beachwood neighborhood were exposed to unacceptable airborne risk factors. Merck produced the Risk Assessment of the BAC site pursuant to an order from the California Department of Toxic Substances Control. Plaintiffs allege that the Risk Assessment satisfies their Phase 1 burden for the post-1993 years:

[W]ithin this Risk Assessment, is confirmation of completed pathways with regard to airborne particles exposing residents to contamination via wind, dust, home gardens and home grown fruits and vegetables. Notwithstanding this knowledge of a completed pathway of airborne exposures to these residents, the Merck defendants made no effort to determine whether or not such exposures could have existed prior to 1994, and specifically in relation to the BAC facility when it was in operation. Indeed, there is no evidence that the Merck defendants at any time performed or even requested permission to perform any testing of residential soil, residential home gardens, residential fruit trees, swab samples of attics or indoor locations, or other sampling or testing which would have confirmed whether these vulnerable residents had been exposed to airborne contaminants during the operational history of the plant. This risk assessment shows unacceptable risk factors to plaintiffs in the Beachwood neighborhood.

(Doc. 792 at 26:3-26:19.)

Plaintiffs do not respond to the Defendants' string citation of authorities holding that "risk assessments are intended to provide conservative estimates of risk and offer a wide margin of protection for human health [thus] federal courts have refused to accept their results as evidence of causation." See, e.g., Rhodes v. E.I. DuPont de Nemours and Co., 253 F.R.D. 365 (S.D.W.V. 2008); see also Gates v. Rohm Haas Co., 265 F.R.D. 208, 226 (E.D. Pa. 2010) (collecting cases).

In their motion for partial summary judgment, Defendants summarized the key portions 1994 Risk Assessment and the reasons why federal courts have excluded such risk assessments to demonstrate causation/exposure:

Plaintiffs contend that the 1994 risk assessment provides evidence of exposure to wind-borne dust for the years 1994 to the present, but this contention has no merit for at least two independent reasons. First, the risk assessment assumed that nothing further would be done to clean up the Site in the years after 1994 (and indeed that the old asphalt on the Site would degrade, exposing soil to the air). However, in 1996, Merck and Amsted excavated contaminated soil from the retort area and installed a four-inch asphalt cap over the entire area around the location of the former pressure-treating facility. The risk assessment specifically noted that this remedial measure would "effectively eliminatethe exposure pathways (and subsequent risk)." Thus, even if one assumed that the risk assessment could somehow be used to predict exposure in the years after it was written, the risk assessment predicted that with a cap over the retort area, there would be no exposure and therefore no risk.
Second, as the risk assessment repeatedly makes clear, it intentionally overestimated potential exposures and risks in order to be protective of human health:
• "The risk numbers are considered overestimates of the potential risks . . .";
• "This assessment employs a health-protective bias that leads to the overestimation of risks rather than realistic evaluation or an underestimation";
• "To minimize the effect of uncertainties in the evaluation, each step is biased towards health protective estimations. Because each step builds on the previous one, this biased approach should more than compensate for risk assessment uncertainties. In addition, these calculations do not represent currently existing or expected future exposure or health risks. Rather, they are estimates of potential risk only if all of the conservative assumptions are realized."
In short, the risk assessment was based on a hypothetical scenario that never occurred, and it does not purport to accurately calculate actual exposures or risks.

(Doc. 677-1 at 22:15-23:19.)

In Rhodes v. E.I. DuPont de Nemours and Co., 253 F.R.D. 365 (S.D.W.Va. 2008) the court held that risk assessments were of "limited utility" in a toxic tort case based on the risk assessment's distinct regulatory purpose:

Risk assessments have largely been developed for regulatory purposes and thus serve a protection function in providing a level below which there is no appreciable risk to the general population. They do not provide information about actual risk or causation. See Bernard D. Goldstein Mary Sue Henifin, Reference Guide on Toxicology in Federal Judicial Center Reference Manual on Scientific Evidence 413 (2000). "Because of their appropriately prudent assumptions when there are limited data, risk assessments intentionally encompass the upper range of possible risks." Id.; see also Sutera v. Perrier Group of Am. Inc., 986 F.Supp. 655, 664 (D.Mass. 1997) (rejecting regulatory standards as a measure of causation because the purpose of regulatory standards is to "reduce public exposure to harmful substances.") (quoting Allen v. Pa. Eng'g Corp., 102 F.3d 194, 198 (5th Cir. 1996)); O'Neal v. Dep't of the Army, 852 F.Supp. 327, 333 (M.D.Pa. 1994) (determining that risk figures based on the EPA's upper-bound estimates for another chemical are "appropriate for regulatory purposes in which the goal is to be particularly cautious [but] overstate the actual risk and so, are inappropriate for use in determining whether medical monitoring should be instituted.").
Id. at 377.

Rhodes concluded:

Because a risk assessment overstates the risk to a population to achieve its protective and generalized goals, it is impossible to conclude with reasonable certainty that any one person exposed to a substance above the criterion established by the risk assessment has suffered a significantly increased risk
Id. at 378.

Rhodes' reasoning is sound. Here, the express purpose of the 1994 Risk Assessment was to provide a regulatory body, the DTSC, with conservative estimates of risk and provide a wide margin of protection for human health. It does not establish general exposure. It is not particularized nor does it include an

identifiable risk of exposure to the Beachwood residents.

Plaintiffs have not provided any legal authority finding a Risk Assessment relevant to the issue of general exposure. Plaintiffs do not address the case law or why/how the Risk Assessment is relevant to prove actual causation to appreciable levels of contaminants from the BAC site. Under the reasoning of Rhodes and Gates v. Rohm Haas Co., 265 F.R.D. 208 (E.D. Pa. 2010), the motion for partial summary adjudication is GRANTED as to the 1994 Risk Assessment, which does not establish general exposure or otherwise satisfy Plaintiffs' Phase 1 burden as to the air pathway.

5. Conclusion on Air Pathway

The qualifications of Plaintiff's air model expert, Camille Sears, and the reliability of her methodology and assumptions are materially at issue in this case. Drawing all inferences in favor of the non-movant, Ms. Sears' testimony presents a scientific dispute not resolvable as a matter of law. Defendants' Daubert motion is DENIED and that portion seeking to adjudicate air contamination is also DENIED.

With respect to the 1994 Risk Assessement, Defendants' motion for partial summary judgment is GRANTED. The express purpose of the 1994 Risk Assessment was to provide a regulatory body, the DTSC, with conservative estimates of risk and a wide margin of protection for human health. It does not establish general exposure or otherwise satisfy Plaintiffs' Phase 1 burden as to the air pathway.

D. Pathway # 2 — Groundwater/MWC-2

The parties' groundwater pathway arguments focus on whether contaminants from the BAC site reached Meadowbrook Well No. 2, the only well alleged to have been contaminated with chemicals from the BAC site. As explained above, scientific sampling was performed at MWC-2 by Meadowbrook Water District from 1964 to 2007. Monitoring was then performed by Arcadis, an environmental consultant (2007-2009). It is undisputed that MWC-2 was taken out of service in March 2008. A material dispute exists as to the reliability and accuracy of the sampling data from the network of monitoring wells and several extraction wells near the BAC Site.

The EPA defines a monitoring well as: "A well used to obtain water quality samples or measure groundwater levels." See (last visited Dec. 29, 2010). Monitoring wells are denoted herein with a "MW #".

According to the EPA's website, extraction wells are "discharge well[s] used to remove groundwater or air." See (last visited Dec. 29, 2010). More specifically, extraction wells are installed to "clean the groundwater and control the migration of contaminated water." Cross Petroleum, Inc. v. United States, 54 Fed. Cl. 317, 324 (Fed. Cl. 2002). Monitoring wells are denoted herein with a "EW #".

Plaintiffs' Exhibit No. 11, introduced on December 28, 2010, demonstrates the importance of the monitoring and extraction well data to the analysis: "Data relied upon and testimony related to Bartlett's opinions regarding modeled results compared to observed data [. . .] EW 4 sampling data is consistent with the model [] EW 5 is sampling data is consistent with the model [] MW 23 had its first high level (450) in June 2008 [. . . .]".

Plaintiffs offer the expert testimony of Douglas Bartlett, a hydrologist and groundwater modeler who opines that hexavelant chromium from the BAC Site entered and contaminated the water pumped by MWC-2 for at least a period of twenty-five years. According to Bartlett's groundwater model, hexavelant chromium reached the MWC-2 in mid-1972 in concentrations exceeding 10 ug/l. MWC-2's contamination levels allegedly increased between 1972 and 1994, reaching a peak value of 389 ug/l in June 1994. From that point on, according to Bartlett's model, hexavelant chromium concentrations decreased. Critical to satisfying the Phase 1 burden, Bartlett reports that hexavelant chromium was present in MWC-2 in excess of the federal (100 ug/l) and state (50 ug/l) Maximum Contaminant Level ("MCLs") for more than twenty years.

According to Defendants, Bartlett's testimony and groundwater model are inadmissible for a number of reasons but primarily because he excludes 46 years of sampling data (49 tests) from the MWC-2 and 16 years of sampling data from monitoring wells surrounding the BAC Site. Characterizing Bartlett's model as "contradicting reality," Defendants claim that Plaintiffs "have produced no data, documents, or percipient witness testimony that could establish exposure to hexavelant chromium or arsenic from the BAC Site through MWC-2 water, and the results of analyses of water from MWC-2 refute plaintiffs' claims that such exposure occurred." (Doc. 677-2 at 10:13-10:16.) Defendants assert that the actual testing data shows that total chromium concentration at MWC-2 never exceeded 12 ug/l, less than one quarter of the MCL for total chromium. Defendants also argue that MWC-2 never captured hexavelant chromium beyond background levels.

Defendants do not dispute that Bartlett is qualified to serve as an expert on groundwater modeling. Bartlett has been employed in the fields of hydrogeologic movement and soil investigation since 1977, focusing on groundwater contamination since 1984. Bartlett's employment history includes working as a consultant with Dames Moore in Denver, Colorado from 1977 through 1989. In 1989, Bartlett moved to Phoenix, Arizona, where he managed several groundwater contamination projects for Dames Moore. Several of these projects involved CERCLA remediation and groundwater modeling, including what is referred to as the "Motorola 52nd St. CERCLA site." In 1999, Bartlett left Dames Moore to form the consulting firm "Clear Creek Associates." As a principal at Clear Creek, Bartlett provides groundwater modeling services to private and public entity clients throughout the United States.

Bartlett obtained his M.S. and B.S. degrees in geology from Colorado State University (1984 1977).

According to his original and rebuttal reports, filed pursuant to Federal Rule 26(a)(2), Bartlett provides the following opinions: (1) the BAC Defendants' wood-treating operations introduced 12,913 pounds of hexavelant chromium to the groundwater over a twenty-five year period; (2) from the BAC site the hexavelant chromium migrated, pursuant to an established pathway, to MWC-2 and was extracted from that well with pumped water; and (3) Bartlett predicts that hexavelant chromium concentrations exceeded 50 ug/l in mid-1975 and peaked at 389 ug/l in June 1994.

This "mass" chromium calculation is taken from Dr. Agardy's calculation, i.e., Dr. Agandy's opinion that a total of 12,913 pounds of chromium was released from the BAC Site during the relevant time period. BAC Defendants do not separately move to strike Dr. Agardy's opinions under Daubert.

Bartlett's relies heavily on the fact that his model is based on and derived from an earlier groundwater model developed by Arcadis, an environmental remediation firm contracted by Merck. According to Bartlett, the Arcadis' model evolved from a previous model created by IT Corp. in the 1990's. Arcadis, however, revised IT's model in the following ways: (1) expanded the model domain to include MWC-2 and nearby wells; (2) incorporated seasonal pumping from nearby production wells, including MWC-2 and Bernardo Dairy Well; (3) expanded the number of "well layers" to simulate various hydrogeologic strata at the site; and (4) hydraulic conductivity distribution was set in "uniform" layer mode based on slug tests, aquifer tests by Shaw, and specific capacity data. Arcadis ran the model in "steady-state" mode, which, according to Bartlett, is "essentially a point-in-time prediction rather than a prediction over a time period."

Bartlett's groundwater model makes a number of significant changes to the Arcadis model. According to Bartlett, these revisions were necessary because the "Arcadis model did not adequately simulate MWC-2 operations." Below is a summary of Bartlett's revisions and the reasons thereto:

1. Layering/Cell Exposure — The Arcadis model was based on a 20 foot by 20 foot size calculation cells, with five layers simulated (94,875/cell). Bartlett increased the cell size to 50 feet by 50 feet and added two more layers to the model (to account for well depth); one representing a "finer grained horizon" from 95 to 105 feet (Layer 6), the other representing the bottom portion of the aquifer from 105 to 140 feet (Layer 7). Bartlett also imported actual elevations for each of the seven layers. In contrast, Arcadis assigned model uniform elevations (flat layers).
2. Hydraulic Conductivity — Bartlett opines that Arcadis underestimated the ease with which water can move through pore spaces or fractures at or around the BAC site, i.e., hydraulic conductivity. According to Bartlett, the figures were "low by a factor of at least 2." The hydraulic conductivity rate for the seven layers was adjusted to account for this discrepancy. Bartlett also assigned the vertical layer a value ten times less than the horizontal layer to reflect the decrease in permeability (caused by interbedded fine grained layers).
3. Boundary Conditions/Groundwater Flow — Bartlett explains that these factors "required extensive revision" because Arcadis's model was run in steady-state mode to reflect a single year, 2005. To evaluate groundwater flow over a number of years, Bartlett ran his model "transiently and boundary conditions [were] changed over time to reflect changes in regional groundwater conditions." The most critical change to the boundary was to assign a "general head boundary," which allowed the predicted head adjacent to the boundary cell to vary during the simulation depending on what stressors were introduced. According to Bartlett, GHB is a "better boundary condition for simulating pumping wells near a boundary because it essentially allows the model to predict an expanding cone-of-depression outside of the model domain and the model does not inadvertently introduce flow balance errors." With respect to groundwater flow, Bartlett submits that flow shifted in the northerly direction in the early 1990's (from a predominately westerly and southwesterly direction in the 1960's through early 1990's). GHB cells were assigned to "all four model boundaries" because the regional flow pattern changes dramatically throughout the simulation period (1969 to 2008, westerly/southwesterly to northerly).
4. Groundwater Pumping — Groundwater pumping was assigned to model domain wells known to intersect shallow strata representing one of the seven model layers. Bartlett noted that two wells, MWC-2 and the Bernardo well, lacked seals immediately above the casing perforations and thus both wells are capable of pumping shallow groundwater above the highest perforations. The vertical movement of groundwater through the open annulus between the steel casing and borehole wall allows the free flow of pumped water to the bottom of the well. In addition, extraction and recharge wells were installed at the site in the early 1990's and are included as part of Bartlett's simulation.
5. Stress Periods — A stress period is defined as a period of time when all boundary conditions including pumping wells, recharge wells, and boundary cells are unchanged. A stress period is further broken down into time steps that the model solves sequentially to predict changes in the model during the stress period. In the case of the revised model, stress periods were assigned the length of one month. A total of four model runs were needed to complete a simulation from 1969 through 2008.

Bartlett adopts several of Dr. Agandy's expert opinions to complete his chromium transport simulations. First, Bartlett acknowledges that other areas of the BAC Site may have contributed as chromium source origins, however, he adopts Dr. Agandy's opinion that the retort area was the primary source area. Second, Bartlett incorporates Dr. Agandy's opinion that a total of 12,913 pounds of chromium was released into the groundwater between 1969 and 1991, when the BAC facility closed. Bartlett then reports that the appropriate source of chromium in the groundwater model is based on the following factors: (1) hexavelant chromium was discharged at the surface and a portion was carried to groundwater by the natural infiltration of rainwater; (2) the hexavelant chromium required no more than one year to migrate to the water table; (3) the calculated total release of chromium occurred evenly between 1970 and June 1991; and (4) taking into account natural precipitation in the source area (1.14 inches × 22,500 sf = 16,000 gal/year). He also refutes Dr. El-Shoubary's "adsorption" theory, arguing that Dr. Argady's total chromium analysis takes into account any and all adsorption determinants.

BAC Defendants move to exclude Bartlett's testimony on grounds that his opinions ignore actual data in favor of unsupported modeling. They note that hexavelant chromium was measured 40 times in MWC-2 but never exceeded 2.5 micrograms per liter (ug/l or parts per billion) and that total chromium was measured 74 times but never exceeded 12 ug/l. In contrast, Bartlett's model predicts concentrations of hexavelant chromium up to 582 ug/l at MWC-2 during this same time period (1964 to 2009). BAC Defendants submit the following sampling from MWC-2, interposed with the relevant MCLs:Date Result Cal MCL Fed MCL

5/12/64 Trace — — 8/74 ND 10 — — 1/27/75 Trace — — 3/1/80 ND 5 50 ug/l 50 ug/l 4/30/80 ND 5 50 ug/l 50 ug/l 3/19/86 ND 5 50 ug/l 50 ug/l 6/9/87 ND 10 50 ug/l 50 ug/l 1/22/90 ND 10 50 ug/l 50 ug/l 6/90 ND 10 50 ug/l 50 ug/l 12/31/92 ND 10 50 ug/l 100 ug/l 3/9/93 ND 10 50 ug/l 100 ug/l 12/21/95 ND 10 50 ug/l 100 ug/l 2/7/99 ND 10 50 ug/l 100 ug/l 5/3/01 12.0 50 ug/l 100 ug/l 9/13/01 2.0 50 ug/l 100 ug/l 2/12/02 12.0 50 ug/l 100 ug/l 2/12/02 1.0 50 ug/l 100 ug/l 3/11/05 ND 10 50 ug/l 100 ug/l 2/27/07 2.1 50 ug/l 100 ug/l 2/27/07 ND 3 50 ug/l 100 ug/l 5/15/07 ND 3 50 ug/l 100 ug/l 5/16/07 1.8 50 ug/l 100 ug/l 9/18/07 ND 3 50 ug/l 100 ug/l 10/16/07 ND 3 50 ug/l 100 ug/l 10/16/07 1.6 50 ug/l 100 ug/l 11/19/07 ND 3 50 ug/l 100 ug/l 12/18/07 ND 3 50 ug/l 100 ug/l 1/16/08 ND 3 50 ug/l 100 ug/l 2/22/08 ND 3 50 ug/l 100 ug/l 3/21/08 ND 3 50 ug/l 100 ug/l 5/29/08 ND 3 50 ug/l 100 ug/l 6/30/08 ND 3 50 ug/l 100 ug/l 7/30/08 ND 3 50 ug/l 100 ug/l 8/20/08 ND 3 50 ug/l 100 ug/l 9/18/08 ND 3 50 ug/l 100 ug/l 10/23/08 ND 3 50 ug/l 100 ug/l 11/21/08 ND 3 50 ug/l 100 ug/l 12/23/08 ND 3 50 ug/l 100 ug/l 1/09/09 1.8 50 ug/l 100 ug/l 1/20/09 ND 3 50 ug/l 100 ug/l 3/19/09 ND 3 50 ug/l 100 ug/l 8/06/09 ND 10 50 ug/l 100 ug/l 8/11/09 2.3 50 ug/l 100 ug/l 8/18/09 2.5 50 ug/l 100 ug/l 8/25/09 2.1 50 ug/l 100 ug/l 9/1/09 1.8 50 ug/l 100 ug/l 9/08/09 1.9 50 ug/l 100 ug/l 9/15/09 2.2 50 ug/l 100 ug/l 9/22/09 1.8 50 ug/l 100 ug/l In their motion, BAC Defendants raise a number of specific objections to Bartlett's justifications for excluding the historic sampling data from his groundwater model. BAC Defendants' objections, which are supported by the expert declarations of Daniel B. Stephens and John L. Wilson, are summarized as follows:

1. Bartlett's model is hostile to sampling from: (1) "monitoring wells" around the BAC site; and (2) MWC-2 from 2007 onward. Bartlett and/or Dr. Laton admit that this data is valid and reliable. However, Bartlett's model predicts appreciable levels of hexavelant chromium in these areas.
2. Defendants contest Bartlett's predictions regarding the size of the groundwater plume. Defendants note the inconsistency between Bartlett's prediction — the BAC Site is "clean" by 2004 — and his modeling data, which predicts high levels of chromium in 2004 and beyond. They also point out that the plume was not observed in upgradient monitoring well sampling.
3. Bartlett represents that the sampling data is unreliable because hexavelant chromium is "complex" and may appear/disappear in a well. Defendants aver that Bartlett fails to identify a similar circumstance where a compound's variability invalidated a single year of monitoring. Defendants further argue that the same fact — variability — annuls Bartlett's model.
4. According to Bartlett, the infrequency of the sampling renders it unreliable. Defendants, however, respond that Bartlett's predictions conflict with every single sample taken at MWC-2 over a 46-year period. Bartlett's model is also inconsistent with the data from MWC-23, the closest monitoring well. There is simply no harmony between Bartlett's model and any sample taken from MWC-2 or any monitoring well.
5. Bartlett suggests that when the well was pumping at lower rates, the water may have flowed away from MWC-2. However, Defendants argue that pumping rates were not incorporated into his groundwater model and Bartlett testified at his deposition that "there was not enough information" to know whether low pumping rates impacted chromium levels.
6. Bartlett argues that the samples were unreliable based on "chain of custody issues," however, he does not know whether the wells were sampled properly and has no evidence to suggest they were falsified. BAC Defendants respond that the "data was reported by Meadowbrook to the [] Department of Health and have been produced from the State's files."

BAC Defendants rely on the opinions and modeling of Daniel B. Stephens and John L. Wilson to challenge Bartlett's testimony. Stephens and Wilson believe, based on their model and observed conditions, that chromium contamination from the BAC site did not reach MWC-2. Their model predictions are inconsistent with Bartlett's model, which Wilson characterizes as "invalid" and violating the basic premise of expert modeling: merging existing data with model assumptions. As to Bartlett's critiques of and modifications to the Stephens/Wilson model, Wilson responds:

Bartlett changed our model by accelerating the rate at which contamination reached groundwater and the manner in which pumping was simulated. Mr. Bartlett's revision of our model predicts hexavelant chromium concentrations in MWC-2 that are higher than the observed concentrations, and the modeled plume has shape that is quite different than the observed plume. Because the revised model conflicts with the observed data, it too is invalid [. . .]
If a groundwater model is to be used to predict the concentrations of contaminants in a well, then a key test of that model is whether it is consistent with data observed in the field. The Bartlett model fails that test.

(Doc. 677-7 at ¶¶ 6-7.)

Daniel B. Stephens' declaration, supporting BAC Defendants' motion for summary judgment, mirrors the positions taken by Mr. Wilson in his declaration. (Doc. 697 at ¶¶ 4-15.)

Plaintiffs dispute BAC Defendants' characterization of Bartlett's modeling and his decision to exclude the sampling at and around the BAC Site, including MWC-2's sampling data. In opposing BAC Defendants' motion and expert arguments, Bartlett declares that his model "incorporated the work of other Merck consultants and was based on sound, state-of-the-science analysis." He also reaffirms his original position that MWC-2 was contaminated with concentrations of chromium in excess of the MCL for a period of twenty-five years.

Bartlett's 23-page declaration, an extension of Plaintiffs' 36-page opposition brief, focuses on the decision to exclude the sampling data as an input into his groundwater model. Bartlett declares that the sampling was disregarded for the following reasons: (1) the samples "appear to be taken" during periods of low pumping; (2) lack of documentation concerning how the samples were collected; (3) the deeper portions of the well were not screened;

and (4) the pulsating nature of the chromium movement combined with the fact that, since 1993, regional groundwater flow is to the northwest. These criticisms are not made in Bartlett's original expert report, submitted to the BAC Defendants on December 9, 2009.

Bartlett's supplemental declaration repeats his criticisms of the Wilson/Stephens groundwater model. Specifically, Bartlett theorizes that chromium travels much more rapidly in the aquifer than either model anticipates and that the monitoring system did/does not adequately sample the deep portion of the aquifer:

The monitoring system is essentially a two dimensional (shallow aquifer only) and does not inform the investigator regarding the deep portion of the aquifer. How can one evaluate a large, three-dimensional aquifer system if the data only reflect[s] the shallowest few feet of the system? We are essentially being asked to evaluate a three-dimensional problem with two-dimensional data.
The predictions of our three-dimensional model should not be limited by the poorly planned monitor well system that was installed at the BAC site by the entity responsible for the contamination. Dr. Stephens (and other contracts that have worked at this site) have portrayed concentrations in groundwater by contouring the data from the BAC monitor wells. These portrayals are limited to the shallow aquifer and ignore the deeper aquifer because there is essentially no data to contour the deep aquifer.

(Doc. 777 at ¶ 8.)

This is puzzling in view of Bartlett's hearing testimony that the source of contamination is not from the deep aquifer below 140 feet, but rather is in the shallow aquifer.

According to Bartlett, the Wilson/Stephens model "does not simulate what has been observed at the BAC site." Bartlett specifically cites sampling from offsite well M A-1 in July 1992, which measured 1,300 ug/l. The Wilson/Stephens model predicted less than 6 ug/l. As to the most critical area, MWC-2, Bartlett argues that the model under-predicts chromium contamination and "[Wilson] should apply his criticism of the Bartlett Model to his own model."

According to the exhibits submitted by Plaintiff on December 28, 2010 (BATES # MER00015037), M A-1 was a "monitoring well installed by All America Trenching on June 9, 1992 [] the well was drilled to 55 bgs [and] screened from 35 to 55 bgs." M A-1 is the closest well to the BAC Site, located approximately 250 feet southwest from the facility. It is one of the furthest monitoring wells from MWC-2, approximately 1,100 feet north. In addition, according to the same exhibits, monitoring well M A-2 was installed by All America Trenching on December 22, 1992 "as part of a continuing environmental investigation of an underground storage tank leak at the off-site market."
It is undisputed that M A-1 and M A-2 had several high chromium readings immediately following installation. Although M A-1 and M A-2 were located furthest from the facility and "M A Market's environmental investigation [wa]s unrelated to the activities at the BAC-Pritchard facility," Mr. Bartlett relies heavily on these wells to establish that MWC-2 was contaminated with chromium above the MCLs during the relevant time-frame.

Bartlett's supplemental declaration purports to reinforce the scientific basis for his model and the exclusion of MWC-2's sampling data. According to Bartlett, the historical operation and construction of MWC-2 is not sufficiently documented to allow for scientifically-accurate groundwater (and flow) modeling. Bartlett also responds to Stephens' criticisms concerning "background concentrations of chromium in regional groundwater." Given the importance of chromium background concentrations to the instant inquiry, the relevant portion of Bartlett's supplemental declaration is reproduced in its entirety:

Dr. Stephens claims in his declaration that MWC-3 and MWC-5, two other Meadowbrook Water Company production wells `provide good examples of the background concentrations of chromium in regional groundwater.' He cites that MWC-3 had a detection of 14 ug/l in 2001 and MWC-5 (which replaced MWC-3 in 2002) had a measured hexavelant CR concentration of 3.6 in 2002. These wells were immediately adjacent to each other and are located about 1 mile due west of the BAC site. Contrary to Dr. Stephens' opinion, we believe that these data suggest that MWC-3 was impacted by chromium from the BAC site and that this detection substantiates the results of modeling that I completed as part of this case.
The [RWQCB] recognizes that the background concentration of total chromium in groundwater at the BAC site is 6 ug/l. This value is based on an analysis that IT Corporation performed for Merck [. . .] Clearly the measurement of 14 ug/l in MWC-3 exceeds the background by a factor of more than 2. Dr. Stephens, in his Cottle declaration, dated March 2009, believed that the background total chromium concentration ranged from 4 to 6 ug/l. However, in his deposition, Dr. Stephens changed his opinion and claimed that background concentrations of total chromium were greater than 12 ug/l [. . .]
We conclude that the measurement of chromium in MWC-3 is an indication of just how far chromium from the BAC site has traveled in this aquifer system. The value measured in MWC-3 suggests the Bartlett model underestimates the dispersal of chromium to the west of the BAC site.

(Doc. 777 at ¶¶ 11-12.)

The BAC Defendants, in their reply, assert that there is an unexplained analytical gap between Bartlett's model and the actual data. Defendants argue that there is no evidence of low-pumping at MWC-2, rather there were several months where the well pumped more than 20 million gallons (above average); that "pulse" movements cannot explain nearly 46 years of near non-detect results; Bartlett's theory that the monitoring wells are too shallow to detect the hexavalent chromium plume is wrong because he predicts high concentrations at depth and in the shallow zone; testing of the "deep" wells near the BAC site — DW-1 and DW-2 — were non-detect when sampled in 1991 and 1992. Bartlett's model, however, predicts chromium concentrations in excess of 5,000 ug/l at those wells, contrary to the actual sampling. At the hearing, Bartlett rationalized the differences by opining that data from the monitoring wells failed to detect chromium in the shallow aquifer, presumably at 60 to 150 ft bgs.

Defendants' final reply argument is that even assuming, arguendo, that there is a valid scientific objection to the pre-2007 sampling data, Plaintiffs fail to reconcile Bartlett's groundwater model with: (1) post-2007 MWC-2 data; and (2) the data from the BAC Defendants' monitoring well network. Bartlett's model is contrary to almost all observed data, California Department of Public Health statutes and test results, and DTSC findings after remediation efforts, all of which concluded hexavelant chromium was not present at levels in excess of the MCL.

Based on the magnitude of the dispute between the parties' experts and the complexity of the groundwater issues in this case, the Court retained, under Fed.R.Evid. 706, an independent expert, geohydrologist Kenneth D. Schmidt, Ph.D., to assist in the understanding and analyzing the dispute in the groundwater pathway issues. Dr. Schmidt's expert report, submitted on November 22, 2010, disagrees with Mr. Bartlett's model. The substantive divergence being Schmidt's opinion that a substantial relatively impermeable clay layer — the "Corcoran clay" — exists beneath MWC-2 (and other wells near the BAC site) at approximately 92 feet. Based on his review of the original well driller's log, the electronic logs ("e-logs") and his extensive professional experience working in the Merced area and Central Valley geohydrology, Dr. Schmidt believes that MWC-2 is an open-bottomed well, with a "blank casing" landed in the clay layer (at approximately 92 feet). The impact and effect of these conditions is that MWC-2 primarily drew uncontaminated water the deep aquifer below the Corcoran clay layer, not the upper aquifer, i.e., there is minimal hydraulic connectivity between the upper and lower aquifers. Because Bartlett believes that the upper aquifer is contaminated, Dr. Schmidt's conflicting opinion puts in material the "source and extent of contamination" in MWC-2.

In the appropriate case, typically if the evidence consists of complex scientific evidence, the Ninth Circuit has found that Rule 706 allows the court to appoint a neutral expert. See Students of Cal. School for the Blind v. Honig, 736 F.2d 538, 549 (9th Cir. 1984), reversed on other grounds by 471 U.S. 148 (1985) ("[Rule] 706 allows the district court to appoint a neutral expert on its own motion, whether or not the expert is agreed upon by the parties.").

The Arcadis report also notes a clay layer. Bartlett's December 2, 2010 supplemental declaration provides, in relevant part:

Based upon my review of the reports and opinions of Dr. Richard Laton and Marvin Glotfelty, as well as numerous historical documents related to this site, I do not believe that the Corcoran clay or other confining clay layer is present at MWC-2. It is my opinion that the lithology is completely interconnected with the uppermost units and therefore all groundwater is interconnected.

(Doc. 958, ¶ 5) (emphasis added).
Bartlett modified his testimony at the hearing to acknowledge that there is clay present, but it is not a continuous impermeable layer.

According to Dr. Schmidt, in "order to successfully complete an open-bottomed well, the casing must be landed in clay." (Doc. 941 at pg. 14.) The clay layer acts a "seal" and only water from below the clay layer can be "tapped." This is directly disputed by Bartlett who opines water enters MWC-2 from the shallow aquifer through the unsealed annulus and slots in the casing.

Bartlett vehemently disagrees. In his testimony on December 15 and 28, 2010, he testified that MWC-2 primarily drew water from the upper aquifer above the non-continuous clay layer. Bartlett contends that there is no evidence of an "annular seal" in the 2009 video log or, separately, the gamma ray log. Because there is no annular seal and slots in the well's casing, water from the upper aquifer flowed into the well, i.e., there was nothing to prevent the downward flow of contaminated water. It is Bartlett's opinion that the water pumped at MWC-2 contained contaminated water from the upper aquifer. This opinion is contrary to Dr. Schmidt's independent expert report.

According to Bartlett, the video log was taken in September 2009. (RT, Dec. 15, 2010, 28:1.) It is undisputed that MWC-2 was out of service in 2009, i.e., the well was in a static "no-pumping" condition.

Dr. Schmidt offers a number of additional contrary opinions to support his conclusion that Bartlett's model is unreliable and inaccurate:

1. To form his opinion on the annular seal, Bartlett relies too heavily on Plaintiff's other experts, Mr. Glotfelty and Mr. Laton, and completely ignores the language of the original driller's log. Bartlett also misunderstands or is unfamiliar with: (1) open-bottomed wells; (2) video logs; and (3) Central Valley hydrogeology.
2. Bartlett does not adequately address the "water budget," instead assuming it is constant or one (inflows = outflows).
3. Bartlett's hydraulic connectivity values (between the contaminated plume and MWC-2) are too high. Bartlett assumes uniform hydraulic connectivity, which overstates actual movement (clay layer, low permeability). He used specific capacity data which is ten times more than the hydraulic connectivity values derived from the slug tests.
4. Bartlett's revisions to the earlier IT and Arcadis models were not comprehensive enough and did not take into account their limited purpose/hydrogeology.
5. Insufficient flow model calibration. Bartlett only considered the historical water level changes, not the direction of groundwater flow in the different layers (including the strata below the Corcoran Clay). Bartlett offered new opinions about direction of groundwater flow.
6. Insufficient transport model calibration. Bartlett's model is did not match the representative results of actual monitoring for a number of monitor wells or the MWC-2 in the southerly area. In addition, there were two open-bottomed wells at the BAC site, which tapped groundwater below the Corcoran clay or its equivalent. Sampling results for these wells indicated very low concentrations of total chromium. Bartlett's model conflicts with this observed data.
7. Schmidt believes that there was a gasoline leak near the M A Market during the relevant time period. According to Schmidt, the gas leak induced a chemical reaction, converting a portion of the hexavalent chromium to trivalent chromium. This was not addressed by either party.

A slug test is an aquifer test where water is quickly added or removed from a groundwater well, and the change in hydraulic head is monitored through time, to determine hydraulic connectivity and field permeability. See generally Champion Laboratories, Inc. v. Metex Corp., No. 02-5284-WHW, 2009 WL 2496888, at 11 (D. N.J. Aug. 13, 2009).

(Doc. 941 at pgs. 4-25.)

Dr. Schmidt's ultimate opinion is that "the Bartlett model cannot be used to determine the historical chromium concentrations in water pumped from MWC-2." (Id. at pg. 25.) According to Dr. Schmidt, his opinion was based on a number of factors, but "primarily [] due to [] the use of values for hydraulic conductivity that are too high [and] the assumption of too much water from above the Corcoran Clay being pumped from MWC-2." (Id.)

Dr. Schmidt's opinions were the focus of argument and testimony on December 3, 15, and 28, 2010, at which time Bartlett presented rebuttal testimony. Bartlett opined that Dr. Schmidt's opinions were incorrect and reiterated that there was no evidence of an annulus seal, i.e., there was groundwater movement between the upper and lower aquifers at MWC-2 (MWC-2 primarily drew water from the contaminated upper aquifer). During his testimony, Bartlett played portions of the 2009 video log, pausing the video to provide commentary at critical points. Bartlett noted that at 92.8 feet bgs he observed the actual bedrock (alluvial formation) outside of the steel casing, evidence of an absence of a casing. Bartlett also observed a number of corrosion holes in the steel casing, which allegedly allowed the water to move freely from the upper aquifer.

Bartlett agreed with Dr. Schmidt's opinion that MWC-2 is open-bottomed. (RT, Dec. 15, 2010, 25:6-23:16.) Bartlett, however, opined that water can also be drawn from slots in the liner and unsealed annulus, i.e., the draw is not limited to the water from the bottom of the well hole.

With respect to the presence of a clay layer, Bartlett opined that no layer exists because, at 98.4 feet, "I can look through the casing and see if there is no seal. There is a bore hole wall. And quite a large gap, probably several inches." (RT, Dec. 15, 2010, 23:3-23:5.) Bartlett also argued that the gamma ray log supported the lack of a continuous and thick clay layer beneath MWC-2. Whether a "substantial" continuous clay layer exists beneath at MWC-2 is heavily disputed between Dr. Schmidt and Mr. Bartlett.

With these parameters of the dispute, the parties' arguments are addressed in the context of Federal Rule of Evidence 702 and the Supreme Court's decisions in Daubert, 509 U.S. 579, and Kumho Tire Co., 526 U.S. 137.

1. Bartlett Groundwater Model — In or Out?

The seminal dispute between the parties is whether Bartlett properly excluded the MWC-2 and monitoring well data from his groundwater model. Plaintiffs state that the BAC defendants' Daubert motion is fundamentally improper because they are challenging the "results" of the model, not the methodology. According to Plaintiff, such a challenge falls outside Daubert's ambit and must be determined by the trier of fact. Plaintiffs further argue that the sampling data and any model excluding that data are not necessarily inconsistent. Rather, "defendants have cited to no authority that somehow such data trumps expert opinion on these issues or that such data is required in addition to expert opinion and analysis in order for plaintiffs to meet their burden at trial":

[W]hile defendants claim the alleged testing results from MWC-2 should be the `reality' by which the Bartlett model must be judged — it is readily apparent that this issue as to what constitutes `reality' is at the heart of this litigation. Such an argument further presupposes the superiority of the alleged testing data from MWC-2 over the conclusions of the Bartlett model. The Bartlett model and Bartlett's opinions however challenge the accuracy and representativeness of the alleged data from MWC-2.

(Doc. 776 at 2:16-2:22.)

Defendants rejoin that they are not challenging Bartlett's conclusions per se, rather they dispute the inconsistencies between Bartlett's model and the actual sampling data. According to Defendants, Bartlett's model is inherently unreliable because: (1) his model contradicts and ignores all of available field testing data; (2) Bartlett developed his model without reviewing the data and then, when it was brought to his attention, simply ignored the data; (3) Bartlett's model departs from well-recognized groundwater modeling protocol because it is not calibrated to actual well monitoring data; and (4) Bartlett failed to perform a sensitivity analysis on his model, another breach of well-recognized modeling protocol. These reasons allegedly disqualify Bartlett's model under Rule 702 and Daubert.

Citing Ramsey v. Consolidated Rail Corp., 111 F. Supp. 2d 1030, 1037 (N.D. Ind. 2000), and In re TMI Litigation, 193 F.3d 613, 671 (3d Cir. 1999), amended on other grounds, 199 F.3d 158 (3d Cir.), cert. denied, 530 U.S. 1225 (2000), BAC Defendants argue that "[m]odel-based opinions are of no value to a trier of fact where they are contrary to real world data." In Ramsey, the court framed the relevant legal inquiry as follows: "The issue before the court isn't whether Dr. Haitjema's [Plaintiff's groundwater expert] analysis is more probably correct than incorrect — the Ramseys are entitled to have that issue decided by a jury rather than a judge — but rather whether the opinion testimony that encompasses his analysis is admissible into evidence."

Ramsey's facts and reasoning provide a starting point. In Ramsey, Plaintiffs attempted to rely on groundwater modeling to prove groundwater contamination. Specifically, Plaintiff attempted to demonstrate that TCE contamination from a rail yard migrated north to Plaintiffs' off-site well. Plaintiffs' well was tested twice during the relevant time-frame, and no contamination was found. Additionally, two upgradient wells (between the rail yard and well) were similarly negative for contamination. The Ramsey court stated that the upgradient wells had been sampled twelve times, and no contamination had been found in any of those wells.

Determining whether proffered testimony is scientifically reliable can be a considerable challenge for a trial judge. See Bitler v. A.O. Smith Corp., 391 F.3d 1114, 1121 n. 1 (10th Cir. 2004). "This requirement will sometimes ask judges to make subtle and sophisticated determinations about scientific methodology and its relation to the conclusions an expert witness seeks to offer — particularly when a case arises in an area where the science itself is tentative or uncertain [. . . .]" Joiner, 522 U.S. at 147-48 (Breyer, J., concurring).

The Ramsey Plaintiffs submitted the testimony of Dr. Haitjema, who opined that groundwater modeling demonstrated that contamination reached the Plaintiffs' well during the years 1983 to 1998. Dr. Haitjema's model directly conflicted with sampling results, and Defendants moved to strike the testimony under Daubert. In response to Defendants' motion, Dr. Haitjema argued that his model was valid because: (1) groundwater contamination provides a "snapshot of contaminant occurrences" and is inherently unreliable because contaminants "move along different pathways at different rates and at different times."; (2) his model incorporated an EPA consultant's model, who was retained at an earlier time to investigate possible contamination near the rail yard; (3) the allegedly contaminated well tapped from the same depth as a contaminated well to the south (towards a rail yard); and (4) contamination from the rail yard would vary spatially and temporally.

Ramsey decided that none of Dr. Haitjema's arguments established the reliability of his model or the opinion that appreciable rates of TCE reached the Plaintiffs' well. The Court held that although Dr. Haitjema's methodology passes the Daubert inquiry with "flying colors," the "record identifies no scientific principle on which Dr. Haitjema based the critical last step of his analysis — that the uniform `non-detect' results of well testing, not only of the Ramseys' well but also wells to the south, don't affect the conclusion that TCE [] from the rail yard entered the Ramsey's drinking water." The Court further explained the basis for excluding Dr. Haitjema's model under Daubert:

[U]se of the groundwater flow model as a comparatively accurate predictor of the general direction of VOC migration doesn't support a finding of reliability when the model is used to support an opinion that VOCs traveled from one point (anywhere on the rail yard) to a specific second point (the Ramseys' well) despite lack of support in years of actual testing [. . .]
the critical link to the Ramseys' causal chain isn't simply that Dr. Haitjema is right (and the others wrong) about whether TCE or CC14 from the rail yard reached the Ramsey property — the point to which the other references in Dr. Haitjema's supplemental report seem pertinent — but rather whether those substances reached the well from which Amanda Ramsey's drinking water was taken. That water was tested, as was water from the wells to the south of her well — twelve tests in all, over the course of eight years — and none of those tests detected such substances.
Id. at 1037.

The Ramsey court concluded:

The court struggles to avoid the superficially logical conclusion that those substances would have been detected in at least one of those twelve tests had they reached the Ramsey well: the law of evidence allows scientists to give expert testimony in court precisely because what science knows to be true is not always intuitive, and may even appear illogical to the non-scientist. Still, it seems that reliable methodology would, as suggested by Mr. Weaver's report, involve further examination after some number of tests revealed no VOCs. The requisite number of negative tests might be greater than one or two, or perhaps even greater than twelve, but this record provides no basis for the court to find that methodology based on reliable principles simply allows all twelve tests to be discounted in their entirety.
Many cases decided under Daubert have excluded opinion testimony from experts who ignored facts or considerations that must be considered under methods based on reliable principles. See, e.g., United States v. Crotteau, 218 F.3d 826, 833 (7th Cir. 2000); Virginia Vermiculite Ltd. v. W.R. Grace Co.-Conn., 98 F.Supp.2d 729 (W.D. Va. 2000). Dr. Haitjema didn't ignore the non-detect tests. He explained why those tests didn't prove that TCE never reached the Ramsey well. But the record contains no explanation as to how any scientific principles support the contrary opinion in the face of eight years of non-detect results in and immediately around the Ramsey well.
Id. at 1037-38.

Ramsey has facial similarity to this case: the expert argued that his "methodology is well-accepted in the [groundwater] field" and that "other modeling studies support his conclusions about the general directions of groundwater movement [around the site]." Id. at 1034. Similarly, Bartlett argues that the MODFLOW methodology is well-accepted in the field, which it is, and that the Arcadis model supports the general directional movement, which it generally does. As demonstrated by the Ramsey opinion, however, allegiance to portions of accepted scientific protocols, but not others, is not dispositive of the Daubert inquiry. The Third Circuit explains: "[u]nder Daubert, any step that renders the expert's analysis unreliable [. . .] renders the testimony inadmissible [. . .] [t]his is true whether the step completely changes a reliable methodology or merely misapplies that methodology." Mitchell v. Gencorp Inc., 165 F.3d 778, 782 (10th Cir. 1999). Bartlett's model is on the razor's edge of that standard. He has now offered explanations why the testing results are not reliable: (1) the monitoring wells are too shallow and do not intercept the

Bartlett's base methodology satisfies Daubert's reliability and relevancy inquiries. Bartlett employed the industry-standard MODFLOW computer program to simulate groundwater flow and to calculate the rate at which hazardous substances migrated from the BAC Site. MODFLOW was formulated by the United States Geological Survey, is sanctioned by the EPA, has been peer-reviewed, and is considered a standard model that is acceptable and commonly used by hydrogeologists. The use of MODFLOW by groundwater experts has been approved by a number of district courts throughout the United States and its base use in this case is not disputed. See, e.g., Aero-Motive Co. v. Becker, No. 1:99-CV-384, 2001 WL 1698998, at 2-3 (W.D. Mich. Dec. 6, 2001) (describing MODFLOW as a "widely-used, well-tested particle-tracking model[] that ha[s] been subjected to peer review and [is] commonly accepted in the hydrogeologic community.").
According to Defendants, Bartlett's use of MODFLOW is not the problem. Rather, Bartlett's groundwater flow model is unreliable and inaccurate because he fails to employ "standard modeling protocol."

contaminated aquifer; (2) groundwater reliability is not accounted for; (3) test methods and test results are suspect and not objectively confirmable; and (4) there is infrequent testing of MWC-2.

The comparisons between this case and Ramsey do not end there. The Ramsey expert justified his model's predictions, accepting the non-detect field data, by opining that the "degree of groundwater contamination downgradient from the rail yard [] would vary spatially and temporally," and "in nearly every case in which he ha[d] been involved, `discrepancies' have been found between expected pathways of contaminants and data from wells." Id. Plaintiffs in this case similarly state: hexavalent chromium is so chemically unbalanced and unstable (and subject to pulses) that it cannot be tested; and the data is unreliable due to pumping irregularities and variability; and inadequate documentation by Meadowbrook. Assuming, arguendo, that those criticisms have merit, there is a scientific, practical, and logical expectation that at least one of the data points from MWC-2 — from the years 1964 to 2009 — would have been positive for hexavelant chromium, if Bartlett's model was reliable and accurate. Borrowing from Ramsey's language: "[the Court] struggles to avoid the superficially logical conclusion" that hexavelant chromium would have been detected at MWC-2, at actionable levels, at least one time during the relevant time period.

During oral argument on December 29, 2010, Plaintiffs' counsel sought to distinguish Ramsey by stating that Ramsey only dealt with "true non-detects." In this case, there are many (many) MWC-2 and test well non-detects significantly lower than the MCL (3-10 ug/l). There is, however: (1) data from the early 1990's that other, more distant wells — extraction and monitoring — were more contaminated; and (2) a 2008 sample from MW-23 shows a high chromium hit of 450 ppb. Plaintiffs reason that Ramsey is distinguishable based on the "`true' non-detect" nature of the sampling data; a single reading above non-detect level deprives the case of any analogous impact.

The analytical gap between Bartlett's predictions and the actual data is more pronounced than in Ramsey. Here, Bartlett's model conflicts with more than 100 samples from MWC-2 and nearby monitoring wells (i.e., MW-23). In contrast, Dr. Haitjema's model conflicted with twelve non-detect samples in Ramsey. Bartlett's model predicts chromium concentrations of chromium at MWC-2 as high as 582 ug/l, 230 times more than ever recorded at that well.

To understand this argument it is necessary to profile the observed data from MWC-2 and the monitoring/extraction wells and, more importantly, describe how it was analyzed/used during the different stages of Plaintiffs' case, i.e., first in the expert reports, second, in summary judgment declarations, and last, in the post-trial hearings (Daubert and rebuttal to 706 experts). A close review of these documents and testimony reveals several inaccuracies and/or unexplained assertions in the Bartlett model (and his testimony). Such a determination, however, is inherently based on credibility, which is the function of the trier of fact.

Bartlett testifies in December 2010 that he did consider all of the test data. Defendants argue that Bartlett's minimization of this data necessarily raises the question whether the Bartlett model was "calibrated" and tested according to well-established modeling protocol. The consideration of the MWC-2 and monitoring/extraction well data is considered in the context of the calibration arguments.

According to well-recognized groundwater modeling principles and the undisputed expert testimony in this case, an expert must test or "calibrate" a model to align the model predictions with field observations. See, e.g., Mary P. Anderson and William W. Woessner, "Applied Groundwater Modeling: Simulation of Flow and Advective Transport." According to this treatise, among others, the larger the disconnect, the greater need to validate or "calibrate" the groundwater model. As explained by one federal district court, "[c]alibration is the adjustment of parameters of a model's geometry or input parameter values in an effort to match model outputs to observed conditions." Chitayat v. Vanderbilt Assoc.'s, No. 03-5314-DRH-MLO, 2007 WL 2890248, at 3 (E.D.N.Y. Sept. 27, 2007).

The degree of variance between Bartlett's model and the well data requires a factual foundation to decide the truth of Bartlett's grounds for minimizing most of the observed test data. This credibility determination will bear on the ultimate admissibility of the model. The Court reserves the right to exclude the model after hearing the evidentiary foundation for the model at trial.

Defense counsel's direct examination of Dr. Stephens on October 6, 2010 provides a starting point to analyze whether Bartlett's model employs flawed methodology, specifically, whether Bartlett improperly excluded/ignored relevant field data so that his model "predicted" higher concentrations of hexavelant chromium at MWC-2. Stephens believes that Bartlett omitted, without scientific justification, several fundamental modeling steps that are built into the modeling protocol to ensure scientific reliability:

Dr. Stephens asserts the omitted steps are "calibration, sensitivity analysis, and validation."

Q. Are there what we might call essential elements of a model that those of you in the field and in Exhibit B would look to to conduct a reliable and valid model? A. Yes, sir. Q. What are those elements? A. In this modeling protocol, one of the first steps is to establish clearly the purpose of the model. The second step would be to develop a conceptual model. Q. What does that mean? A. The conceptual model is where you get to look at all the field data, you assemble all the field data that's available regarding the geology, regarding the climatological conditions, regarding pumping rates, the locations of rivers and streams, the construction of wells, et cetera. You get all that data together and you form in your mind a framework as a hydrogeologist and your understanding of the subsurface and what's moving into it, what's moving out through the pores and into wells or streams, et cetera. Q. Okay. What's the next step? A. The next step is to select a computer code, a mathematical model that you're going to use to capture that concept about the behavior of the hydrogeologic system. Q. Okay. Next. A. And the next step would be to design the model. Meaning in the case of a numerical model, how many elements or cells the model will have, how many layers, how large an area, how deep the model will go, et cetera. Q. Okay. Next. A. And the next step is to calibrate the model. Q. What is calibration? Uh, please, go ahead. A. Calibration is probably one of the most important parts, one of the most important steps in the process. In the protocol among all the steps. Calibration is a crucial one. And this is one in which the model is run based on your initial understanding of the hydrogeology, the conceptual model. You run the numerical model and then you make a calculation of, let's say, the contaminant plume. And you compare the contaminant plume to the observed data. Q. When you say the observed data, are you talking about measured data? A. That's correct. Yes. Q. Like monitoring well analytical data? A. Yes. The plumes that one might draw from those point-wise measurements in the monitor wells. Q. So when you — when you construct a model and go to this calibration step and you look at a plume, you overlay that plume that you've modeled with existing data. What if it doesn't match? A. Then if it doesn't match, the next step would be to go back and revisit your conceptual model. Something wasn't quite right with your assumption. Something didn't quite behave as you expected it to. The geology wasn't right. They had the layers too thick, they were too permeable, not permeable enough. The well was pumping too much, it was pumping too little, you started the contaminant source too soon, too late, depending on whether the plume was larger or smaller than what was measured or observed. So you go back and revisit all the assumptions and revise your conceptual model and then make another forecast. And this is an iterative process, typically until you get a good agreement, a reasonable agreement for whatever your goal is. It's got to be consistent with what your goal is. So you get a reasonable agreement to go forward with a prediction. Q. Okay. And what's the next step after you've calibrated the model and evaluated it, the output of that model, against the measured data? What's the next step? A. The next step typically, it's kind of coupled with the calibration stage. That's typically the sensitivity analysis, where you try to understand what is the most important — or what is the most important factors that are affecting my outcome. That's the — some adjustments you might make to, let's say, pumping rate or permeability to see how that might affect the size or shape of the plume. Just to get your understanding of what is the most important information and what type of uncertainty is associated with your understanding about that information. So you get a good estimate of — a good expectation or at least intuitive feel for the reliability of your results. Q. Okay. And after the sensitivity analysis or analyses are completed, what's the next step? A. Another step typically is called verification or testing, in which you run the model that you feel is calibrated to the best of your ability. Then you might make a prediction to compare the model against a different data set that you have available from what you used to calibrate the model. Q. Okay. And is that the last step? A. Well, another step would be getting toward the prediction into the future, let's say, if you're trying to make a forecast of the future condition, that you make some predictions then with more confidence. There then could be other steps, such as validation. That would be, let's say we made a prediction with a model two years ago and now today we have a result that we can use to compare against what the model predicted when we ran it two years ago and see if, in fact, that happened. That will be a validation step. Q. Did you have the opportunity to read the deposition testimony of Mr. Bartlett? A. Yes, sir. Q. Did Mr. Bartlett testify whether he did or did not calibrate his transport model? A. He did not. Q. He — just so the record is absolutely clear. He testified on the subject and testified that he did not calibrate his transport model? A. That's correct. He did not calibrate his transport model, my understanding is what he said in his deposition. Q. Did — do you recall whether he testified whether he did a sensitivity analysis? A. I don't believe he could have done a sensitivity analysis without calibrated — without going through the calibration step. Q. If he testified that he did not do a sensitivity analysis, would that be consistent with your understanding? A. Yes. (RT, Oct. 6, 2010, 124:20-129:9.)

Defendants argue that Bartlett, armed with this knowledge and without scientific justification, elected not to "test" his model against field data revealing an absence of hexavelant chromium at or near MWC-2. The result, according to Defendants, is a model and opinion that is inherently unreliable and only connected to existing data "by the ipse dixit of the expert." See Joiner, 522 U.S. at 146.

It is undisputed that Bartlett did not conduct a sensitivity analysis. A sensitivity analysis is performed on a calibrated model to "establish the effect of uncertainty." Bartlett claims that a sensitivity analysis is not required because his model is based on the Arcadis model, which performed a sensitivity analysis.

Plaintiffs suggest that Bartlett's model conforms to well-recognized groundwater methodology because he calibrated the "flow" portion of his model and, critically, used his "professional judgment" to exclude chromium concentration data from MWC-2 and the monitoring wells, i.e., the "transport" component.

Bartlett's model consisted of two components: "flow" and "transport." The "flow" portion models groundwater flow in saturated and unsaturated zones around the target site, including confined, unconfined and perched aquifers. The "transport" portion models the supply or source of the chemical to groundwater, focusing on mass per unit of time (entering the groundwater). The flow and transport components work synergistically to assist the modeler in estimating whether a contaminant entered the groundwater table and, if so, whether the contamination migrated from the source to off-site locations.

Plaintiffs' argument has some support in the record. They rely on the Anderson and Woessner treatise which annotates the calibration, calibration sensitivity analysis, and model verification steps as follows:

5. Calibration. The purpose of calibration is to establish that the model can reproduce field-measured heads and flows. During calibration a set of values for aquifer parameters and stresses is found that approximates field-measured heads and flows. Calibration is done by trial-and-error adjustment of parameters or by using an automated parameter estimation code.
6. Calibration Sensitivity Analysis. The calibrated model is influenced by uncertainty owing to the inability to define the exact spatial (and temporal) distribution of parameter values in the problem domain. There is also uncertainty over definition of boundary conditions and stresses. A sensitivity analysis is performed in order to establish the effect of uncertainty on the calibrated model.
7. Model Verification. The purpose of model verification is to establish greater confidence in the model by using the set of calibrated parameter values and stresses to reproduce a second set of field data.

While Bartlett specifically delineated "field-measured heads and flows," there is no mention of calibrating the "transport" portion of the model to observed concentrations of contaminants from specific wells or monitoring well networks. However, according to the California EPA's groundwater investigation manual, titled "Ground Water Modeling for Hydrogeologic Characterization," both the flow and transport components of the model must be calibrated to field data:

It is undisputed that the "field-measured heads and flows" concern water level data and the "flow" model, not the transport model or the chromium concentration levels.

The California EPA manual was first cited by the Court, not the parties. It was thereafter referred to by the parties during post-trial hearings concerning the reliability of the Bartlett groundwater model.

Calibration is the iterative process of adjusting the parameters in the model, such as hydraulic connectivity, transmissivity and dispersivity, so the model adequately approximates the real ground water system. This is accomplished by comparing the model results to a set of field observations. The calibration data set should include measurements over the lateral and vertical extent of the model area. For a flow model this data will often consist of water level measurements from monitoring wells and piezometers. Calibrations to observed hydraulic head gradients, rather than head measurements is more difficult, but may be more representative for problems dealing with flow velocity and transport. Contaminant concentrations measured in ground water samples will be used to calibrate a contamination transport model. The calibration data set, including all data point locations (monitoring wells, etc.) and the values of potential head or contaminant concentration that are being used for calibration, should be specified in the professional report.
California EPA, Ground Water Modeling for Hydrogeologic Characterization (Oct. 15, 2010, 9:35 p.m.),

Although these two authoritative sources do not provide a definitive answer on the issue of "calibration," i.e., whether well-established groundwater modeling protocol requires an identical calibration of the "transport" and "flow" models, it is undisputable that calibration is a "critical" and "valuable" step that ensures that model simulation matches the field observation to a reasonable degree.

The importance of calibrating model results to actual data is not limited to the field of groundwater modeling. In a number of different contexts, Federal Appellate Courts throughout the United States have emphasized calibrating/harmonizing model predictions with actual data to ensure reliability. See Eleven Line, Inc. v. North Texas State Soccer Ass'n, Inc., 213 F.3d 198, 206-8 (5th Cir. 2000) (In the antitrust context, stating that a plaintiff may prove damages [. . .] so long as the estimates and assumptions used rest on adequate data [however] the evidence of actual injury [in this case] impermissibly consists of estimates based on assumptions that are based on estimates and assumptions."); Inland Empire Public Lands Council v. Schultz, 992 F.2d 977, 982 (9th Cir. 1993) (in the NEPA context, the Forest Service's findings were not arbitrary or capricious because "[t]he agency constructed mathematical models to predict increases in runoff and stream flows as a function of acres logged and rate of regeneration following logging [. . .] [i]t conducted extensive field investigations to calibrate and verify its models.") (emphasis added); Sterling v. Velsicol Chemical Corp., 855 F.2d 1188, 1199 (6th Cir. 1988) ("The plaintiffs carefully devised, calibrated, and tested their model, based upon physical data generated by Velsicol's own consultants, to determine the physical and chemical characteristics beneath the landfill."); Ohio v. United States Environmental Protection Agency, 784 F.2d 224, 226 (6th Cir. 1986), reaff'd, 798 F. 2d 880, 881 (6th Cir. 1986) (EPA acted arbitrarily in using a model to set emission limits "without adequately validating, monitoring, or testing its reliability or its trustworthiness in forecasting pollution [. . .]"); Boucher v. U.S. Suzuki Motor Corp., 73 F.3d 18, (2nd Cir. 1986) (excluding expert testimony under Rule 702 because the testimony was "not accompanied by a sufficient factual foundation" and the "expert's projection thus was based on assumptions about Boucher's employment prospects that represent a complete break with his work history."). In each of these cases, the Court has recognized the impact of calibration on the model integrity.

Defendants focus on Bartlett's evolving rationalization for his approach to "testing" the transport portion of the groundwater model, originally excluding all of the data from MWC-2 data and monitoring well network. Defendant are correct that Bartlett's current position is not completely consistent with his expert report or his deposition testimony. Although Bartlett's testimony is contradicted in some respects, the credibility of a witness is an issue for the jury. See, e.g., Bourns, Inc. v. Raychem Corp., 331 F.3d 704, 714 (9th Cir. 2003).

During his April 7, 2010 deposition, Mr. Bartlett stated: "As I said, the transport part of the modeling is not calibrated." (Bartlett Dep., April 7, 2010, 115:20-115:21) (emphasis added). Bartlett further testified: "In other words, we don't directly compare concentrations in the model, predicted concentrations to the model observations. We are using, because we have to, a professional judgment." (Id. at 115:21-115:25.)

For instance, in his expert report, submitted on December 9, 2009, Bartlett devotes more than twenty-four pages to his revisions of the Arcadis model. The "model calibration" section spans nineteen lines and is limited to a discussion of the "flow" component:

The revised Arcadis model was constructed to simulate measured water level data from the monitor well network. As discussed, the general head boundaries were set to reflect the regional and local water levels. The purpose of this modeling exercise was to revise the existing Arcadis groundwater model to properly simulate the influence of well MWC-w. The comparison of model simulated and observed water levels is illustrated on Figures 5 through 10 and in Appendix B.
Figure 5 illustrates a comparison of model simulated and measured water level data for the fourth simulation period (1996 through 2008). Table 8 displays general model calibration statistics for the 1996 to 2008 simulation.
Figure 6 illustrates the locations of wells for which hydrographs were prepared. These hydrographs (figure 7) compare measured water levels with model simulated values for nine wells across the model domain. In general, the response of the water levels simulated by the model is similar to the measured values during the course of seasonal changes in pumping at well MWC-2. These hydrographs suggest that the assumption that well MWC-2 draws groundwater from all seven model layers is reasonable. The overall impacts of pumping at MWC-2 are about 2 to 4 feet of change in water levels over the course of a year. This is clearly reflected in the simulated water levels.
Figures 8 through 10 display the simulated water levels and residual error (measured value minus simulated value). In general residual values are low, with most less than 2 feet. These results suggest the model is well calibrated, with overall errors of less than 2 feet.

(Doc. 777-2, Bartlett Export Report, pgs. 25-26.)

Missing from the this discussion, headed "Model Calibration," is any mention of calibrating the "transport" portion of the groundwater model or any reason not to calibrate it. In addition, under the "Results" section, Bartlett states:

When comparing the predictions of the transport model to observed concentrations of chromium in the monitoring and extraction well network, any variance can be explained as a result of the highly mobile characteristic of hexavalent chromium when moving through the groundwater and the heterogeneity of the aquifer itself. Taking all of this into consideration, it is my opinion that the prediction[s] of this model are within the realm of scientific certainty and demonstrate the amounts of hexavelant chromium over time present in Meadowbrook Well No. 2

Section 10 of Bartlett's expert report — "References Consulted" — catalogues the documents/data reviewed by Bartlett and his staff. On December 28, 2010, Mr. Bartlett acknowledged that this section did not reference a number of reports and data from environmental/groundwater consultants retained by the BAC Defendants, including IT Corp., Dames Moore and Kleinfelder. Bartlett asserted that they were incorporated by reference to another expert's report, in particular, Dr. Laton's report. (Rough RT, Dec. 28, 2010, 28:5-31:6.) As to the omission of the Kleinfelder data taken from monitoring wells installed in 1983 and sampled in 1986 — showing non-detects at the BAC site — Bartlett testified that "[w]hat I discussed in my report related to the modeling that was done, the construction of the model, the Arcadis — I focused on the Arcadis work because that's where we were developing the model from." (Id. at 50:9-50:12.) It is undisputed that the Kleinfelder data is inconsistent with Bartlett's prediction of chromium concentrations at 5,000 to 40,000 ppb at the BAC site in 1986. In response to this large discrepancy and why he excluded the 1986 data, Bartlett stated: "I took the [Kleinfelder] data into account as I evaluated the results of the model and compared not only that data, but all of the data. I, as a scientist, am obligated to look at all the data and decide, in terms of the — both the space and time element fo that piece of data, as to whether it is representative [. . .] [a]nd looking at all of the data going forward in time as more information was collected, it was my opinion that the earlier data from these early shallow wells [the Kleinfelder wells] was not representative." (Id. at 50:18-50:25.) Bartlett further clarified on December 28, 2010 that he "made no use of" the Kleinfelder data. There is no mention of the Kleinfelder data in Bartlett's expert report or any reason to exclude it. Bartlett's justifications to exclude the Kleinfelder data were provided for the first time on December 28, 2010.

(Id. at pg. 32.)

Notably absent from these two sections, and the entire expert report, is any mention of the MWC-2 data from the years 1964-2007. Not one sample is mentioned. There is also no indication in the report that the MWC-2 sampling data was deficient because of infrequent testing, poor record-keeping, pumping irregularities, flow variations, or human error. These observations were only identified by Bartlett over three days of testimony in the Fed.R.Evid. 104 hearing. In view of the importance of the MWC-2 data to this model's reliability and whether contamination reached MWC-2's lower aquifer, these initial omissions — and lack of explanation — raise material credibility issues.

Bartlett now offers these reasons for excluding the MWC-2 data, which appeared for the first time after Bartlett's expert reports. In contrast to his expert report, which mentions none of these alleged foundational deficiencies, Bartlett now opines that the data "lacked information as to who drew the samples, when they were drawn, what testing methods were employed and other missing underlying lab data which undermine the data's credibility." This post hoc challenge to the credibility of the well data, required to be tested pursuant to a state statutory water quality monitoring regime, with confirmation by an independent laboratory testing and California Department of Health Services confirmation by a geohydrologist who is unlicensed in and unfamiliar with California well testing and public water quality standards is puzzling.

Two additional concerns are raised by the omission of the MWC-2 data from Bartlett's expert report that impact his credibility but do not exclude his testimony under Daubert. One, Defendants assert that the MWC-2 data was not mentioned in the expert report because Plaintiffs' counsel did not make the data available to Bartlett prior to building his model or, alternatively, he received the data but did not review it prior to modeling. Defendants maintain that such oversight invalidates the groundwater model. Two, Bartlett's reliance on the water level data from the California Department of Water Resources and the U.S. Geological Survey — to calibrate the "flow" component —, conflicts with his later assertion ( after the expert report) that the MWC-2 sampling data is unreliable because it lacks "chain of custody" documentation. Chain of custody evidence is a legal inquiry relevant to authenticity of underlying subject matter. Normally it goes to weight not admissibility of evidence.

This may have been the result of Plaintiffs' attorneys' failure to prove all the data or a choice by Dr. Bartlett not to utilize the data. (See, e.g., Bartlett Dep., April 7, 2010, 115:20-115:21) ("As I said, the transport part of the modeling is not calibrated.").

Bartlett's alleged lack of awareness of the MWC-2 data was developed during cross-examination on October 7, 2010:

Q. Now, the book talks about gathering data. Do you agree with me that before embarking on preparation of the model, all data should be considered? [. . .] A: Yes [] Q. Before preparing your expert report, which I'll — if I remember it correctly, it was December 9th, 2009. Or was it December 2nd? It was early December. A. Early December. Q. 2009. Before preparing that expert report, you never asked Mr. Marderosian for Meadowbrook 2 data; did you? A. No. We had the data. And we looked at it. It was in our database. My staff looked at it. I — I know where you're going to go with this because it was a misstatement in my deposition. But I want to make sure it's clear to the Court that, in fact, we did have the data. We did evaluate it. We were aware that the data from MWC-2 showed low or no concentrations predicted. I'm sorry. Measured. Q. Referring to your deposition, so that the Court is aware of what you're referring to as the misunderstanding. On page 124, line 21. "Question: So I am understanding that correctly, it was the Wilson and Stephens reports that tipped you off to the fact that there was more data than just the Arcadis 2007/2008 data?" "Answer: Yes. That is correct." And that's been changed from "yes, that is correct" to "No, that is not correct. My office had previously been provided with the Arcadis data along with the records from the previous California State of California Drinking Water Program related to Meadowbrook 2. The Stephens and Wilson reports contained information that was not referenced in the records found in the State of California Drinking Water Program System or the Arcadis data." Isn't it true, Mr. Bartlett, that though you're now testifying that your offices had the Meadowbrook 2 data, you personally had not seen that data before preparing your model and your expert report in December 2009? A. Our effort in this model was a team effort. There was several staff involved. I was part of that team. We — my colleague, when I got back from my deposition, reminded me that he had shown me this data. We had discussed it. I had forgotten that at the time of this deposition. This is several months after the expert report was issued. Q. So your testimony now is that you had reviewed it prior to preparing the model and your expert report? A. We were aware of it, yes. Q. You had reviewed it? Not aware of it. A. I had not seen the actual analytical sheets. Is that what you're referring to? Q. The chemical concentration data that was reported in those sheets or the sheets themselves. A. The data that we had was in a printout from the Department of Health Services, I believe. Q. Staff saw it. You didn't. Correct? A. Yes. Q. You received those data, those Meadowbrook 2 data in March of 2010; isn't that correct? A. I don't specifically recall. You mean as far as the data sheets that go with the output from the Department of Health Services? Q. Yes. A. I don't recall, Mr. Barg, I'm not sure. Q. Okay. Referring to your deposition again, page 135, line 4. "Question: And when did you receive those documents that you have in your hand, Exhibit 230?" And I'll represent to you that's the Meadowbrook data that's referred to on pages 133 and 134 [. . .]" "Answer: This would have been early, looks like probably early March." "Question: 2010?" "Answer: Yes. The narrative reports were issued, I think, in late February, so that would have been the first time I would have seen anything as far as reported values. And I don't recall whether Dr. Wilson or Dr. Stephens reports have the backup attached to them, they may have. I just don't recall." WITNESS: Now, I believe this is referring to data prior to 1990. MWC data prior to 1990. We did not have at the time the model was constructed. COUNSEL: Okay. So you obtained those data in March of 2010. That's four months after your December expert report. Correct? A. Correct. Q. And more than four months after you completed your model; correct? A. That sounds about right [. . .] Q. When you received the assignment from Mr. Marderosian, you knew that the assignment concerned possible contamination of a municipal supply well; correct? A. That's correct. Q. And you knew, based on your experience in working with suppliers of water, municipal water suppliers, that such wells were routinely tested for chemical constituents; correct? A. Yes.

(RT, Oct. 7, 2010, 168:15-173:9) (objections omitted)

In this Circuit, Bartlett's deposition and court testimony are subject to impeachment, not exclusion. Assessing the credibility of a witness is properly left to the jury. Whether Mr. Bartlett did not receive or consider the data is a disputed fact based on Mr. Bartlett's assertion that the data was received and reviewed by his staff and considered in relation to the model's formulation. Mr. Bartlett testified he knew that municipal wells underwent extensive testing and public compliance regulation, but he did not personally analyze/review the data prior to his model's completion. Nor was the data an any reason for excluding it mentioned in his expert report. The reasonableness of Mr. Bartlett's handling of the data ultimately present credibility issues for the jury.

The second area is Mr. Bartlett's review of and reliance on the water level data from the California Department of Water Resources and the USGS. According to his expert report, Bartlett used this data to calibrate the "flow" component of his groundwater model. (Doc. 777-2, pgs. 25-26.) Specifically, during oral argument on October 7, 2010, Bartlett explained that he calibrated his flow model with sampling data from the California Department of Water Resources and the USGS:

There is quite a bit of water level data collected in these monitoring wells over all these years. They basically begin when the first monitor wells were installed in [the] 1986 time period.
And then the water level data was used to calibrate the flow part of the model. So it's very important, of course, to get the flow components calibrated. And the way we do that is we make model runs and then compare the measurement of water levels to the prediction. And then we do a statistical evaluation of the water level that — the observed water level versus what the model predicts and do a statistical evaluation, which would include something called the percent root mean squared error. And in this case, that worked out to about 7 percent once we have the flow part of the calibration completed.

(RT, Oct. 7, 2010, 397:1-397:15.)

This data was downloaded from the internet without any foundational or chain-of-custody documents. Bartlett now claims that the MWC-2 sampling data is unreliable because it lacks the same foundational support:

Q. For the flow model, the flow portion of your model, you relied on water level measurements, water level data from various sources. Correct? A. Correct. Q. And water level data is not contamination data, it's not chemical concentration data, it is the level of the groundwater above mean sea level; correct? A. Yes, it's measured in a monitor well converted to elevation. Q. Okay. And I think you testified that you obtained such water level measurement data from the California department of water resources and the US geological survey. A. As well as the on site data that was provided from both IT and Arcadis. Q. Yes. Okay. Now, in what form were the water level data that you received from the department of water resources and US geological survey? That is to say, was it narrative, was it in tables, was it in maps, was it in diagrams? What was the form? A. My staff obtained that from online database sources. I wasn't directly involved in downloading that, so I can't speak specifically to the form of the database. But it was an online database source. Q. Okay. So you don't know the form that it took; that is, whether it was tabular, map, diagram, narrative? A. I don't. Q. Some of the water level data dated back to the 1960s and `70s that you obtained from — I'll just say DWR and USGS. A. Yes. Q. Now, how are water level measurements obtained in the field? How do you go about that? A. You take a water level sounder to a well. You put the instrument down the well and find out where the water level is. Measure that distance and convert it to elevation. Q. Okay. A. Knowing the elevation of the point at which you're doing the measuring from. Q. Okay. And the rest is just arithmetic? A. Yes. Q. So it sounds like the two components of the water level measurement are, one, the water level measured in that well. And, two, the precise location of that well location. A. Yes. Q. Who actually performed the water level measurements that were reflected in the department of water resources water level data? A. It's a compilation. I could not say. Q. Did the data that was obtained on line indicate the name of the person or entity who collected that water level measurement? A. I do not know. Q. Who determined the precise location where the water level data was measured for the department of water resources? A. I don't know. Q. And who was it that conveyed the information about the water level measurement and its location to the department of water resources? A. I don't know that either. Q. How do you know that the water level measurement data that you received from the department of water resources was accurate, both in terms of measurement and location? A. Well, often we'll plot out the data and it's often that we find data points that don't seem to match. We don't always use them. If they fit the regional pattern and it makes intuitive sense, then based on my professional judgment I feel it's a valid point, then we'll use it. But otherwise we may not use it. I'm not aware that there's any points like that in this case. We did — did screen out some data points that we could not reconcile with the regional water level pattern because, in this case, we weren't concerned about local small variations in the water level pattern. We're looking at the bigger picture and what's going around the model domain. Q. Did the water level measurement that you relied on from the department of water resources include water level measurements from municipal supply wells? A. I — I'm not exactly sure. Possibly. Q. And you relied upon these water level measurement data from the department of water resources and the US geological survey without knowing who conducted the test, who conducted the — not only the measurement, but the location information and who conveyed that to the department of water resources. You relied on that anyway? A. Yes, that's correct. Q. Even in the old data, the 1960s and `70s data. A. Yes. (RT, Oct. 7, 2010, 420:8-423:18)

Although Bartlett now argues that the MWC-2 data lacks the requisite foundation, he fully accepts the water level data without knowing anything about the foundation for that data. Bartlett offered little explanation why he distrusts the State of California water quality test data for MWC-2. There is a similar lack of clarity or justification why Bartlett calibrated the "flow" component of the groundwater model but not the "transport" component.

The Court in Dwyer ex rel. Dwyer v. Secretary of Health and Human Serv., No. 03-1202-V, 2010 WL 892250, at 148 (Fed. Cl. Mar. 12, 2010), stated: "An occasional misstatement of a study might be expected in any expert's analysis of such complex issues [. . .] A scientist might well pick data from many different sources to serve as circumstantial evidence for a particular hypothesis, but a reliable expert would not ignore contrary data, misstate the findings of others, make sweeping statements without support, and cite papers that do not provide the support asserted."

The next dispute is whether Bartlett's selective use and/or exclusion of monitoring and extraction well data invalidates the groundwater model. On December 28, 2010, Plaintiffs introduced Exhibit No. 11, a catalogue of citations to the December 15, 2010 testimony to support their groundwater arguments, in response to the Dr. Schmidt's independent expert report. On the monitoring/extraction well dispute, this document provides, in relevant part:

• There is no question that the data shows chromium concentration under the Beachwood Neighborhood [. . .]
• EW 4 sampling data is consistent with the model [. . .]
• EW 5 is sampling data is consistent with the model [. . .]
• M A sampling data is consistent with the model [. . .]
• The shallow depth of the off site monitoring wells is critical to evaluating the model because they are inadequate to evaluate the full 3 dimensional characteristics of the aquifer — this has been the problem all along with this site [. . .]
• MW 23 had its first high level (450) in June 2008. This is consistent with the drawdown theory since this occurred once MWC-2 was turned off in March of 2008 and contamination began to rise with water levels [. . .]
• The historical low levels in MW-23 can be explained by the shallow nature of monitoring well — only 58.5 feet deep and therefore only skimming the 8 feet of the aquifer [. . . .]

(Pl.'s Exhibit No. 11, Dec. 28, 2010, 3:22-4:11.)

On December 28, 2010, Bartlett explained that the true problem in this case is the "misleading" monitoring well network. It is Bartlett's position that the contaminated water is "below" the monitoring well network, i.e., the monitoring wells screen at shallow depths:

[A]s I've stated on several occasions through these hearings, the — problem with just contouring data and not evaluating what the data is telling you s that you can be misled. And I think that's what's happening here. We're being misled by the monitoring well network. Because the monitoring well network is too shallow, the — and as you get closed to MWC-2, they're very shallow. The contamination resides beneath the existing monitoring system.

(Rough RT, Dec. 28, 2010, 72:19-73:2.)

Bartlett, however, supports his model predictions with the citations to these same "shallow" monitoring wells. In particular, he relies on the "high hits" from MW-23 and MW-24 in June 2008 to establish the presence of chromium contamination in MWC-2 in 2008 (and prior to 2008). He also relies heavily on data from a number of extractions wells near the BAC site, over 1,000 feet away from MWC-2:

Bartlett testified at the Daubert hearings that his model was "conservative" based on a comparison between observed concentrations at two extractions wells (EW-4 and EW-5) and his "model predictions" at MWC-2 during certain points in time. For example, in 2001, Bartlett testified that EW-4 observed data showed chromium concentrations of 3,150 ppb, "much higher than what we predicted at Meadowbrook Well No. 2 [. . .] the hexavelant chromium value [at MWC-2] would have been roughly in this area here, which is about 50." (Rough RT, Dec. 28, 2010, 149:12.) It is undisputed that the two extraction wells used for comparison are located more than 1,000 feet north of MWC-2 and that several monitoring wells are located between the extraction wells and MWC-2. Besides the June 2008 "hit" at MW-23, Bartlett did not compare the observed concentrations at the monitoring wells with his model's predictions (at MWC-2).

The data that was obtained from MW-23, MW-24 [in 2008] showed concentrations as high as 450 parts per billion. Other concentrations were observed in wells farther to the north from this location. The Beachwood neighborhood data support the model. I want to make that perfectly clear. The problem and crux of the problem for this site is the monitoring well system is inadequate to properly characterize the site.

(Id. at 66:9-66:16.)

Bartlett further explained the impact of the MW-23 and MW-24 "high hits" on his model:

So those spikes [in MW-23 and MW-24] mean something to me. They're telling me there's contamination there. And in the case of MW-23, it was — it was first — that high peak was observed in June of 2008. That well was taken off line in March of 2008. And in October of 2008, MW-24 also saw a spike.
And so I would related the occurrence of that spike to the succession of operations at MWC-2.

(Id. at 71:1-71:6.)

Defendants, however, contend that this is one example of Bartlett's "selective" incorporation of observed monitoring well data, i.e., he incorporates the few data points that "fit" his model predictions to the exclusion of all others. Any data that is inconsistent with his model predictions is excluded based on his "professional judgment." MW-23 is illustrative of this criticism. As explained, Bartlett specifically relies on the 450 ppb sampling in June 2008, which Dr. Schmidt testified was an outlier, and excludes the (many) non-detect chromium observations at MW-23 (before and after June 2008):DATE RESULT

March 1994 10 June 1994 14.9 T July 1994 12.8 August 1994 8.1 Oct 1994 4.8 Nov 1994 3 Dec 1994 5.4 Feb 1995 3.1 April 1995 3 July 1995 7 Oct 1995 3 Jan 1996 3.1 April 1996 3.2 July 1996 3 Oct 1996 3 Jan 1997 3 April 1997 3 July 1997 3 Oct 1997 3 Jan 1998 3 April 1998 3 August 1998 3 Oct 1998 3 Jan 1999 3 April 1999 3 July 1999 3 Oct 1999 3 Jan 2000 3 Apr 2000 3 July 2000 3 Oct 2000 3 Jan 2001 3 April 2001 3.9 August 2001 3 Oct 2001 3 Jan 2002 3 April 2002 3 July 2002 3 Nov 2002 3 Jan 2003 3 May 2003 3.7 Sep 2003 3.6 Nov 2003 3 Feb 2004 3 April 2004 3 July 2004 3 Oct 2004 3 Jan 2005 3.5 May 2005 3 August 2005 4.1 Nov 2005 4.3 Feb 2006 1.9 J May 2006 2.3 J Nov 2006 2.2 J August 2007 3 Sep 2007 3 Oct 2007 3 Nov 2007 3.8 Dec 2007 3 Jan 2008 3 Feb 2008 3 March 2008 3 Jun 2008 450 July 2008 53 Aug 2008 9.5 Sep 2008 6.2 Nov 2008 (7) 2.0 (hex) Nov 2008 (24) 8.2 Dec 2008 18 Bartlett opines that the "spike" of 450 ppb in June 2008 was a result of the MWC-2's being taken off-line, i.e., the cessation of pumping activities at MWC-2 increased the water levels in MW-23 and other monitoring wells so that the chromium plume was fully present in the first layer (50 feet bgs). However, the abeyance of pumping at MWC-2 does not fully address the inquiry in view of: (1) MW-23 was screened to approximately 58 feet bgs (chromium should be present in portions of the second layer); (2) the low chromium observations in August, September, November, and December 2008; and (3) there were periods of low pumping at MWC-2 prior to June 2008, however, there was not a single observation above actionable levels (specifically, above 14.9 total chromium).

Despite the arguments to the contrary, Bartlett's opinions concerning the monitoring well network represent a factual scientific dispute to be determined by a jury. If Bartlett is to be believed, the monitoring well network was too shallow; the network screened the upper well, missing the chromium contaminations present in the lower layers. Although less persuasive, Bartlett states that the MWC-2's being taken off-line in March 2008 allowed the actual chromium contamination to be observed at monitoring well depths in June 2008. The present showing is marginal to demonstrate that the monitoring well data is "misleading" or otherwise scientifically objectionable (and not considered). At trial, Plaintiffs must demonstrate clear scientific justification to omit the monitoring well data, including the fourteen years of non-detects (1994 to 2008) and the restoration of low chromium observations at MW-23 immediately following the June 2008 hit. If such evidence is not developed, Mr. Bartlett's testimony is subject to the Court's reserved authority to strike it under Fed.R.Evid. Rules 104, 203, 206, 403, and 702.

The same scrutiny applies to the decision not to calibrate or "test" the model predictions to the field data from MWC-2, which reveal non-detect chromium levels for thirty years. Although the BAC Defendant's Daubert motion is denied without prejudice, subject to a reservation to strike the testimony at trial, Mr. Bartlett, on December 28, 2010, clarified that in his professional judgment the MWC-2 data was unreliable based on its infrequency, not chain of custody or other foundational deficiencies. His December 28, 2010 testimony raised a second concern, that the model is barred under Daubert's reliability prong because there is an absence of a known or potential rate of error:

See, e.g., Boyd v. City and County of San Francisco, 576 F.3d 938, 946 (9th Cir. 2009) (discussing the rate of error standard). Defendants also argue that the Bartlett model does not survive the "peer review" element of the Daubert inquiry given Dr. Schmidt's expert report. As discussed, however, those opinions present a factual scientific dispute concerning the presence of a continuous clay layer and annular seal in MWC-2.

Bartlett: So at MWC-2, we would expect to see something similar to what we observed in various monitoring wells at the site where you had quite a bit of variability over time. The Court: And the reason we don't is because there aren't enough samples. Bartlett: Correct. D's Counsel: So a graph of Meadowbrook 2 should look like the graph of Extraction Well 4 with some spikes and some lower concentrations and some spikes and some lower concentrations? Bartlett: Yes. D's Counsel: Okay. Well, let's look at your Figure 22. Isn't this a graph of Meadowbrook 2 predicted concentrations from your model? Bartlett: Yes. D's Counsel: The red are the concentrations; correct? Bartlett: That's correct. D's Counsel: It's not coming and going in spikes; is it? It's showing continuous contamination? Bartlett: Well, I would disagree. There are some spikes, there are some jumps in it. In general, it's not as severe as you might expect or as indicated in some of the monitoring wells. D's Counsel: But you're showing contamination in Meadowbrook 2 continuously, not coming and going. Correct? Bartlett: Yes. And as I've explained, the model is — simulates a — well, there's simplifying assumptions in it, so the concentrations that we're seeing are — they're accurate to the degree that they're illustrating what could potentially have occurred at MWC-2 over time. The Court: If we take this figure and we start in — let's start in 1975 where it gets close to a thousand micrograms per liter. Then we should have hits that are somewhere proximate to that range for the next 20, actually would be 30 years, to 2004, because there isn't any reduction in the model. And yet, even though we don't have many, we have a 1980, a 1989, a 1990, 1993, `95, 2001 and 2002, it gets to about the 9th, that you observed and none of those are even remotely within that range. Shouldn't they be showing somewhere in this range? Bartlett: Not necessarily, Your Honor. Because, as I pointed out, there are — there's more variability than the models simulate in the real world. And that's a function of the simplification we'd have to make to the model. And the predictions of the model. So — The Court: And so could we say, then, that that makes this model, at least in terms of its predictive ability, a guesstimate as opposed to an estimate? Because, of course, it doesn't show us anything that is remote to reality. Just looking at the model. You have to use your real world experience and knowledge. The model doesn't give you any of it [. . .] In other words, why do we not see even one test data observation that shows above the MCL for chromium in a 30 year period? Bartlett: My answer to that is frequency. Frequency of sampling. It's very infrequently sampled. And as we indicated, there's only nine samples collected. The Court: But if it's there at that level, wouldn't you have at least one? Bartlett: Not necessarily, Your Honor. The Court: Then this depiction, if you will — and I'm not arguing, I just want to understand. This depiction ought to show the spikes up and down, where it goes down to less than 1,000, even less than 100. It shouldn't show what looks like — almost like a bell shaped curve. Bartlett: Well, the model as we've been going through, the data, you can see there are concentrations in the aquifer. They are high. And so if you compare the observed concentrations in monitor wells, this is not unreasonable in that the potential for concentrations in excess of 500 micro grams per liter in that well is demonstrated by the model. The Court: If you were to do a statistical evaluation, a probability analysis that you would get not one hit that would be at a level that the model is supposed to predict, what would that be? THE WITNESS: I couldn't say. I don't know. (Rough RT, Dec. 28, 2010, 176:7-179:21.)

Although a close call, a scientific factual dispute remains whether the MWC-2 field data is a true reflection of chromium concentrations at MWC-2 during the relevant time-frame. According to Bartlett, he "considered" the MWC-2 field data, but ultimately concluded it was unreliable based on testing infrequency. The BAC Defendants point out that these alleged justifications were absent from his expert report and conflict with his deposition testimony and "review" of the USGS water level data. These arguments bear directly on Bartlett's credibility as a scientist and his knowledge and review of documents/data prior to composing his expert report. While Bartlett's justifications are late-developing, his review of the data was an exercise of professional judgment and cannot be decided as a matter of law. Whether this evidence was properly excluded under the guise of "professional judgment" is in total dispute. It must be weighed and decided by the trier of fact.

There is an additional controversy giving rise to the existence of a disputed issue of scientific material fact. Mr. Bartlett and Dr. Schmidt, the Court's independent expert, dispute the source for well water for MWC-2 and, critically, the presence and extent of Corcoran clay layer beneath MWC-2. Bartlett says there is a non-continuous clay layer at approximately 90 feet bgs and that the primary water source for MWC-2 is the contaminated shallow aquifer above 90 feet. Bartlett also opines that water enters the well through the open annulus above the well casing and through holes in the casing. Dr. Schmidt opines that a relatively impermeable Corcoran clay layer at 88 to 94 feet bgs acts as a barrier and MWC-2 draws its production water from a deeper, unconfined and uncontaminated aquifer. The scientific opinions on the breadth/expanse of the clay layer are conflicting and susceptible of at least two reasonable interpretations. These two contrasting views from scientists represent a factual scientific dispute, an issue that a jury must decide.

Under Daubert, a partially-calibrated model may not be scientifically sound. In the groundwater context given that field observations play such a critical role in ensuring reliability, calibration is scientifically important. See, e.g., Natural Resources Defense Council, Inc. v. Muszynski, 268 F.3d 91, 102 (2nd Cir. 2001) ("If the model is not closely calibrated, [i.e., if an agricultural use of land, for example, contributes significantly more than the expected amount of phosphorus to the adjoining water body], regulation of phosphorus concentrations in the reservoirs becomes an uncertain matter."). A half-calibrated model is, by definition, inadequately calibrated and excluded under Daubert. See Mitchell, 165 F.3d 778, 782 ("Under Daubert, any step that renders the expert's analysis unreliable [. . .] renders the testimony inadmissible.").

Plaintiffs contend that the BAC Defendants' Daubert motion fails because Defendants challenge the "matters clearly going to the weight of the[] testimony and not the admissibility of the[] testimony under Rule 702 and Daubert." (Doc. 792 at 1:17-2:2.) Under Daubert, however, any scientifically required step that renders the expert's analysis unreliable renders the testimony inadmissible. This is true whether the step completely changes a reliable methodology or merely misapplies that methodology. Bartlett's failure to calibrate his entire model to field data is premised on the alleged unreliability of the MWC-2 well test data and the monitoring well data. This is a highly risk-laden choice. If Plaintiffs cannot prove the unreliability of the well water quality (metals) test data, their groundwater expert and modeler has left them without an alternative explanation and justification for the existing model or an alternative model. Daubert does not prohibit proof of this theory, but will still exclude the Bartlett model. This must be determined at trial.

Two federal cases analyzing the admissibility of expert groundwater modeling under Rule 702, City of Wichita, Kansas v. Trustees of APCO Oil Corp. Liquidating Trust, 306 F. Supp. 2d 1040 (D. Kan. 2003) and Sterling v. Velsicol Chemical Corp., 855 F.2d 1188, are instructive.

In City of Wichita, 306 F. Supp. 2d 1040, the City brought a private CERCLA action against potentially responsible parties, seeking contribution for prior response costs and declaration of responsibility for future response costs. Id. at 1107. Prior to trial, defendants moved in limine to bar the testimony of Michael J. Smith, the City's expert witness regarding groundwater modeling. Id. The motions were denied without prejudice to reassertion in post-trial submissions. Following an eight-week bench trial on the CERCLA issues, defendants renewed and supplemented their motions to exclude Smith's trial testimony. Id.

The Court "decline[d] to credit the majority of Smith's testimony because it `fail[ed]' to meet the requirements of the Federal Rules of Evidence and applicable case law." Id. The expert used three proprietary codes: DYNFLOW for the flow component, DYNTRACK for the contaminant transport, and CHAIN, a government model, to analyze contaminant transport when there is degradation of the contaminants. The Court analyzed the utility and application of computer modeling to groundwater modeling:

Computer modeling is an accepted and, in appropriate circumstances, reliable method for use in determining groundwater flow and contaminant transport in an aquifer, and to evaluate the effectiveness of remedial alternatives. This is especially true when there is extensive and overlapping contamination from multiple sources, where contamination cannot be traced entirely to a specific source and when the extent of contamination is difficult to determine by `field methods' such as geoprobing and drilling test wells. Nevertheless, even in the best of circumstances, a model is only an estimate and the accuracy of the estimate depends to a considerable extent on the data selected for use in the computer model, the quality and reliability of that data and, of course, the skill of the modeler. Smith acknowledged that modeling results are not a perfect match to actual conditions but rather an approximation of what happens in reality. He conceded that his modeling techniques and methodology are not based on any specific guidelines or standards, but rather on his `professional judgment.'
Id. at 1108.

The Court described the calibration process as "a reasonable fit to the field data," Id. at 1109, and found that Smith took considerable leeway in his "professional judgment," which resulted in the Court giving "Smith's testimony and his modeling [. . .] little weight":

Examples of field data include source locations and degradation rates of the contaminant, in this case chlorinated solvents, which degrade from PCE to TCE to DCE to VC. If the area of contamination or "plume" simulated by the model does not match the area defined by field data, the modeled plume can be "truncated" to, in effect, eliminate any areas where the model says there is contamination but the field data says there is not. However, apparently using his `professional judgment,' Smith did not always `truncate' areas where the model showed contamination but the field data did not. He acknowledged, for example, that his model at APCO was not consistent with reality because it modeled contamination north or upgradient of APCO which was not confirmed by field data. Yet he did not truncate that area. The effect of this use of `professional judgment' was to allocate contamination to APCO which APCO could not have caused.

In this case Bartlett has rejected and explained the field data based on the exercise of "professional judgment."

In Sterling, 855 F.2d 1188, the Sixth Circuit affirmed the denial of a motion to exclude a groundwater model on the basis that "district court carefully considered the plaintiffs' model, including its components and the data it utilized." Id. at 1199. Sterling considered a class action against a chemical manufacturer for personal injuries and property damage allegedly caused by contaminant exposure from its chemical waste burial site. Judgment was entered in favor of plaintiffs. On appeal, the corporation attacked the plaintiffs' expert model as unreliable because it failed to utilize all of the relevant domestic well data. The Sixth Circuit disagreed, finding that the failure to incorporate selected well samplings conducted after 1978 did not invalidate the model and conclusions based on the model:

In 1978, the EPA tested nineteen of twenty-two domestic wells (not including four USGS wells) for chemical contamination. After 1978, only one of the five wells from which the representative plaintiffs drank or otherwise used water and seven of the nineteen remaining wells were ever again tested. Data showing the level of chemical contamination of these remaining wells was available only upon a limited and sporadic basis. While the plaintiffs did not incorporate these post-1978 well samplings into their model, they thoroughly reviewed and utilized the data to confirm the water table contour component of their model. Plaintiffs also did not utilize soil column studies, such studies having been abandoned by Velsicol itself and found non-credible by the district court. However, the plaintiffs utilized all other relevant data including, but not limited to, the 1964 Rima Study, the 1978 Sprinkle Study, the data generated by the State of Tennessee in 1970, all the information generated by AWARE, and all of the monitoring information by Velsicol, the State of Tennessee, and the EPA. Because the district court carefully considered the plaintiffs' model, including its components and the data it utilized, we cannot say that it erred in using the model in concluding that Velsicol's chemicals contaminated the plaintiffs' wells as early as 1970.

These two cases demonstrate the level of scrutiny that must be applied to the Bartlett model. First, like the expert in City of Wichita, Bartlett exercised his professional judgment to depart from the field data, which revealed that there was no hexavalent contamination at that MWC-2 well, at that time. Bartlett's use of his professional judgment resulted in an allocation of hexavalent chromium contamination at MWC-2 that manifestly deviated from the field data. Scientific reliable methodology, as described by Mr. Stephens and the authoritative treatises, suggests further examination after some number of tests revealed no hexavalent chromium contamination at MWC-2 or the nearby monitoring wells. Although it is undisputed that modeling is not an exact science, the sheer volume of non-detects followed by a lack of calibration or adjustment raises serious concerns.

Sterling is distinguishable from this case. There, the expert excluded truly "irregular" testing data, however, the expert calibrated his model to the EPA's testing in 1978, as well as the remaining scientific studies and monitoring data. The exclusion of the post-1978 data was supported by scientific principles. Neither of those steps were taken in this case.

Plaintiffs next distinguish Ramsey, 111 F. Supp. 2d 1030, and the other groundwater modeling cases cited by the BAC Defendants on grounds that those cases did not involve challenges to the reliability of the underlying data. Plaintiffs argue that the factual dispute on the data's reliability creates a triable issue of fact over the issue of groundwater contamination in the Beachwood neighborhood. See Ralston v. Smith Nephew Richards, Inc., 275 F.3d 965, 970 n. 4 (10th Cir. 2001) (the proponent of expert testimony bears the burden of showing that its proffered expert's testimony is admissible). Plaintiffs proffer several arguments to support of their position.

Plaintiffs first argue that the sampling data was properly excluded from its expert's model because there was insufficient documentation and Defendants "lack credibility" with respect to "information related to the MWC-2 site." In particular, Plaintiffs contend that Defendant Meadowbrook's sampling data is untrustworthy because it was conducted pursuant to the "honor system":

[P]rior to 2007 all the sampling of MWC-2 which was submitted to the state, was performed by Meadowbrook under the "honor system." Clearly the maintenance of an illegal well is hardly indicative of a company who should be trusted under the "honor system." Further, Meadowbrook is the same entity which had previously been reprimanded for using a "pool kit" to sample its wells — again hardly indicative of a company concerned about abiding by proper sampling protocol. The complete lack of expected and required supporting documentation which should accompany this alleged data, coupled with this conduct by Meadowbrook completely undermines any sense of reliability to this data. Certainly, plaintiffs and their experts are well justified in viewing the scientific reliability of this "honor system" sampling from Meadowbrook with great skepticism.

(Doc. 792 at 4:4-4:17.)

Plaintiffs further theorize that BAC Defendants, specifically Defendant Merck, engaged in a "conspiracy to destroy and/or manipulate the BAC site sampling records":

Plaintiffs focused investigation in this litigation has uncovered a strategic, well-organized and methodic approach by the Merck defendants to impede the ability of any future litigant to establish exposure to contaminants from the BAC facility. The Merck defendants' strategically avoided the collection of contemporaneous offsite data and did not provide the residents surrounding the BAC Facility with any information which would have placed them on notice of the need to collect such data themselves. The Merck defendants knew that with the passage of time, the ability to obtain such critical and contemporaneous exposure data would be forever lost to plaintiffs. The avoidance of the collection of this data was clearly designed to protect defendants in any future litigation by forcing plaintiffs to forensically reconstruct exposure to contaminants and then accuse plaintiffs of having no evidence to support their claims. This conduct by the Merck defendants raises genuine issues of fact concerning the reliability and credibility of their evidence submitted in support of their motion, and certainly precludes any argument that plaintiffs should be precluded from proceeding with their claims due to the lack of evidence that the Merck defendants themselves should have collected.
Finally, the Merck defendants piously warn this court as to efforts by plaintiffs to somehow "distract" the court from issues that are relevant to their pending motions. This defensive statement is no doubt aimed at the arguments related to the egregious conduct of the Merck defendants in relationship to the manipulation and control of information made available to the public through state and federal agencies concerning this contaminated site. The Merck defendants no doubt have grown accustomed to dominating and controlling all the information related to the site, and plaintiffs' close scrutiny of their conduct is both threatening and unnerving to the defendants.

(Id. at 6:7-7:9.)

The allegations of improper documentation and conduct are based on hindsight and not supported by any specific facts. Plaintiffs fault the RWCQB for not instituting special procedures for consultants paid by the allegedly offending-entities. They criticize RWQCB's uniform sampling standards, as far back as 1964, including the documentation requirements at the monitoring site, lack of state personnel at the site, and means of authentication. Plaintiffs, however, overlook the fact that identical sampling evidence has been relied on in a number of federal court cases, including City of Fresno v. United States, ___ F. Supp. 2d ___, 2010 WL 1662476 (E.D. Cal. April 22, 2010), Santa Clara Valley Water Dist. v. Olin Corp., 655 F. Supp. 2d 1066 (N.D. Cal. 2009) and Waste Management of Alameda County, Inc. v. East Bay Regional Park Dist., 135 F. Supp. 2d 1071 (N.D. Cal. 2001). Plaintiffs have cited no legal authority holding that sampling authorized and then certified by the RWQCB is unreliable, otherwise lacking credibility, or that it is subject to manipulation by independent testing laboratories or state employees who certify testing results. The MWC-2 intervals in this case are criticized for not meeting the tri-annual testing frequency called for by State regulations.

Plaintiffs cite United States v. San Diego Gas Elec. Co., Nos. 06-CR-0065-DMS; 07-CR-0484-DMS, 2009 WL 4824489 (S.D. Cal. Aug. 31, 2009) to support their "inadequate testing" arguments. That case is factually distinguishable. First, it involved with allegations of criminal violations of the asbestos NESHAP work practice standards. Those standards specify the government is required to prove absestos content — an essential element of the charge — through: (1) a representative sample; and (2) a specified test method, as prescribed by the federal regulations. In San Diego Gas Elec. Co., the government's experts did not satisfy NEHSAP's standards and the evidence was excluded. No similar criminal statute applies in this case. Second, the Court noted that "the Government should be held to precise evidentiary standards before evidence of this nature is admitted and Defendants are subjected to criminal exposure." Id. at 9. This is not a criminal case.

Here, the samples collected at MWC-2 and monitoring wells at or near the BAC site were tested and/or verified by the RWQCB. At the time the samples were taken, they complied with the RWQCB's scientific and chain-of-custody requirements. There is no evidence, other than argument, to show otherwise. An attack on all samples ever collected and/or verified by the RWQCB is without support in the record. In addition, Plaintiffs' allegations of a Merck-led conspiracy at the BAC site are speculative and unsupported by any factual citations. To create a genuine issue of disputed fact on the Phase 1 issues, the disputed conduct must be based on facts not argument.

Plaintiffs levy a number of non-scientific attacks on the entities/agencies involved in collecting and verifying the data in an attempt to prove a Merck-led conspiracy. Specifically, Plaintiffs characterize the RWQCB for being "over-worked and understaffed" and Meadowbrook Water Company as a "small and unsophisticated water company." At the same time, Plaintiffs note Merck's size, corporate status, and alleged experience with environmental disputes. These arguments are all disputed.

Second, Plaintiffs refer to the supplemental declaration of their groundwater expert, Mr. Bartlett, to argue that there is a triable issue on the global groundwater issue because there is an alleged dispute over the reliability of sampling data from MWC-2. In his supplemental declaration, Bartlett opines:

I argue [. . .] that [] water quality data from well MWC-2 is both suspect and not representative of the historical concentrations of chromium that were pumped from this well. The data collected by Meadowbrook Water Company prior to 2007 is not properly documented; there are no records concerning how or where the samples were collected, who collected the sample, the operating condition of the pump in the well, the quality assurance/quality control data from the laboratory, the name of the lab that did the analysis, etc.

(Doc. 777 at ¶ 5.)

Bartlett is not licensed in California and there is no record evidence that he has an expertise in California water sample collection and water quality assurance formalities required by the State of California. It is unknown if he has ever personally taken or observed sampling at groundwater, domestic, or monitoring wells in California or any other state. Second, he takes inconsistent positions concerning the validity of data provided to the RWQCB and there is a dispute whether he had personal knowledge of the MWC-2 data prior to building his model. As discussed above, none of MWC-2's alleged foundational deficiencies are delineated in his Rule 26 expert report.

Following the October 15, 2010 oral argument, Meadowbrook Water Company was directed to produce the relevant foundational documents, i.e., the company's laboratory records for the relevant time periods, authenticated by a declaration from its custodian of records. (Doc. 863.) Meadowbrook Water Company complied with the request on October 20th by filing Constance Farris' declaration and attaching the lab results. In her declaration, Ms. Farris states that she maintains the business records of Meadowbrook Water Company and the laboratory results were kept in the ordinary course of business. These laboratory results are summarized as follows: Well Collection/ Sampler Lab Reading (Chr. Test Date VI)

MWC-2 Aug. 1974 Signed by R. Merced Brooks County ND 10 MWC-2 March 19, Signed by M. Twining 1980 Banuelos Labs, Inc. ND Cal ML MWC-2 June 9, P. Espinosa Cal. Water 1987 Labs ND 10 MWC-2 Jan. 3, D. Adams Cal. Water 1990. Labs ND 10 MWC-2 March 9, T. Stevens APPL., 1993 Inc. ND 10 MWC-2 Dec. 21, T. Stevens APPL., 1995 Inc. ND 10 MWC-2 Feb. 17, T. BSK 1995 Stephenson Laboratori ND 10 es, Inc. MWC-2 Sep. 13, MWC — BSK 2.0 ppb 2001 Laboratori es, Inc. MWC-2 Feb. 12, MWC BSK Total Chrom 2002 Laboratori (12 ppb) es, Inc. Chrom IV (1 ppb) MWC-2 March 11, MWC BSK Total Chrom 2005 Laboratori (ND) es, Inc. MWC-2 March 2, BC BC Total Chrom 2007 Laboratories Laboratori (ND, 3.0) es Chrom IV (1.8 and 1.6 ppb) MWC-2 12 samples Arcadis Arcadis Total Chrom (May 15, (ND) 2007 to Sep. 18, Chrom IV (2.1 2008) ppb) The declaration of Ms. Farris and the attached sampling records rebut the quality control criticisms and other alleged foundational deficiencies, however, they indicate that Meadowbrook did not produce well samples a tri-annual basis, the regulatory standard.

2. Conclusion on Groundwater Pathway

The degree of variance between Bartlett's model and the well data requires a factual foundation to decide the truth of Bartlett's grounds for minimizing most of the observed test data. This credibility determination will bear on the ultimate admissibility of the model. The Court reserves the right to exclude the model after hearing the evidentiary foundation for the model at trial.

The BAC Defendants' Daubert motion is DENIED subject to the express reservation described in detail in this Memorandum Decision. If proceedings during trial ultimately reveal that Bartlett's model is not supported by a proper scientific foundation, it will be subject to a motion to strike at trial under Daubert.

Because a scientific factual dispute remains as to the admissibility of the Bartlett model, which was proffered to demonstrate chromium contamination via the groundwater pathway (specifically, in MWC-2 from 1969 onward), that portion of the motion for partial summary judgment is also DENIED without prejudice.

E. Pathway # 3 — Private Wells

The BAC Defendants argue that summary judgment is proper because Plaintiffs have provided no evidence of contaminant exposure via the private well pathway. In particular, Defendants assert that "neither plaintiffs nor their experts have produced a shred of sampling data to show that there was any elevated concentration of any contaminant in any private well used for drinking water at any time."

The terms "private well" and "domestic wells" are synonymous.

Defendants include a water sampling history in their motion, (Doc. 677-1), but a more specific "private well" fact recitation is required: Defendants assert that two offsite well surveys were conducted by consultant IT Corp., the first in 1992. This survey was conducted under the supervision of the RWQCB in the Beachwood neighborhood immediately after the discovery of off-site contamination at the M A Market, which is located across Santa Fe Drive from the BAC Site. In carrying out the survey, BAC and IT searched public records and walked door-to-door through the northeast portion of the Beachwood neighborhood looking for private domestic wells. Most of the wells were shallow and dry. The survey found only one well in use at 2942 Elm Avenue. The well water was tested and no hexavalent chromium or arsenic were found. The results were provided to the county and homeowners, and those homes that were not already connected to the Meadowbrook system were connected at BAC expense.

This summary is based on a review of the parties' briefing and arguments advanced at the December 29, 2010 hearing.

In 1999, IT Corp. conducted a second well survey under the supervision of the RWQCB. They surveyed a well to the north of the BAC Site because the groundwater flow had shifted to the north (north was then the downgradient direction of flow). The second survey also included a door-to-door survey to confirm the presence of abandoned or active wells. Two active domestic wells were identified and tested. According to Defendants, background concentration levels of hexavalent chromium were detected in one of the wells (3.9 ug/l). The second well contained no detectable levels of hexavalent chromium. Arsenic was not found in either well.

According to Defendants, these facts demonstrate that concentrations of hexavalent chromium and arsenic in domestic wells were at or below background concentrations and well below the state MCL for chromium (50 ug/l).

Plaintiffs rely on the expert testimony of Richard Laton, a purported expert in the field of hydrology and hyrdogeology. Laton opines that domestic wells utilized by plaintiffs have been impacted by chemicals of concern from the former BAC facility resulting in significant pathways of exposure to those plaintiffs; and given the detections in the monitoring well network, "it is a scientific certainty that hexavalent chromium and chromium reached residential wells in the Beachwood neighborhood." (Doc. 778 at ¶ 8.)

Dr. Laton next opines that the monitoring well data sufficiently establishes that contamination reached several domestic wells in the Beachwood neighborhood:

While the Merck defendants attempt to use the lack of monitoring well data to justify their opinion that no private well contamination existed, the lack of data should not be interpreted as a lack of contamination. Given Merck's failure to implement a sufficient investigation of the private/domestic wells surrounding the BAC facility, the Court must consider the limited data that was available in surrounding wells and the Bartlett model confirming the contamination in the private/domestic wells. Further, many of these wells had been taken out of service by the time the contamination was discovered in the M A monitoring well. There is no indication that Merck undertook a comprehensive effort to sample the wells prior to abandonment.

(Id. at ¶ 10.)

Although Dr. Laton's criticisms concerning the "failure to investigate" add nothing to the Phase 1 inquiry, his comments concerning the Bartlett model are significant. According to Plaintiffs, the monitoring well data, together with the Bartlett model, establish that MWC-2 and the private domestic wells drew water from the same contaminated plume:

And let's not forget that prior to their discovery of off site contamination in 1992, let's not forget that there is no off site data that we located or that existed between 1969 and 1992. That's a 23-year period of time that th[e] plume, according to Doug Bartlett and Richard Laton, my hydrologist, who has not been challenged in this case. They are of the opinion that the plume existed prior to June 9, 1992, when the M A Market well detected the presence of hexavelant at eight times drinking water level, at 400 part per billion. The significance, as Mr. — Mr. Bartlett said was the very first day, it showed 400 parts per billion. Ergo, the plume could very well have been there prior to that period of time.

(Rough RT, Dec. 29, 2010, 89:18-90:7.)

There is some merit to Plaintiffs' position. Although it is Plaintiffs' burden to establish private well contamination, that proof requires admission — and acceptance — of the Bartlett model. The Bartlett model is relevant to a degree, because it suggests that MWC-2, the monitoring wells, and the domestic wells drew water from the same contaminated plume causing injury to the well users and the community. At the same time, however, Plaintiffs have a proof problem. Despite their arguments concerning the lack of sampling and alleged "plume" sharing, neither Dr. Laton nor Dr. Bartlett identify a specific private well allegedly containing hexavalent chromium or arsenic during the relevant time-period. Nor do they identify a specific concentration of contamination allegedly present in any private well.

There is a proper approach to this evidence and how to present it to the jury, particularly in view of its dependence on the Bartlett model. It has been previously determined that Bartlett's testimony is conditionally admissible, subject to a motion to strike. The same conditions applies to the surface water pathway. If Bartlett's model and testimony are ultimately inadmissible at trial, the modeling evidence showing domestic well contamination will also be excluded.

There is a sharp dispute over the depth and reliability of the monitoring well network. Bartlett excludes most of the monitoring well data, characterizing it as flawed and inaccurate. The Defendants disagree, arguing that Bartlett's exclusion of the data has no scientific basis and reflects his model's lack of scientific reliability. Whether the exclusion of the monitoring well data is a "function of a scientist's professional judgment" is a subject for trial.

The nature and extent of the "shared plume" remains in dispute. To date, however, Plaintiffs have not provided evidence except argument and hyperbole addressing Merck's alleged corporate misconduct and failure to provide environmental notice. Plaintiffs failed to prove the BAC Defendants' lack of diligence and corporate misfeasance during the Daubert hearings. It was marginally relevant, at best.

The BAC Defendants' motion for partial summary judgment on the private well pathway is DENIED subject to the express reservations described in this Memorandum Decision. Plaintiffs must make a particularized showing of private domestic well contamination based on admissible evidence.

In their opposition to Defendants' motion, Plaintiffs' state:

In future plaintiff-specific phases, plaintiffs will provide evidence related to specific domestic wells from which a specific plaintiff was exposed and at what levels.

(Doc. 792 at 23:3-23:15.)
Defendants argue that this an express acknowledgment of Plaintiffs' failure to meet their Phase 1 burden, i.e., Plaintiffs have no evidence concerning the level of contamination from the domestic well pathway. However, for the reasons explained in this Memorandum Decision, this in part depends on conditionally admissible evidence.


For the reasons stated:

1) Defendants' Daubert motion to exclude Sears' air model is DENIED;
2) Defendants' motion for summary judgment is GRANTED with respect to the 1994 Risk Assessment;
3) Defendant's motion for summary judgment is DENIED with respect to the air pathway;
4) Defendants' Daubert motion to exclude Bartlett's ground water model is DENIED, subject to exclusion at trial as discussed in this Memorandum Decision;
5) Defendant's motion for summary judgment with respect to the groundwater pathway is DENIED, subject to the admissibility of Bartlett's groundwater model; and
6) Defendant's motion for summary judgment with respect to the private well pathway is DENIED; subject to the admissibility of Bartlett's groundwater model.


Summaries of

Valencia v. Franklin County Water District

United States District Court, E.D. California
Jan 5, 2011
761 F. Supp. 2d 1007 (E.D. Cal. 2011)

rejecting defendant's challenge to expert's opinion that relied on data and information as unreliable

Summary of this case from Victorino v. FCA U.S. LLC
Case details for

Valencia v. Franklin County Water District

Case Details

Full title:ABARCA, RAUL VALENCIA, et al., Plaintiffs, v. FRANKLIN COUNTY WATER…

Court:United States District Court, E.D. California

Date published: Jan 5, 2011


761 F. Supp. 2d 1007 (E.D. Cal. 2011)

Citing Cases

Romero v. S. Schwab Co.

Rule 702 requires that an expert possess "knowledge, skill, experience, training, or education" sufficient to…

Romero v. S. Schwab Co.

Rule 702 requires that an expert possess "knowledge, skill, experience, training, or education" sufficient to…