The opinion of the court was delivered by: Oliver W. Wanger United States District Judge
FINDINGS OF FACT AND CONCLUSIONS OF LAW RE PLAINTIFFS' REQUEST FOR PRELIMINARY INJUNCTION AGAINST IMPLEMENTATION OF RPA COMPONENT 2 (a/k/a Action 3)(Doc. 433)
Plaintiffs, San Luis & Delta Mendota Water Authority (the "Authority") and Westlands Water District ("Westlands"), move for a preliminary injunction ("PI") against the implementation of Reasonable and Prudent Alternative ("RPA") Component 2 set forth in the United States Fish and Wildlife Service's ("FWS") December 15, 2008 Biological Opinion, which addresses the impacts of the coordinated operations of the federal Central Valley Project ("CVP") and State Water Project ("SWP") on the threatened delta smelt (Hypomesus transpacificus) ("2008 Smelt BiOp" or "BiOp"). Doc. 433.
Plaintiffs State Water Contractors; Metropolitan Water District of Southern California; Kern County Water Agency and Coalition for a Sustainable; Stewart & Jasper Orchards, et al.; and the Family Farm Alliance join in the motion. Docs. 449, 451 & 453. Plaintiff-Intervenor Department of Water Resources ("DWR"), the operator of the SWP, partially joins. Doc. 452.
Federal Defendants and Defendant Intervenors opposed. Docs. 469, 473. Plaintiffs replied. Docs. 487, 491, 495, 497 & 507. The motion came on for an evidentiary hearing on April 2, 5, 6, and 7, 2010. Docs. 644, 652, 653 & 654. The parties were represented by counsel, as noted in the record.
After consideration of the testimony of the witnesses, the exhibits received in evidence, the written briefs of the parties, oral arguments, and the parties' proposed findings of fact and conclusions of law, the following findings of fact and conclusions of law concerning the motion for interim relief/preliminary injunction are entered.
To the extent any finding of fact may be interpreted as a conclusion of law or any conclusion of law may be interpreted as a finding of fact, it is so intended.
The 2008 Smelt BiOp, prepared pursuant to Section 7 of the Endangered Species Act ("ESA"), 16 U.S.C. § 1536(a)(2), concluded that "the coordinated operations of the CVP and SWP, as proposed, are likely to jeopardize the continued existence of the delta smelt" and "adversely modify delta smelt critical habitat." BiOp at 276-78. As required by law, the BiOp includes an RPA designed to allow the projects to continue operating without causing jeopardy to the species or adverse modification to its critical habitat. Id. at 279. The RPA includes various operational components designed to reduce entrainment of smelt during critical times of the year by controlling exports out of and water flows into the Delta. Id. at 279-85.
Component 1 (Protection of the Adult Delta Smelt Life Stage) consists of two Actions related to Old and Middle River ("OMR") flows.
* Action 1, which is designed to protect upmigrating delta smelt, is triggered during low and high entrainment risk periods based on physical and biological monitoring. Action 1 requires OMR flows to be no more negative than -2,000 cubic feet per second ("cfs") on a 14-day average and no more negative than -2,500 cfs for a 5-day running average. Id. at 281, 329.
* Action 2 of Component 1 is designed to protect adult delta smelt that have migrated upstream and are residing in the Delta prior to spawning. Action 2 is triggered immediately after Action 1 ends or if recommended by the Smelt Working Group ("SWG"). Flows under Action 2 can be set within a range from -5,000 to -1,250 cfs, depending on a complex set of biological and environmental parameters. Id. at 281-82, 352-56.
At issue here is Component 2 (Action 3) (Protection of Larval and Juvenile Delta Smelt), which requires OMR flows to remain between -1,250 and -5,000 cfs, beginning when Component 1 is completed, when Delta water temperatures reach 12° Celcius ("C"), or when a spent female smelt is detected in trawls or at salvage facilities. Id. at 282, 357-58. Component 2 remains in place until June 30 or when the Clifton Court Forebay
Id. at water temperature reaches 25° C. 282, 368.
Component 3 (Improve Habitat for Delta Smelt Growth and Rearing) requires sufficient Delta outflow to maintain average mixing point locations of Delta outflow and estuarine water inflow ("X2") from September to December, depending on water year type, in accordance with a specifically described "adaptive management process" overseen by FWS. Id. at 282-83, 369.
Under Component 4 (Habitat Restoration), DWR is to create or restore 8,000 acres of intertidal and subtidal habitat in the Delta and Suisun Marsh within 10 years. Id. at 283-84, 379.
Under Component 5 (Monitoring and Reporting), the Projects gather and report information to ensure proper implementation of the RPA actions, achievement of physical results, and evaluation of the effectiveness of the actions on the targeted life stages of delta smelt, so that the actions can be refined, if needed. Id. at 284-85, 328, 375.
Plaintiffs' request temporary injunctive relief on the following grounds:
1) the district court has already found that the United States Bureau of Reclamation ("Reclamation") failed to comply with the National Environmental Policy Act ("NEPA") in implementing the 2008 Smelt BiOp RPA; and.
2) the 2008 Smelt BiOp violates the ESA and is arbitrary, capricious, and contrary to law because:
a) various aspects of the BiOp's baseline and effects analysis are flawed, undermining the overall jeopardy conclusion, causing overstatement of the effects of the proposed action and imposition of overly-broad and overly-restrictive RPA Components;
b) the severe OMR flow restrictions in RPA Components 1 and 2 are unsupported by the best available science and the data in the 2008 Smelt BiOp; and
c) Component 3 ("The Fall X2 Action") is arbitrary and capricious, because it is without factual or scientific justification and/or is not supported by the best available science, compelling a finding of likelihood of success on the merits.
Plaintiffs further claim that the implementation of RPA Components 1 and 2 will cause them continuing irreparable harm and that the public interest and balance of hardships favor injunctive relief.
RPA Component 1 has ended for the 2009-2010 water year, mooting any request for injunctive relief against its imposition. Component 3 is not set to begin until September, and Plaintiffs do not presently seek injunctive relief against its operation. Barring unforeseen circumstances, the parties' cross-motions for summary judgment will be heard and decided before September. Components 1 and 3 are not addressed in this decision.*fn1
Plaintiffs' injunction request has been modified over time. Originally, Plaintiffs sought an injunction against implementation of RPA Component 2 and enforcement of the incidental take limits in the BiOp. See Doc. 435 at 2-4.
* In place of Component 2, Plaintiffs sought to require Federal Defendants and DWR to use a Potential Entrainment Index ("PEI") to estimate cumulative entrainment loss of delta smelt. If the PEI estimate of cumulative loss is less than or equal to 7%, no pumping restrictions should be imposed; if the PEI estimate of cumulative entrainment loss exceeds 7%, FWS shall be responsible for setting OMR flows under the range specified in Component 2 of the BiOp. Doc. 435 at 3.
* Plaintiffs requested that the Incidental Take Statement ("ITS") be recalculated based on a higher Cumulative Salvage Index ("CSI") of 11.36 for adults. Doc. 435 at 4.
* In the alternative, if the above remedies are not imposed, DWR requested that that the Court impose the interim remedial operational conditions imposed following summary judgment in NRDC v. Kempthorne, 1:05-cv-1207. Doc. 452 at 2.
Although Plaintiffs never filed a written
modification of their request for relief, at the evidentiary hearing Plaintiffs withdrew their request to enjoin enforcement of the ITS and their request to implement the PEI in place of RPA Component 2 of the RPA. 4/2/10 Tr. 90:4-12; 4/7/10 Tr. 243:23-244:8. Instead, Plaintiffs now propose that Component 2 be replaced by a flat -5,600 cfs ceiling on negative OMR flows during the remainder of the implementation period for Component 2. Id.; see 4/2/10 Tr. 208.
Injunctive relief, whether temporary or permanent, is an "extraordinary remedy, never awarded as of right." Winter v. Natural Resources Defense Council, 129 S.Ct. 365, 376 (2008); Weinberger v. Romero-Barcelo, 456 U.S. 305, 312 (1982). Four factors must be established by a preponderance of the evidence to qualify for temporary injunctive relief:
1. Likelihood of success on the merits;
2. Likelihood the moving party will suffer irreparable harm absent injunctive relief;
3. The balance of equities tips in the moving parties' favor; and
4. An injunction is in the public interest.
Winter, 129 S.Ct. at 374; Am. Trucking Ass'n v. City of Los Angeles, 559 F.3d 1046, 1052 (9th Cir. 2009).
1. The agency action is the coordinated operation of the CVP and SWP, pursuant to an Agreement for the Coordinated Operation of the two projects ("COA").
2. According to the Rivers and Harbors Act of 1937, the dams and reservoirs of the CVP "shall be used, first, for river regulation, improvement of navigation and flood control; second, for irrigation and domestic uses; and, third, for power." 50 Stat. 844, 850.
3. The CVP was reauthorized in 1992 through the Central Valley Improvement Act ("CVPIA"), which modified the 1937 Act and added mitigation, protection, and restoration of fish and wildlife as co-equal project purposes. Pub. L. 102-575 § 3402, 106 Stat. 4600, 4706 (1992). One of the stated purposes of the CVPIA is to address impacts of the CVP on fish and wildlife. § 3406(a). The CVPIA made environmental protection and water deliveries co-purposes.
4. This case presents a critical conflict between these dual legislative purposes, providing water service for agricultural, domestic, and industrial use, versus enhancing environmental protection for fish species whose habitat is maintained in rivers, estuaries, canals, and other waterways that comprise the Sacramento-San Joaquin Delta.
5. It is of manifest significance to the public interest that DWR, a co-operator and the State contractual partner of Reclamation, disagrees with at least some portions of the RPA and seeks injunctive relief against the calendar-based ceiling in RPA Component 2.
B. Facts Relevant to NEPA Claim.
6. It is undisputed that neither FWS nor Reclamation engaged in any NEPA analysis in connection with preparation or implementation of the 2008 Smelt BiOp.
7. It is also undisputed that on November 13, 2009, the Court entered an Order granting San Luis Plaintiffs' motion for summary judgment on their claim that Federal Defendants violated NEPA when they implemented the 2008 Smelt BiOp without conducting the required NEPA analysis. Doc. 399.
8. FWS did not engage in a systematic consideration of impacts to the human environment and/or consideration of alternatives that took into account those impacts, ordinarily performed as part of a NEPA review.
C. Facts Relevant to ESA Challenges
(1) Status of the Species.
9. The delta smelt was listed as a threatened species under the ESA on March 5, 1993. 58 Fed. Reg. 12,584 (March 5, 1993). Critical habitat was designated for the delta smelt on December 19, 1994. 59 Fed. Reg. 65,256 (Dec. 19, 1994).
10. The threatened delta smelt, one of the most abundant species in the Bay-Delta ecosystem as recently as thirty years ago, is in imminent danger of extinction. Doc. 94, Findings of Fact Re Plaintiffs' Motion for Preliminary Injunction, ## 1-2. The experts agree that there is no current population count for delta smelt. 4/2/10 Tr. 174 (Feyrer); 4/5/10 Tr. 67 (Newman); 4/5/10 Tr. 231 (Hilborn); 4/6/10 Tr. 95 (Deriso). However, the species' relative abundance from year-to-year is monitored using the Fall Midwater Trawl index ("FMWT") prepared by the California Department of Fish and Game ("CDFG"), as well as other abundance indices. 4/2/10 Tr. 174-75. The FMWT shows a continuously and precipitously declining trend in delta smelt abundance in recent years, registering a series of record-breaking lows. 4/2/10 Tr. 176-78. That trend has continued in the last two years, with the FMWT declining from 23 in 2008 to 17 in 2009, the lowest value ever recorded. Id. The population growth rate for delta smelt has been "quite negative" for the last ten years. 4/5/10 Tr. 232. The stock-recruitment relationship for delta smelt, which shows the relationship between adults (i.e., the "stock" of the population) to juveniles recruited into the population, is "trending toward the origin," the opposite direction from recovery. 4/2/10 Tr. 187-88. "There's no question that [the present abundance levels of delta smelt] are very low." 4/5/10 Tr. 232 (Hilborn).
11. FWS recently determined that delta smelt warranted uplisting from threatened to endangered, but that the action was currently precluded by higher priority listing actions. 4/7/10 Tr. 163; 75 Fed. Reg. 17,667 (Apr. 7, 2010). The direct mortality of delta smelt by entrainment at the CVP and SWP pumps, as well as the destruction and adverse modification of its habitat caused by water exports, were important factors in this determination. 75 Fed. Reg. at 17,671 ("The operation of State and Federal export facilities constitute a significant and ongoing threat to delta smelt through direct mortality by entrainment"). As a result of the "immediate and high magnitude threats" confronting the species, the delta smelt was assigned a listing priority number of 2.*fn2 Id. at 17,675.
12. Evidence submitted during trial indicates that, as of the dates of the March Spring Kodiak Trawl (March 8-11, 2010) and 20 mm surveys (March 15-18, 2010), delta smelt were collected in the northern and western portions of the Delta, not in the danger zones of the central or south Delta. SWC Exs. 918 & 919. Through March 28, 2010, the SWP had an expanded salvage of 16 delta smelt, and the CVP had an expanded salvage of 28 delta smelt. SWC Ex. 915.
13. Plaintiffs are correct that during the three years that restrictions on spring exports have been in place, the FMWT index has continued to trend downward.
4/7/10 Tr. 94:8-14. However, Mr. Grimaldo testified that improved conditions may not immediately translate into improved survival and population growth. 4/7/10 Tr. 120:9-25.
a. Comparison of CalSim and Dayflow Data
14. CalSim II ("CalSim") is a computer model developed jointly by DWR and Reclamation. The model simulates SWP and CVP operations and is the standard planning tool for evaluating project operations. 4/2/10 Tr. 101:24-102:6. The first version of the CalSim model was available in May 2002. It is continuously updated. 4/2/10 Tr. 102:7-13.
15. CalSim simulates SWP and CVP reservoir operations, project exports and water deliveries, flow through the Delta, and salinity requirements in the Delta, including the location of X2. 4/2/10 Tr. 102:14-20; BiOp at 207.
16. X2 is the location in the Delta where the salinity is two parts per thousand. It is measured as the distance upstream from the Golden Gate. 4/2/10 Tr. 102:21-24.
17. The CalSim model assumes 82 years of hydrology, 4/2/10 Tr. 101:23-102:3, 103:14-18, 161:2-6, provides the model with data regarding inflow to reservoirs and other information affecting the water supply, 4/2/10 Tr. 103:19-23. The model also assumes a level of development, which reflects water demand resulting from a particular urban population level, agricultural production, and wildlife refuge needs, 4/2/10 Tr. 104:1-7, as well as the existence and effect of environmental regulations and environmental programs, 4/2/10 Tr. 103:14-18. The assumptions used in the CalSim studies were developed by representatives from FWS, the National Oceanic and Atmospheric Administration ("NOAA"), Reclamation, CDFG, and DWR. 4/2/10 Tr. 105:8-12.
18. The CalSim model assists scientists in making planning decisions by allowing comparisons between studies based on differing assumptions. See 4/2/10 Tr. 102:25-103:6. According to Aaron Miller, P.E., an expert qualified to offer opinions on the subject of the formulation and application of CalSim, CalSim is not designed, or intended to be used, to compare CalSim study outputs to actual "historic" data or to outputs from different models, including the Dayflow model. 4/2/10 Tr. 95:7-14; DWR Ex. 511 at ¶8.
19. CalSim study 7.0 was developed as the baseline study for the 2008 OCAP Biological Assessment ("2008 OCAP BA" or "BA"). Study 7.0 represents existing conditions, and assumes a 2005 level of development and a full environmental water account ("EWA"). 4/2/10 Tr. 104:8-20; 123:21-24, 146:3-6; BiOp at 207. Study 7.1 is a near-future conditions study. It assumes a 2005 level of development and a limited EWA. 4/2/10 Tr. 104:8-23; 123:21-25; BiOp at 207-08. Study 8.0 is a future conditions study. It assumes a 2030 level of development and a limited EWA. 4/2/10 Tr. 104:8-25; 123:21-124:2; BiOp at 208.
20. CalSim study 6.0 was designed to look at the differences between the prior CalSim model used in the 2004 OCAP BA and the new model used in the 2008 OCAP BA. 4/2/10 Tr. 104:8-15, 157:11-18.
21. Study 6.1 is similar to 6.0, but did not include the EWA and used an older version of the X2 estimate. 4/2/10 Tr. 104:8-17. Study 6.1 was prepared at the request of Reclamation biologists to assess changes in water project operations during the pelagic organism decline ("POD") era. 4/2/10 Tr. 149:18-24, 150:16-151:17, 158:8-13. Reclamation biologists compared study 6.1 against the 7.0 and 8.0 studies on pages 13-10 though 13-17 of the 2008 OCAP BA. 4/2/10 Tr. 149:12-24; AR 011057-011064.
22. Mr. Miller testified that study 6.1 should not have been used for comparison because it was not comparable to the other studies. 4/2/10 Tr. 156:25-157:8. Study 6.1 used the Kimmerer Monismith equation to estimate X2 and it, as well as study 6.0, did not completely reflect the new enhancements in the CalSim model developed after the 2004 OCAP BA. 4/2/10 Tr. 157:10-18; SLDMWA Ex. 12 at 205-206.
23. The CalSim 9.0 series of studies represents climate change scenarios. Study 9.0 represents a future condition to serve as a basis of comparison of the effects of climate change to sea level rise, without the inclusion of (b)(2) or EWA. Study 9.1 represents a one-foot sea level rise, without the inclusion of (b)(2) and EWA. Studies 9.2 through 9.5 look at predicted changes in precipitation and temperature for the period 2010 to 2030, relative to conditions for the period 1971 to 2000. The 9.0 climate change scenarios were not intended to be directly compared to studies 7.0--8.0. 4/2/10 Tr. 105:1-5; BiOp at 208. Such a comparison is not valid because the studies make different assumptions regarding environmental programs. 4/2/10 Tr. 123:10-16.
24. In the BiOp, CalSim studies were compared to simulations of historic conditions generated using the Dayflow model. 4/2/10 Tr. 107:4-7, 142:6-9. Dayflow is a model that estimates historic outflow based on historic precipitation, inflow, and exports, and estimates of delta island diversions. Dayflow also provides an estimate for the location of X2. 4/2/10 Tr. 107:8-14.
25. In the BiOp, FWS purports to quantify adult entrainment by comparing OMR flows from CalSim studies to historic OMR flows during 1967-2007. BiOp at 212-13. The BiOp depicts these results in Tables E-5b and E-5c in the BiOp, which are labeled "difference from historic median value to CalSim II model median value" and "difference from historic median salvage to predicted salvage based on... CalSim II," respectively. Id. at 214. Tables E-5b and E-5c purport to quantify, as effects of the action, changes in OMR flows and entrainment using the Dayflow-generated historic data as the baseline and comparing that to CalSim study results. Based on these comparisons of CalSim data and Dayflow-generated historic data, the BiOp concludes, "adult entrainment is likely to be higher than it has been in the past under most operating scenarios, resulting in lower potential production of early life history stages in the spring in some years." BiOp at 213.
26. In another analysis in the BiOp, FWS purports to quantify the effects of the action on delta smelt habitat by comparing CalSim model projections of the location of X2 under the proposed action to the median location of X2 over the historical period 1967-2007, as simulated by Dayflow. BiOp at 235-36. Based on this comparison, the BiOp concludes "[t]he median X2 [locations] across the CalSim II modeled scenarios were 10-15 percent further upstream than actual historic X2 (Figure E-19)." Id. at 235. In reliance on these percent differences between CalSim-created data and historical data, the BiOp concludes "proposed action operations are likely to negatively affect the abundance of delta smelt." Id. at 236.
27. In the BiOp, FWS performed similar comparisons of CalSim data to Dayflow-simulated historic baseline data to quantify the effects of the action on larval and juvenile delta smelt. See, e.g., BiOp at 219 (examining effect of action on larval and juvenile entrainment and stating "[t]he analysis is based on comparison of historical (1967-2007) OMR and X2 to the proposed action's predictions of these variables provided in... [CalSim] studies 7.0, 7.1, 8.0, and 9.0-9.5").
28. Mr. Miller explained that outputs from a CalSim study should not be compared to outputs from the Dayflow model because the assumptions used in the two models are significantly different. 4/2/10 Tr. 107:18-23, 136:10-18.
a. The CalSim model assumes a constant level of development. In contrast, the Dayflow model incorporates a continuous change in the level of development because the Dayflow model is using historical information as input. When comparing models to determine the effect of project operations, the best scientific practice is to keep the assumed level of development constant. 4/2/10 Tr. 107:15-108:15.
b. A CalSim study also assumes a constant regulatory environment, whereas Dayflow uses a regulatory environment that has changed over time. This difference renders any comparison between CalSim and Dayflow outputs unreliable. 4/2/10 Tr. 108:16-109:23.
c. CalSim also operates on a monthly time step, whereas Dayflow operates on a daily time step. The two models also operate to different guidelines. The Dayflow model incorporates a conservative operation to avoid violating a regulation. In contrast, the CalSim model operates strictly to that regulation. 4/2/10 Tr. 107:23-108:3, 109:24-110:9. Operating conservatively results in higher modeled outflow. 4/2/10 Tr. 110:10-14.
d. The differences in the model assumptions and in the way the models operate, as described above, cannot be quantified to calibrate the models. CalSim does not model or simulate historical conditions, so it cannot be calibrated to history. 4/2/10 Tr. 121:18-122:6, 161:2-6. Calibration would be "very difficult, nearly impossible, to do without  developing a model designed to simulate historical conditions." 4/2/10 Tr. 110:15-111:1. The CalSim model cannot currently predict X2 for historic years because it would require a new model. 4/2/10 Tr. 122:7-16.
e. The Dayflow historic time window that FWS reported using in the BiOp was 1967 to 2007. CalSim studies model water years 1992 through 2003. The BiOp's comparison of CalSim-modeled data to Dayflow-modeled data resulted in comparing different sets of water years. Mr. Miller testified that the best scientific practice regarding years of comparison would have been to use consistent time windows. 4/2/10 Tr. 116:18-117:21; 142:13-15.
f. The artificial neural network ("ANN") and the Kimmerer Monismith equation ("KM equation") are two methods of estimating X2. 4/2/10 Tr. 111:2-16. The CalSim studies used ANN to estimate the position of X2, because ANN can be adapted to address sea level rise. 4/2/10 Tr. 111:19-25. The Dayflow model uses the KM equation to estimate X2. 4/2/10 Tr. 111:2-8; DWR Ex. 510 at Fig. 2; DWR Ex. 511 at ¶15. The KM equation was developed using historical data, making the KM equation invalid for a sea level rise study. 4/2/10 Tr. 111:19-25.
g. At locations less than 75 kilometers ("km") from the Golden Gate, the KM equation results in an X2 estimate greater than (or farther upstream than) the ANN estimate. In contrast, at locations greater than 75 km from the Golden Gate, the KM equation provides an estimate less than the ANN estimate. 4/2/10 Tr. 112:1-113:18, DWR Ex. 510 at Fig. 2.
29. Mr. Miller calculated the magnitude of error introduced into the BiOp by FWS's application of both the KM and the ANN methods of estimating X2. He replicated the 87 km value as the median estimate of X2 from CalSim study 7.0 using the ANN method, and, consistent with the BiOp, calculated the difference between the reported historic median of X2 [79 km] and the study 7.0 median [87 km] to be 10% [(87 km - 79 km)/79]. He then calculated the median X2 for the CalSim 7.0 study using the KM equation (instead of using ANN) to be 84 km (instead of 87 km). Finally, he identified the percent difference between the reported historic median estimate of X2 using the KM equation [79 km] and the CalSim study 7.0 median estimate of X2 using the KM equation [84 km] to be 6% [(84 km--79 km)/79 km]. 4/2/10 Tr. 114:6-25; DWR Ex. 511 at ¶¶ 14-16; BiOp at 235-36.
30. FWS did not calculate X2 using the KM equation for the CalSim studies, as did Mr. Miller. Instead, it undertook a direct comparison. DWR Ex. 511 at ¶15. The BiOp reported a 10% difference between the reported historic median X2 and the CalSim study 7.0 X2 median. Calculating the percent difference between the historical median X2 and study 7.0 median X2 using the KM equation resulted in only a 6% difference. From this, Mr. Miller concluded that 40% of the difference between X2 as estimated by study 7.0 and the historical X2 baseline reported in the BiOp is error attributed entirely to the use of the KM equation to calculate the historical baseline X2 and the ANN equation to calculate the CalSim study 7.0 baseline. 4/2/10 Tr. 114:6-25; DWR Ex 511 ¶ 15.
31. Mr. Miller testified that the differences in the KM equation and the ANN method of estimating X2 has an effect on the BiOp's analysis of habitat area, which in turn effects the BiOp's prediction of smelt abundance (as measured by the Summer Townet Survey Index). 4/2/10 Tr. 113:19-114:5; BiOp at 235--236, 266-269.
32. Mr. Miller explained that correcting for the differences between the use of the KM and ANN methods to estimate X2 does not correct for all the biases inherent in comparing CalSim data to "historic" data. It is unknown which portion of the remaining 60% of difference is attributable to the proposed action, and which portion is due to the other identified biases. 4/2/10 Tr. 115:1-8; DWR Ex. 511 at ¶16.
33. Mr. Miller testified that when using CalSim study 7.0 -- designed as a current conditions baseline -- instead of the "historical" baseline in the BiOp, and comparing study 7.0 to the near-future 7.1 study, X2 moved upstream 0.7 km. The percentage change in X2 from current to near-current conditions was 0.8%. Further, when comparing study 7.0 to study 8.0 (a 2030 level of development scenario), X2 moved upstream only 1.1 km, with a resultant percentage change in X2 of 1.2% from current to future conditions. 4/2/10 Tr. 128:18-129:11; DWR Ex. 511 at ¶20; BiOp at 235, 265. The 0.7 km change and the 1.1 km change, respectively, were vastly different from the approximately 8.7 km and 9.1 km changes shown in the BiOp (Figure E-19) using historical Dayflow as the baseline. BiOp at 265; DWR Ex. 511 at ¶7.
34. Using the equation identified in Figure E-20 in the BiOp, Mr. Miller calculated the reduction in suitable habitat consistent with the change in the position of X2. A comparison of CalSim study 7.0 with study 7.1 yielded a reduction in habitat area of 128 hectares, and a comparison of study 7.0 with study 8.0 yielded a reduction in habitat area of 289 hectares. 4/2/10 Tr. 129:12-130:5; DWR Ex. 511 at ¶20; BiOp at 266.
35. Plaintiffs assert that, prior to issuance of the BiOp, FWS was put on notice that comparing historical data to CalSim simulated data was an inappropriate and invalid methodology. 4/2/10 Tr. 133:15-134:11, 137:16-138:16, 138:21-139:14; SLDMWA Ex. 351 at 7; SLDMWA Ex. 261 at 5; SWC Ex. 933 at 3.
a. The 2008 OCAP BA did raise some cautionary notes:
CalSim II is intended to be used in a comparative mode. The results from a "proposed operation" scenario are compared to the results of a "base" scenario, to determine the incremental effects. The model should be used with caution to prescribe seasonal or to guide real-time operations, predict flows or water deliveries for any real-time operations. The results from a single simulation may not necessarily represent the exact operations for a specific month or year, but should reflect long-term trends.
b. DWR Deputy Director Jerry Johns, on October 24, 2008, submitted comments to FWS on the draft effects analysis, generally cautioning against the comparison of modeled data with actual data:
USFWS is using historic data for comparison to CalSim II simulations. Great caution should be taken when comparing actual data to modeled data. CalSim II modeling should be used in a comparative mode. In other words, it should be used to compare one set of model runs to another. For example, it would be appropriate to compare CalSim II modeling of one demand alternative to another to analyze the incremental effects.
AR 8671; see also AR 8668 (further explaining unreliability problems comparing historic and modeled data).
c. The State Water Contractors also cited a letter that they sent to FWS before the BiOp was completed. However, that letter only critiqued the comparison of simulated data to historical salvage data, and did not dispute with the comparison of CalSim-simulated to Dayflow-simulated historic data. 4/2/10 Tr. 133-34.
d. Mr. Miller acknowledged that, despite his heavy involvement in the modeling analysis underlying the BiOp, he did not present his current criticism of the use of the data to FWS during preparation of the BiOp. 4/2/10 Tr. 115-16.
36. FWS was not on notice of Mr. Miller's critiques regarding comparing simulated Calsim runs to simulated Dayflow runs, and was not put on notice by him that they were improperly using the specialized models. FWS did not have an opportunity to correct its modeling or address Plaintiffs' concerns.
37. The BiOp explains why FWS looked beyond CalSim. When CalSim was used to identify current Project operations, and these results were then compared to the results of a CalSim modeling run purportedly simulating past operations, the results "were nearly identical" despite significant operational changes in current operations as compared to past. BiOp at 204-05. The BiOp explains that "[t]he inaccuracies in CalSim [led FWS] to use actual data to develop an empirical baseline." Id. at 206. FWS "also developed historical time series data for hydrologic variables used in this effects analysis based on the Dayflow database... and OMR data obtained from USGS." Id.
38. Mr. Miller asserts that best scientific practice would preclude FWS from comparing CalSim output to historic data generated by Dayflow. However, Mr. Miller acknowledged that in the 2008 OCAP BA, DWR and Reclamation compared CalSim output to historic data, albeit for a different purpose, namely to show that the timing and magnitude of reservoir and export operations were similar to historic operations. 4/2/10 Tr. 119-20. Mr. Miller acknowledged that other modelers involved in preparing the BA expressed concerns about using only CalSim data, and that the BA itself questioned the use of that data alone, as CalSim simulations did not provide "an especially satisfactory representation of pre-POD water project operations." Id. at 150-51. The BA, prepared by DWR and Reclamation, states: "While we have not adopted an alternative statistical approach [to the use of CalSim model runs] in this biological assessment, we believe it would be a useful way to further assess changes in water project operations during the POD era and we recommend that [FWS] consider such an analysis as further refinement to this BA." Id. Other reputed scientists in the field agree with FWS and the BA that the CalSim-generated modeling studies did not "generate baselines with a high degree of reliability." Id. at 160. Neither Mr. Miller nor DWR offered any alternative to Dayflow to FWS to address that serious shortcoming during preparation of the BiOp. Id. at 160-61.
39. Mr. Miller acknowledged that, even if the CalSim comparison had been conducted in the manner he recommends, it would have confirmed FWS's conclusions that Project operations as proposed in the BA move X2 further upstream in the fall, reducing the amount of habitat for delta smelt and modifying the quality of critical habitat by shifting the low salinity zone away from higher-quality habitat and further into the central Delta. Id. at 130. Mr. Miller did not suggest that this revision would result in a de minimis shift of X2.
40. Mr. Miller presents substantive criticisms of the BiOp's CalSim runs. These specific concerns were not raised before the agency prior to the BiOp's issuance. Moreover, FWS expressed legitimate concerns, shared with other scientists, about the exclusive reliance on CalSim runs. Mr. Miller concedes that even if his recommended approach had been taken, the same fundamental result would have obtained: project operations shift the position of X2 upstream.*fn3
41. This highly technical dispute was not raised before the agency, and there were legitimate concerns about comparing Calsim modeling runs to other Calsim runs. This choice of competing methodologies is not sufficiently clear error to justify the court's intervention.
b. Treatment of "Other Stressors"
42. Plaintiffs raise a generic concern about how the BiOp treated the many other factors that are undeniably contributing to the decline of delta smelt including: (a) presence of aquatic macrophytes (submerged aquatic vegetation such as Egeria densa that may overwhelm delta smelt habitat); (b) predation; (c) introduction and propagation of invasive species, including inland silversides and the overbite clam that compete with the delta smelt; (d) presence of contaminants, such as pesticides and wastewater, in the Delta; and (e) presence of large blooms of blue-green algae toxic to the copepods eaten by delta smelt. BiOp at 182-86; 4/7/10 Tr. 148:17-19, 149:20-25.
43. Plaintiffs take particular issue with a statement in the very first paragraph of a section of the BiOp entitled "Effects of the Proposed Action."
The Status of the Species/Environmental Baseline section of this document described the multitude of factors that affect delta smelt population dynamics including predation, contaminants, introduced species, entrainment, habitat suitability, food supply, aquatic macrophytes, and microcystis. The extent to which these factors adversely affect delta smelt is related to hydrodynamic conditions in the Delta, which in turn are controlled to a large extent by CVP and SWP operations. Other sources of water diversion (NBA, CCWD, local agricultural diversions, power plants) adversely affect delta smelt largely through entrainment (see following discussion), but when taken together do not control hydrodynamic conditions throughout the Delta to any degree that approaches the influence of the Banks and Jones export facilities. So while many of the other stressors that have been identified as adversely affecting delta smelt were not caused by CVP and SWP operations, the likelihood and extent to which they adversely affect delta smelt is highly influenced by how the CVP/SWP are operated in the context of annual and seasonal hydrologic conditions. While research indicates that there is no single primary driver of delta smelt population dynamics, hydrodynamic conditions driven or influenced by CVP/SWP operations in turn influence the dynamics of delta smelt interaction with, these other stressors (Bennett and Moyle 1996).
BiOp at 202 (emphasis added).
44. The BiOp concludes that "the CVP and SWP have played an indirect role in the delta smelt's decline by creating an altered environment in the Delta that has fostered the establishment of non-indigenous species and that exacerbates these and other stressors that are adversely impacting delta smelt." BiOp at 203; 4/7/10 Tr. 152:5-12. Ms. Goude further testified that it is not possible to quantify the level of effects of those other factors. 4/7/10 Tr. 150:1-3.
45. When asked by the Court to identify any information in the record that supports the BiOp's conclusion that project operations exacerbate the effect of other stressors, Dr. Thomas Quinn, an expert appointed under Federal Rule of Evidence 706, concluded that "there does not appear to be evidence in the record demonstrating that project operations exacerbate the effect/impact of other stressors." Doc. 633, Order Transmitting Responses from 706 Experts, Ex. A, at 20. Ms. Goude testified that she disagreed with this conclusion, but could not identify any evidence from the See record to support her assertion. 4/7/10 Tr. 201:22-203:9.
46. Dr. Andre Punt, another court-appointed expert, further explained the BiOp's notion that indirect effects of the Projects may contribute to effects such as high water toxicity, suppression of phytoplankton, increase of overbite clams, and increase in encounters with unscreened agricultural diversions in the Delta are plausible hypotheses, but that "there are no direct data available to test them." Doc. 633 at 21.
47. In contrast to the BiOp's general statements assigning the blame for at least some, unquantified portion of the negative effects cause by these "other stressors" to the projects, elsewhere, the BiOp acknowledges that there is "no single primary driver of delta smelt population dynamics," id. at 202, but rather that there are "multiple factors" and that "not all are directly influenced by operations of the CVP/SWP." Id. at 328. "Other stressors" are discussed in detail throughout the BiOp. See, e.g., id. at 182-88, 198, 201- 2. Specifically, FWS considered the effects of "predation, contaminants, introduced species..., habitat suitability, food supply, aquatic macrophytes, and microcystis." Id. at 202, 277. The BiOp expressly recognizes that the long-term decline of the species "was very strongly affected by ecosystem changes caused by non-indigenous species invasions and other factors...." Id. at 189.
48. Although the BiOp acknowledges that "not all" of the multiple factors negatively impacting the species "are directly influenced" by Project operations, the general assertion in the BiOp that other stressors are the result of (or at least exacerbated by) Project operations is not supported by the record. This error compounds the agency's failure to address alternative approaches to avoiding jeopardy, including whether other stressors can be mitigated or eliminated, which NEPA requires.
(3) Challenges to Component 2 (Action 3)
49. Component 2 (Protection of Larval and Juvenile Delta Smelt) requires OMR flows to remain between -1,250 and -5,000 cfs beginning when Component 1 is completed, when Delta water temperatures reach 12° Celsius, or when a spent female smelt is detected in trawls or at salvage facilities. Id. at 282, 357-358. Component 2 remains in place until June 30 or when Clifton Court Forebay water temperature reaches 25° Celsius, whichever first occurs. Id. at 282, 368.
50. The objective of Component 2 (which corresponds to Action 3 in Attachment B of the BiOp), is to "improve flow conditions in the Central and South Delta so that larval and juvenile delta smelt can successfully rear in the Central Delta and move downstream when appropriate." BiOp 282.
51. The most recent smelt working group recommendation for the week of April 12, 2010 recommends OMR flows no more negative than -5,000 cfs because the "risk to larval delta smelt was low, given that no salvage of larvae has occurred so far this year and the latest survey data suggest that the greatest densities of delta smelt are in the Sacramento River and downstream of the confluence, and, therefore, outside the influence of the pumps."*fn4
a. Use of Raw Salvage to Justify the Quantitative Flow Restrictions
52. The BiOp quantitatively analyzed the effects of pumping at the Banks and Jones pumping plants. 4/6/10 Tr. 19:1-3; BiOp at 208-209.
53. The results of that quantitative analysis, which compared OMR flows with gross salvage numbers, are described in Figures B-13 and B-14 of the BiOp. BiOp at 348, 350. These figures were presented as part of a three and-a-half page section of the BiOp entitled "Justification for Flow Prescriptions in Action 1." BiOp at 347-51. It also appears that this analysis was relied upon to set the calendar-based flow prescription in Component 2 (Action 3), as no other basis for the -5,000 cfs ceiling is presented. Because this portion of the BiOp is critical to the present challenge, it is reproduced here in its entirety:
Justification for Flow Prescriptions in Action 1
Understanding the relationship between OMR flows and delta smelt salvage allows a determination of what flows will result in salvage. The OMR-Salvage analysis herein was initiated using the relationship between December to March OMR flow and salvage provided by P. Smith and provided as Figure B-13, below. Visual review of the relationship expressed in Figure B-13 indicates what appears to be a "break" in the dataset at approximately -5,000 OMR; however, the curvilinear fit to the data suggest that the break is not real and that the slope of the curve had already begun to increase by the time that OMR flows reached -5,000 cfs.
Further, a nonlinear regression was performed on the dataset, and the resulting pseudo-R2 value was 0.44-suggesting that although the curvilinear fit is a reasonable description of the data, other functional relationships also may be appropriate for describing the data. Fitting a different function to the data could also determine the location where salvage increased, i.e. identify the "break point" in the relationship between salvage and OMR flows. Consequently, an analysis was performed to determine if the apparent break at -5,000 cfs OMR was real. A piecewise polynomial regression, sometimes referred to as a multiphase model, was used to establish the change (break) point in the dataset.
A piecewise polynomial regression analysis with a linear-linear fit was performed using data from 1985 to 2006. The linear-linear fit was selected because it was the analysis that required the fewest parameters to be estimated relative to the amount of variation in the salvage data. Piecewise polynomial regressions were performed using Number Cruncher Statistical Systems ((c) Hintz, J., NCSS and PASS, Number Cruncher Statistical Systems, Kaysville UT).
The piecewise polynomial regression analysis resulted in a change point of -1162, i.e. at -1162 cfs OMR, the slope changed from 0 to positive (Figure B-14). These results indicate that there is a relatively constant amount of salvage at all flows more positive than -1162 cfs but that at flows more negative than -1162, salvage increases. The pseudo-R2 value was 0.42, a value similar to that obtained by P. Smith in the original analysis.
To verify that there was no natural break at any other point, the analysis was performed using a linear-linear-linear fit (fitting two change points). The linear-linear-linear fit resulted in two change points, -1,500 cfs OMR and -2,930 cfs OMR. The -1,500 cfs value is again the location in the dataset at which the slope changes from 0 to positive. The pseudo-R2 value is 0.42 indicating that this relationship is not a better description of the data. Because of the additional parameters estimated for the model, it was determined that the linear-linear-linear fit was not the best function to fit the data, and it was rejected. No formal AIC analysis was performed because of the obvious outcome.
A major assumption of this analysis is that as the population of Delta smelt declined, the number of fish at risk of entrainment remained constant. If the number of fish in the vicinity of the pumps declined, fewer fish would be entrained and more negative OMR flows would result in lower salvage. This situation would result in an overestimate, i.e. the change point would be more positive. In fact, if the residuals are examined for the relationship in Figure B-13 above, the salvage for the POD years 2002, 2004, 2005, and 2006 are all below the line. 2003 is above the line although the line is not extended to the points at the top of the figure, and these data points occur when the curve becomes almost vertical. The negative residuals could ...