I.    Forecasts of Smolt Outmigration Timing in Real Time

Since 1994, this project has been forecasting the timing of smolt outmigration and posting the predictions in real time on the Internet for fish managers and the public.  The forecasts include two forms of information:  (a) Percentage of run to date at a location for a specific fish stock, (b) Date of passage for specific percentiles of the run at a location for a specific fish stock.  This information can be used by fish managers to time spill augmentation in order to best benefit fish and conserve water resources.  For example, at some hydroprojects (e.g., Rocky Reach and Rock Island under the HCP), spill must occur during the middle 95% of the run; forecasts of run timing are therefore used to time the onset and completion of a spill program. Under the 2008 BiOp, spill timing was prespecified within the FCRPS (i.e., RPA 29); however, under the FCRPS Adaptive Management Implementation Plan of 2009, spill programs will be reviewed annually to determine the best operation for the fish.

In 1994, these forecasts began simply with spring Chinook salmon smolts at Lower Granite Dam.  Today the forecasts include over 30 stocks of fish: 

1. Timing tracked beginning at Lower Granite Dam:
   1. PIT-tagged spring/summer Chinook stocks in the Snake River.  The individual stocks include the following:
      1. Big Creek
      2. Catherine Creek
      3. Imnaha River
      4. Imnaha Trap
      5. Johnson Creek Trap
      6. Lake Creek
      7. Lemhi River
      8. Lookingglass Creek
      9. Lostine River
      10. Meadow Creek
      11. Minam River
      12. Valley Creek 
   2. Snake River PIT-tagged, run-at-large of wild spring/summer Chinook.
   3. Clearwater River subyearling Chinook salmon
   4. Snake River subyearling Chinook salmon.
   5. Snake River composite of hatchery-reared sockeye salmon from Redfish Lake Creek trap
   6. Snake River PIT-tagged, run-at-large of wild steelhead
2. Timing tracked beginning at McNary Dam:
   1. PIT-tagged run-at-large of wild Upper Columbia River fall Chinook salmon
   2. PIT-tagged run-at-large of wild Snake River fall Chinook salmon
   3. PIT-tagged run-at-large of wild Snake River spring/summer Chinook salmon
   4. PIT-tagged run-at-large of wild Upper Columbia River steelhead
   5. PIT-tagged run-at-large of wild Snake River steelhead
   6. PIT-tagged run-at-large of wild combined Upper Columbia River and Snake River steelhead
   7. PIT-tagged run-at-large of wild Snake River sockeye
3. Timing tracked beginning at Rock Island Dam in the Mid-Columbia based on FPC index counts:
   1. Fall Chinook salmon
   2. Spring/summer Chinook salmon
   3. Sockeye salmon
   4. Coho salmon
   5. Steelhead
4. Timing tracked beginning at Lower Granite Dam based on FPC index counts:
   1. Fall Chinook salmon
   2. Spring/summer Chinook salmon
   3. Sockeye salmon
   4. Coho salmon
   5. Steelhead
5. Timing tracked beginning at McNary Dam based on FPC index counts:
   1. Fall Chinook salmon
   2. Spring/summer Chinook salmon
   3. Sockeye salmon
   4. Coho salmon
   5. Steelhead
6. Timing tracked beginning at John Day Dam based on FPC index counts:
   1. Fall Chinook salmon
   2. Spring/summer Chinook salmon
   3. Sockeye salmon
   4. Coho salmon
   5. Steelhead
7. Timing tracked beginning at Bonneville Dam based on FPC index counts:
   1. Fall Chinook salmon
   2. Spring/summer Chinook salmon
   3. Sockeye salmon
   4. Coho salmon
   5. Steelhead

For distinct PIT-tagged subpopulations, run timing is tracked downriver as the fish pass through hydroprojects with PIT-tag detectors. 

In the future, hatchery and wild PIT-tag runs from the upper Columbia River will be available for forecasting because of the installation of a PIT-tag detector at Rocky Reach Dam in the mid-Columbia in 2010.  Program RealTime needs a minimum of three historical years of outmigration data to properly forecast inseason run timing.  Forecasting should begin in 2013 for these fish stocks.

Forecasts of run timing are posted on the UW DART website (Inseason Forecasts, http://www.cbr.washington.edu/crisprt/index.html) in a user-friendly format for easy public dissemination (Figure 1). 

Figure 1.  Index of Inseason Predictions webpage on UW DART.

The forecasts are presented in various tabular and graphical formats in an interactive web environment.  Tables of anticipated dates to specific percentiles of the run are posted by stock and location (Table 1).  Graphical summaries of inseason run-timing predictions are also provided (Figure 2).

a. Forecast of percent passage by date.  Graphs include the daily history of predictions, along with the current day's prediction of run timing to date.

b. Comparison of current season’s run-timing predictions with historical cumulative arrival distributions.

Figure 2.  Graphical summaries of inseason run-timing predictions for select composite wild yearling Chinook salmon at Lower Granite Dam as of May 15, 2010. 

Predictive performance is generally good.  For example, in 2009, the last year with complete information, of the 25 stock-by-location combinations where FPC indices were monitored, 5 had mean absolute deviances (MADs) for inseason predictions between 5-10%, 5 had MADs between 3-5%, and 15 had MADs below 3%.  Each year, a detailed report on forecast performance is preferred for BPA and submitted as part of this project’s deliverables (e.g., Townsend et al. 2010).

II.   Annual Summaries of Ambient and Fish Performance Measures

The Status & Trends Overview webpage (http://www.cbr.washington.edu/trends/index.php) on the UW DART website is annually updated with summaries of ambient and fish performance measures.  The interactive website allows investigators to assess time trends in performance measures and cross-reference performance measures by species or location.  Users of the website can query the relational database from multiple dimensions, including performance measure, species, location, and hydroproject (Figure 3). 

Figure 3.  Status & Trends Overview with pull-down menu of performance measures.

This project uses the statistical methods developed, in part, by Project 1989-107-00, Statistical Support for Salmonid Survival Studies, to perform state-of-the-art statistical analyses of tagging and other monitoring data.  Annual updates of performance measures are performed by this work element for public dissemination on the DART website.

Examples of the performance measures include the following:

 A. Adult passage performance counts (example, Figure 4)  

1. Bonneville, 1938 →
2. Ice Harbor, 1962 →
3. John Day, 1969 → 
4. Little Goose, 1970 → 
5. Lower Monumental, 1969 →
6. Lower Granite, 1975 →
7. McNary, 1956 →
8. Priest Rapids, 1961 → 
9. Prosser, 1983 →
10. Rock Island, 1977 →
11. Roza, 1940 →
12. Rocky Reach, 1977 →
13. The Dalles, 1960 →
14. Tumwater, 1999 →
15. Wells, 1977 → 

Figure 4.  Example:  Adult passage of steelhead at Wells Dam, 1977-2009.

B. Compliance with flow targets (example, Figure 5)

1. Bonneville, 1949 →
2. Lower Granite, 1975 →
3. McNary, 1956 →
4. Priest Rapids, 1961 →

Figure 5.   Example:  Days out of compliance with flow targets at Lower Granite Dam during summer, 1975-2009.

C. Compliance with temperature targets (example, Figure 6)

1.  Bonneville, 1995 →
2.  Ice Harbor, 1995 →
3.  John Day, 1995 →
4.  Little Goose, 1995 →
5.  Lower Granite, 1995 →
6.  Lower Monumental, 1995 →
7.  McNary, 1995 →
8.  The Dalles, 1995 →

 Figure 6.  Example:  Days out of compliance with temperature target of 20^oC at Bonneville Dam forebay. 

D. Compliance with total dissolved gas targets (example, Figure 7)

1.  Bonneville, 1995  →
2.  Lower Granite, 1995  →
3.  McNary, 1995  →
4.  Priest Rapids, 1995  →
5.  Ice Harbor, 1995  →
6.  John Day, 1995  →
7.  Little Goose, 1995  →
8.  Lower Monumental, 1995  →
9.  The Dalles, 1995  →

 Figure 7.  Example: Days out of compliance with total dissolved gas targets at Bonneville Dam forebay, 1995-2009.

E.  Smolt survival, travel time, and detection probability (example, Figures 8 and 9)

1. Clearwater Trap, 1993 →
2. Dworshak, NFH, 1997 →
3. Entiat NFH, 1993 →
4. Imnaha Trap, 1994 →
5. Leavenworth NFH, 1995 →
6. Lookingglass Hatchery, 1995 →
7. Lyons Ferry Hatchery, 2004 →
8. McCall Hatchery (Knox Bridge), 1997 →
9. Priest Rapids Hatchery, 1997 →
10. Rapid River Hatchery, 1993 →
11. Ringold Hatchery, 1997 →
12. Sawtooth Trap, 1993 →
13. Wells Hatchery, 1996 →
14. Winthrop NFH, 1996 →  

 Figure 8.  Example:  Smolt survival of spring Chinook salmon, 1995-2009, from Lookingglass Hatchery to Lower Granite Dam.

 Figure 9.  Example:  Detection probabilities for wild Chinook salmon from Clearwater Trap at Little Goose Dam, 1993-2008.  

F.  ROSTER life-cycle analyses (examples, Figures 10–12)

1. Adult survival by release year, 1999-2006
2. Adult survival by return year, 2000-2009
3. D (system wide, LGR, LGS) by release year, 1997-2006
4. Ocean survival by release year, 1999-2006
5. SAR by release year, 1996-2006
6. Smolt survival by release year, 1998-2006
7. T/I (system wide, LGR, LGS) by release year, 1997-2006  

Analyses are performed on hatchery and wild spring Chinook salmon, hatchery summer Chinook salmon, and hatchery steelhead PIT-tag releases above Lower Granite Dam.  These near life-cycle analyses use both juvenile and adult PIT-tag detections in a multistate mark-recapture model (Buchanan and Skalski 2010).

Figure 10.  Example:  Differential mortality (D) associated wtih Lower Granite Dam for Snake River hatchery spring Chinook salmon from ROSTER analyses.  This is the ratio of post-Bonneville survival for fish transported at Lower Granite Dam relative to post-Bonneville survival for fish not transported at Lower Granite Dam. 

 Figure 11.  Example:  Ocean survival (i.e., Bonneville-to-Bonneville survival) for Clearwater River hatchery spring Chinook salmon, 1999-2006.

Figure 12.  Example:  Transport/inriver ratio for Lower Granite Dam for Clearwater River hatchery spring Chinook salmon, 1998-2006. 

G.  Smolt-to-adult ratios (SARs) (example, Figure 13)

CWT data are used to estimate smolt-to-adult ratios (SARs) for over 350 different hatchery salmonid stocks based on year of release.

Figure 13.  Illustration of SARs by year of release for Priest Rapids hatchery fall Chinook salmon, 1975-2005.

H. Harvest

Harvest data for a total of 647 different fish stocks, locations, and fishery types were compiled (Figure 14).

Figure 14.  Example:  Recreational harvest estimates of steelhead, Snake River, Sections 1 and 2, starting at the WA/ID line and ending at the confluence of the Clearwater River.  Data from Bill Horton, IDFG. 

 I. Adult Escapement

Escapement data were summarized at 58 locations and compared to NMFS interim recovery goals (Figure 15).  The method of estimation is coordinated with NMFS to be consistent with BiOp recovery goals.

Figure 15.  Adult escapement of wild and hatchery summer steelhead in the Okanogan River.  Data from the Interior Columbia Technical Review Team (ICTRT) at NWFSC.

In 2005, BPA requested UW staff to stop assembling and disseminating escapement and harvest data because of potential agency sensitivities.

III.  Perform value-added analyses of historical tagging data.

PIT-tag data in PTAGIS and CWT data in RMIS have become important sources of RME data beyond the initial intents of the tagging studies.  These data provide the opportunity to:

1. Examine long-term relationships between fish performance and hydrographic conditions, hydrosystem operations, and mitigation actions. 
2. Evaluate long-term trends in fish performance. 
3. Extract life-history information from the joint juvenile and adult detections.
4. Evaluate past study performance in order to improve the design of future tagging studies.

The use of PIT-tag detections inseason to forecast smolt outmigration timing is another value-added analysis beyond the original intent of the tagging studies (see Section 1 of Major Accomplishments). 

This project performs these value-added analyses to extract fish performance measures that could not be done without the use of meta-analysis.  An example is the ROSTER (River-Ocean Survival and Transportation Effects Routine) analyses that uses joint juvenile and adult detections over multiple years for a release group of PIT-tagged fish to estimate transport effects, differential mortality (D), ocean survival (i.e., Bonneville-to-Bonneville survival), and the effects of transportation on age composition and upriver survival of adults (Buchanan and Skalski 2007).  Another example is Ryding and Skalski (1999) who used age-at-return information from CWT data to extract early ocean survival and relate it to oceanographic effects.  These types of meta-analyses are performed not only to extract additional information from historical tagging studies but to also assess the feasibility of such analyses in order to obtain the most cost-effective information from the studies funded by the FWP.  The results of these analyses are also used as preliminary information to help design future tagging studies. 

BPA requests this project to perform value-added analyses to assess critical uncertainties related to the management of programs and operation of the hydrosystem.  These analyses are also performed to provide BPA with independent information on required sample sizes for proposed work and independent assessment of fish performance and/or hydrosystem operations as part of the peer-review process.  PIT-tag data are analyzed annually to estimate detection probabilities at hydroprojects to help design future tag-release studies and the results are posted on the UW DART website. 

Of the 18 peer-reviewed papers produced by this project, 7 are associated with value-added analyses.  Of the 24 BPA technical reports, 18 are associated with real-time forecasting of smolt run timing.  Today, most of the value-added analyses are reported on the CBR Status & Trends Overview webpage for public dissemination.

IV. Develop Improved Statistical Methods for Monitoring and Evaluation

An important aspect of this project is assuring the monitoring methods used in the Columbia Basin are the best available science.  Project 1989-107-00, Statistical Support for Salmonid Survival Studies, is concerned with the proper design and analysis of tagging studies.  This project focuses on non-tagging aspects of monitoring in the Snake/Columbia River Basin.

Hydroacoustic monitoring of fish passage at hydroprojects in the Columbia Basin started, as elsewhere, as descriptive studies.  Soon quantitative requirements were placed on these studies because hydro managers wanted to assess the effects of hydro operations and mitigation activities at dams (e.g., sluiceways, top spill, screens, weirs, etc.) on fish passage performance.  This project ushered in the use of finite sampling techniques (Hansen et al. 1953, Cochran 1977) in conjunction with hydroacoustic sampling.  Project-wide sampling designs using as many as 50-60 hydrophones were developed from statistical first principles.  The sampling designs often consisted of multistage, stratified designs requiring derivation of both point and variance estimates for a host of performance measures (e.g., fish guidance efficiency, fish passage efficiency, spillway effectiveness, etc.)  Between 1997 and 2007, over 25 statistical sampling plans were developed for hydroacoustic monitoring of fish passage at 14 major hydroprojects in the Snake/Columbia River Basin.  These efforts resulted in 3 peer-reviewed papers (Skalski et al. 1992, Johnson et al. 1994, Skalski et al. 1993) on applications of finite sampling to hydroacoustic monitoring and 16 co-authored technical reports. 

Another aspect of this project has been to consider how to most appropriately sample the environment for different monitoring objectives.  This project has reviewed the use of panel designs/rotational designs (Skalski et al. 1990, Skalski 2010) for long-term monitoring, spatially extensive designs for regional resource monitoring (Skalski 1994), longitudinal sampling for damage assessment (Skalski 1995), and time series analysis (Ryding et al. 2007).  In 2010, this project developed a rotational panel design to monitor trends in smolt abundance in the estuary.

In 2005, the BPA requested guidance on how best to use data on habitat restoration projects funded by the FWP to assess potential benefits to salmon recovery.  To this end, we developed the most extensive collection of habitat, fish performance, and ambient data ever assembled for one location.  We chose the Columbia Cascades Province as a case study (Figure 16).  The database included 1,389 habitat restoration projects over time and by activity type, plus harvest, escapement, adult dam passage counts, hatchery smolt production, hydrographic data on stream flows, as well as spill and flow information at hydroprojects.  The spatial-temporal information was put into a GIS database for interactive display and analysis (Figure 16).  Important findings included the following:  1) benefits of the FWP habitat restoration projects cannot be evaluated in a vacuum but much be considered in context with all restoration activities; 2) spatial scale is critical in assessing restoration benefits—too large or too small an area can miss the recovery response; 3) accurate information on both harvest and escapement is essential in assessing the benefits of habitat restoration activities on salmonid recovery.

 a. Location of monitoring data and habitat restoration projects in the Columbia Cascades Province.

b. Screen capture illustrating interactive data display.

Figure 16.  Screen captures of a prototype interactive program designed to help assess restoration activities in the Columbia Cascades Province.

Finally, this project has assisted the Action Agencies in evaluating the proposed recovery criteria in the various versions of the BiOp.  In the 2000 version of the BiOp, recovery was going to be evaluated in 2005 and 2008, in part, by simple t-tests of pre- and post-2000 means of various performance measures.  This project, in collaboration with NOAA staff, evaluated the statistical power and performance of such tests of recovery.  The evaluation found the proposed tests to have very low statistical power (Skalski et al. 2001b), and alternative multivariate tests of recovery were recommended with much better statistical performance (Skalski et al. 2001a).  These evaluations contributed to NOAA’s revisions of recovery criteria in the BiOp.  Knowledge acquired by UW Project 1989-107-00, Statistical Support for Salmonid Survival Studies, is currently being used in the design of tagging studies needed to assess compliance with survival standards in the new FCRPS 2008 BiOp. 

V. Provide Statistical Consulting on Monitoring and Evaluation Projects in the Columbia Basin

Each year, research staff of this project are asked to assist fisheries investigators in the design and interpretation of monitoring and evaluation studies.  These statistical consulting activities are used to help improve the precision and cost-effectiveness of studies supported by the FWP.  These efforts include attending technical support meetings in the community, direct person-to-person assistance, collaboration on monitoring and evaluation studies, and peer review of study plans and reports.

Since 2005, we have provided statistical support to agencies, tribal nations, and organizations including the following:  NOAA, ACOE, EPA, USFWS, WDFW, IDFG, Washington Department of Transportation, Confederated Tribes of the Colville Reservation, and PNAMP. 

As part of our annual reporting to BPA, we summarize our levels of consulting support to the agencies, tribal nations, and organizations.  For the years 2005-2009, the levels of consulting are as follows:

Direct consulting is defined as consulting services provided by funds from this project.  Indirect consulting is defined as applying technologies and skills developed by this project to agencies that compensate for our effort in separate contracts. 

During the period 2005-2009, the average annual effort was 337 man-hours per year.  Over the same time period, these efforts results in 20 co-authored technical reports and 13 presentations on RME issues in the Columbia Basin.  These products do not include the peer-reviewed papers and BPA technical reports produced specifically by this project.