Marks, D., Olis, M. and Wyman, K., 2004. An Evaluation of the Skagit Basin Coho SSHIAP-LFA Database and an Assessment of the DNR Last Fish Water Type Map. Skagit River System Cooperative, La Conner, WA. pp. 7.

Recent changes to Forest Practices Regulations (Forests and Fish Report 1997) spawned out of concerns of inadequate protections to fish resources and pending Endangered Species Act (ESA) listings of local salmon stocks have led to the development of a potential predictive fish habitat model for the purpose of identifying and classifying all fish-bearing and end of fish habitat waters in western Washington. The “model”, a GIS-based logistic regression model, was to be developed using existing data (10m digital elevation network, precipitation, gradient, and basin size) coupled with survey-verified end of fish distribution points scattered throughout the western portion of the State. The model was given a performance target for accuracy of 95% with± 5% likelihood –“that the line demarcating fish and non-fish habitat waters will be drawn so as to be equally likely to be over and under inclusive…”. In preparation for an evaluation of the new fish habitat model, the Skagit River System Cooperative (SRSC) Forest and Fish program set out to determine the accuracy of the existing coho salmon upper extent distribution portion of the Salmon and Steelhead Habitat Inventory and Assessment Program (SSHIAP) database for the Skagit basin and compare those results with the preliminary fish habitat model output maps.
The SSHIAP database is a spatial data system that characterizes salmonid habitat conditions and distribution of the salmonid stocks in Washington. SSHIAP is a co-management-based dataset operated jointly by the Washington Department of Fish and Wildlife (WDFW) and the Northwest Indian Fisheries Commission (NWIFC) that covers WRIAs (Water Resource Inventory Areas) 1-62. In WRIAs 3 and 4, the salmonid distribution portion of the SSHIAP database was primarily established through the efforts of a technical advisory group (TAG) that convened to provide input to the habitat limiting factors analysis (LFA) that affect the natural production of salmonids. Salmonid distribution is arguably one of the most important outcomes of the LFA.
The goal of our study was to verify and determine the accuracy (ground truth) of the LFA coho distribution database and to further use the verified data as a base to expand and evaluate the preliminary fish habitat model maps. Our first concern with the proposed fish habitat model was to make certain that the output maps would be inclusive of known fish distribution and designate it as such (i.e., type F), in addition to designating potential or likely “fish habitat” that may or may not have fish present during our “snap-shot” sampling period. Protection of fish and fish habitat is a key goal of the Forests and Fish Report.
Coho salmon LFA distribution points were used in this investigation for several reasons. It is a large data set in the Skagit basin with 519 points (562 points if you include the Baker River system; not included in our study). Coho typically utilize the furthest upstream reaches compared to the other salmon species; therefore, coho protection should be inclusive for all salmon species and provide a “conservative” standard for what should be called fish habitat waters. We also have a better understanding of coho distribution than we do for resident fish species (e.g., cutthroat and rainbow trout, native char, dace