WCT Multi-state Assessment February 1 0, 2003 



Oregon and Washington, Puget Sound in Washington, the Klamath and Goose Lake Basins in 

 southern Oregon and the Bear Lake Basin in southeastern Idaho (PNW Reach File, Gladstone, 

 Oregon: Stream Net, August 2002. http://www.streamnet.org/pnwr/pnwrhome.html). River 

 reach files for Montana east of the Continental Divide were obtained from Montana Fish 

 Wildlife and Parks (Streams. Helena, MT: Montana Fish Wildlife and Parks, August 2002 and 

 are available at http://fwp.state.mt.us/insidefwp/fwplibrary/gis/). This LLID hydrography layer 

 routes stream segments by uniquely identifying each stream. Delineating lower and upper 

 segment boundaries as distances above each stream's mouth identified each stream segment 

 occupied by WCT. All known fish barriers were located as points, also using distance upstream 

 from a stream's mouth. 



For a few LLID streams we found that the streams were routed in reverse, from the headwaters 

 to the mouth. These errors became apparent when we computed lengths of stream segments and 

 a negative length resulted. We used the absolute values of length to correct for this problem and 

 neither the computed lengths nor the map locations were affected by this problem. 



Scale issues 



Using a standard 1 : 100,000 base-layer allowed for consistent summaries among states and other 

 entities. However, summaries based on this scale will underestimate "true" field lengths of 

 stream habitats due to scale-based error. There are several potential sources of bias associated 

 with using 1:100,000 scale LLID hydrography. First, map-derived stream lengths under-estimate 

 actual stream lengths. Finnan and Jacobs (2002) found that while hip-chained measurements of 

 Oregon coastal streams were significantly correlated to stream lengths computed using 

 MapTech® Terrain Navigator software and 1 :24,000-scale maps, map lengths needed to be 

 multiplied by about 1.14 to estimate measured stream lengths. 



Secondly, there are scale-differences between 1:100,000 and l:24,000-scale hydrography. To 

 evaluate the magnitude of these scale-differences, we compared lengths of 30 streams from three 

 different 4''' code HUC's (10 per FTUC) and found that lengths of streams derived from 

 1 : 100,000-scale hydrography were only about 1% shorter than estimates of that same stream 

 using 1 :24,000-scale hydrography (Appendix C). Thirdly, there was some variability across the 

 study area in designating which streams were included within the LLID hydrography layer 

 (Appendix C). All named streams were included in the LLID layer for Idaho and Montana, 

 while unnamed streams were not included. All named and most unnamed streams were included 

 in the LLID hydrography for streams within Washington and Oregon. Unnamed streams were 

 also included for those watersheds that spanned the border between Idaho and Washington. To 

 evaluate potential differences between LLID information that included and excluded unnamed 

 streams we compared stream densities between a HUC where unnamed streams were included 

 (Priest) and one where unnamed streams were not included (Upper Coeur d'Alene). We found 

 that inclusion of unnamed streams resulted in 35% higher stream densities (1.86 miles versus 

 1.20 miles per 1,000 acres; Appendix C). Therefore, stream lengths computed for basins located 

 in Washington and Oregon will be higher relative to the rest of the study area, but these two 

 states contain less of the historical and current range of WCT than Montana and Idaho (Figure 

 1). We assume that comparisons among proportions of habitats occupied by various classes 

 should be relatively unbiased within HUC's since these proportions should have consistent 

 biases due to the strong correlation between map length and field-measured length (Firman and 



Page - 3 



