Scientific Investigations Report 2006–5101–D
U.S. GEOLOGICAL SURVEY
Scientific Investigations Report 2006–5101–D
Back to Table of Contents
Methods used during this study followed guidelines established for the National EUSE Program (Cuffney and others, 2000; McMahon and Cuffney, 2000; Tate and others, 2005). Site selection started with an initial 206 “candidate” streams and was eventually refined to 28 “final” streams. Data collection began in November 2003 and continued through March 2005, with the 28 final streams sampled for physical habitat, water chemistry, and aquatic biology. The final streams, and their associated watersheds, represented varying degrees of urbanization, yet shared similar geospatial and socioeconomic characteristics. During the study, 96 water-quality samples were collected and stream habitat was evaluated at each site. In addition, the biological community (algae, benthic macroinvertebrates, and fish) in each stream was sampled during low-flow conditions. Stream stage and water temperature were monitored in each stream from March through November 2004. Semipermeable membrane devices were installed at each location for about one month.
Streams selected for this study represented a full gradient of urbanization, and met the predefined geospatial and ecological characteristics developed for the National EUSE Program defined below.
To limit natural variability between watersheds, certain geospatial and ecological constraints were required. The two primary site selection constraints were “watershed size” and “percent coverage in the Willamette Valley-Level III Ecoregion (Omernik, 1987)”. Watersheds could cover no less than 10 km2 and not exceed 130 km2. Additionally, all watersheds included a minimum of 20 percent of the Willamette Valley ecoregion. With these guidelines, site selection and watershed processing proceeded using Geographic Information Systems (GIS) and the USGS National Elevation Dataset (30-m resolution NED). An initial 206 candidate sites were identified and respective watersheds were delineated (fig. 4). Several GIS datasets were processed against the candidate watersheds, producing an assortment of land cover variables for each site. The GIS datasets included layers of socioeconomic (census variables), climatic (precipitation), ecologic (ecoregion), topographic (slope), hydrologic (hydrologic landscape regions), infrastructural (census road variables), and soil characteristics (erosion potential), which were calculated for (1) the entire basin, (2) the proximate segment of the upstream area, and (3) the adjacent riparian area (appendix A, table A1). For more detailed descriptions of techniques, conversions, and guidelines, as well as a complete list of geospatial variables and their descriptions, refer to Sprague and others (2006) and Falcone and others (2007).
The main objective of this study was to examine the effect of different degrees of urbanization on stream ecosystems. Rather than investigate long-term temporal changes in one system, this study monitored a large number of sites at progressively higher levels of urbanization in a short timeframe. Theoretically, the relation of multiple sites at increasing degrees of urbanization should mimic similar development of one site over progressive years of increasing urbanization (Sprague and others, 2006). Therefore, selecting watersheds that represented a range in urbanization from minimum development to maximum development was paramount.
Classifying urbanization was more complicated than basing it solely on urban land use. For this study, watershed and riparian land cover, infrastructural, and demographic variables were integrated into a “candidate” urbanization intensity index (UII) (McMahon and Cuffney, 2000). This UII was calculated from data for all 206-candidate watersheds. Seventeen GIS variables with at least a 0.5 Spearman’s correlation to population density and that were not correlated above 0.5 with watershed size were included in the calculation. After the UII was generated, 70 of the 206 watersheds were selected for reconnaissance.
Each of the 70 remaining candidate sites were visited after their watersheds were delineated, characterized, and ranked with a UII. Field reconnaissance was used to verify GIS results and to evaluate accessibility and safety restrictions. Some sites were relocated upstream or downstream of the original location to obtain reaches with cobble or riffle substrate, or to avoid culverts or other undesirable obstacles. Some sites were excluded because the stream was ephemeral, watershed conditions were impossible to survey, or landowners would not permit access. Whenever possible, sites were selected to provide an even distribution along the UII.
Of the remaining candidate sites, a prerequisite 30 final sites were selected throughout the Willamette River basin and surrounding area. Initially, all 30 watersheds met the required geospatial and ecological characteristics. However, after adjustments for accessibility were incorporated into the drainage delineation, two sites did not fulfill the 20 percent coverage of the Level III ecoregion requirement—South Scappoose Creek in Oregon (6 percent) and Rock Creek in Washington (17 percent). The sites remained in the study after analysis of land cover, topographic, and hydrologic characteristics revealed that the watersheds shared similar characteristics with the other 28 sites and were necessary to fill vacancies in the urbanization gradient. Two other sites were eventually eliminated late in the study because conditions were not favorable for biological sampling in summer after water chemistry samples had been collected the previous winter. The final 28 sites and their basic watershed characteristics are listed in table 1.
A “final” UII was generated and applied to the 28 final sites based on the procedure developed by McMahon and Cuffney (2000). To calculate the final UII, a select group of 24 input GIS variables (table A2) were normalized to watershed area and then sorted by ascending variable percentage. The sorted values then were ranked on a scale of 0 to 100. Variables that correlated with population density (according to a criterion of a Spearman’s rank correlation of greater or equal to 0.70) and remained uncorrelated with watershed area (absolute value of Spearman’s correlation less than or equal to 0.50) were used as inputs for the UII. For each site, all ranked GIS variables that met these constraints were averaged to produce a raw UII. The averaged raw values then were scaled from 0 to 100. The resulting value represented the final urban intensity index value for each site. The final UII differed from the candidate UII because fewer sites were factored into the analysis, different GIS variables were incorporated into the UII calculation, and a higher Spearman’s rank correlation was used (increased from 0.50 from 0.70).
Back to Table of Contents