Stream ecosystems have experienced significant negative impacts from land use, resource exploitation, and urban development. Statistical models allow researchers to explore the relationships between these landscape variables and stream conditions. Weighting the relevant landscape variables based on hydrologically defined distances offers a potential method of increasing the predictive capacity of statistical models. Using observations from three grouped watersheds in the Portland Metro Area (n=66), I have explored the use of three different weighting schemes against the standard method of taking an areal average. These four different model groups were applied to four stream temperature metrics: mean seven-day moving average maximum daily temperature (Mean7dTmax), number of days exceeding 17.8 °C (Tmax7d>17.8), mean daily range in stream temperature (Range_DTR), and the coefficient of variation in maximum daily temperature (CV_Tmax). These metrics were quantified for the 2011 dry season. The strength of these model groups were also examined at a monthly basis for each of the four months within the dry season. The results demonstrate mixed effectiveness of the weighting schemes, dependent on both the stream temperature metric being predicted as well as the time scale under investigation. Models for Mean7dTmax showed no benefit from the inclusion of distance weighted metrics, while models for Range_DTR consistently improved using distance weighted explanatory variables. Trends in the models for 7dTmax>17.8 and CV_Tmax varied based on temporal scale. Additionally, all model groups demonstrated greater explanatory power in early summer months than in late summer months.
| Identifer | oai:union.ndltd.org:pdx.edu/oai:pdxscholar.library.pdx.edu:open_access_etds-4128 |
| Date | 16 August 2016 |
| Creators | Watson, Eric Craig |
| Publisher | PDXScholar |
| Source Sets | Portland State University |
| Detected Language | English |
| Type | text |
| Format | application/pdf |
| Source | Dissertations and Theses |
Page generated in 0.0018 seconds