Combining Stochastic Techniques with Spatial Analysis for Predicting Stream Temperature

The correlation between the stochastic components of air and water temperatures has been broadly applied in simulation and prediction of water temperature for a diversity of catchment sizes. Advantages of stochastic water temperature models can be summarized by the limited amount of input data required, simplicity of model development, and the ability to provide output on a daily time step conducive to conservation and impact analyses. Stochastic models are also capable of offering insights into the thermal regime of waterways by highlighting temperature fluctuations unexplained by model outputs. While many studies have focused on the functionality of stochastic approaches, few, if any, have attempted to incorporate the effects of spatial scale and variability on the thermal behavior and heterogeneity of water resources. This study investigates the performance of a stochastic model in predicting stream temperatures in an 842 Km² perennial salmon bearing watershed in Southeastern Washington, USA. Specifically, this study combines stochastic methodologies with an array of spatial statistic techniques to identify an optimal relationship between meteorological, geomorphic, and landscape drivers that affect stream temperatures along critical salmonid habitat tributaries within an inland Northwest watershed.