Quantifying the Effect of Catchment Snow Cover on Stream Temperature Dynamics in a Mountainous Region

Abstract

Stream temperature is an important water quality parameter, particularly as it influences thermal habitat suitability for a range of species. Empirical models are commonly used for estimating stream temperature at locations where no or limited data exist and for making projections of stream temperature response to future climate scenarios. Previous research has shown that snow dynamics strongly influence stream temperature in mountainous regions. The objective of this study is to evaluate the use of catchment-scale fractional snow cover ($f_{\mathrm{sc}}$) as a predictor in temporal stream temperature models. The study focused on 26 catchments in the southern Coast Mountains of British Columbia, where stream temperature has been monitored for at least 3 years for the months of April through October between 2016 and 2020. Daily mean air temperature series were estimated for each location using the ECMWF Reanalysis v5 (ERA5) daily surface product. Daily time series of fractional snow cover in the catchment areas were extracted from the MODIS snow cover product. Mean fractional snow cover for June of each year was used as a predictor in models for the following July to October to represent the thermal memory associated with catchment snow cover. Statistical modelling indicated strong support for including $f_{\mathrm{sc}}$ in May, June, September and October. For May and June, $f_{\mathrm{sc}}$ was significant for catchments larger than about 10 km², but there did not appear to be an area threshold for significance for September and October. Mean June snow cover as an indicator of antecedent snow conditions was a significant predictor for some locations for July and August. Further research should explore the utility of $f_{\mathrm{sc}}$ for sites with longer periods of record, and should also explore the use of alternative snow indices, such as snow cover predicted by a hydrological model.