Assessment of snow water equivalent characteristics in time and space over the Mackenzie River basin

This study quantifies the spatiotemporal variability of snow water equivalent (SWE) using a set of complementary statistical approaches across site, grid, and sub-basin scales. It incorporates in-situ measurements, remote sensing, reanalysis, and machine learning-derived datasets for the period 2000–2020 across the Mackenzie River Basin (MRB). Employing a comprehensive evaluation framework, we assessed the performance of multiple gridded SWE datasets against area-averaged in-situ observations at both lowland and highland sites over time, and ERA5-Land reanalysis. Spatiotemporally, the gridded SWE datasets reproduce well-observed seasonal and interannual cycles and the patterns of spatial variability; however, the random forest (RF) model demonstrates superior accuracy and reliability, achieving the highest correlations (R2 = 0.72 and R2 = 0.95) and the lowest errors (RMSE = 43.1 and 37.2 mm) when compared to the lowland and highland in-situ measurements, respectively. Annual and seasonal SWE variations were prominent, with peak values exceeding 700 mm in highland regions during late winter and early spring, reflecting the region’s climatic conditions. The RF feature importance and principal component analysis indicate that snow-related parameters, including snow depth, density and snowfall are the primary drivers of SWE estimates in MRB. Our results reveal a slight, though statistically insignificant, decreasing trend in SWE (for example, a Sen’s slope of −0.301 mm/year with a p-value of 0.26 in highland in-situ sites), which could suggest potential impacts of climate change on snow accumulation. These findings underscore the critical role of SWE in the basin’s hydrological processes and future water resources. The insights provided could further enhance our understanding of SWE spatiotemporal variations, providing a basis for improved inputs in the future numerical hydrologic modeling within the Mackenzie River basin.