Runoff generation over seasonally-frozen ground: trends, patterns, and processes

Abstract

Understanding and modeling runoff generation over seasonally-frozen hillslopes is a major challenge in hydrology. On the Canadian Prairies, snowmelt drives up to 80% of annual runoff, but the hydrological regime is vulnerable to changing precipitation states, snowpack persistence, snowmelt timing and rates, and frozen ground states. Our ability to understand and predict water partitioning and availability is being challenged by a lack of hillslope-scale climate-runoff observations, the presence of multiple interacting controls, and occurrence of spatial and temporal nonlinearity in runoff responses. I undertook long-term analyses of a 52-year dataset (1962-2013) of climate, snow cover, soil water content, and runoff from three 5 ha hillslopes in Saskatchewan. The aim was to determine how recent changes in climate have impacted upon hillslope rainfall- and snowmelt-runoff, and to unscramble the hierarchy of controls on hillslope snowmelt-runoff generation. These analyses then provided a multi-decadal contextual backdrop to an intensive field campaign that I led during the 2014 snowmelt season. I measured the spatial patterns of controls on runoff to assess the mechanisms behind connectivity and threshold delivery of snowmelt over frozen ground. There are three main conclusions from this research. First, differences between frozen and unfrozen soil infiltrabilities caused contrasting long-term snowmelt- and rainfall-runoff trends: no statistically significant changes were observed for rainfall-runoff amounts, but snowmelt-runoff showed statistically significant decreases over the 52-year record. Second, snowmelt-runoff was driven by hierarchical and condition-dependent controls related to snowfall, snow cover, antecedent soil moisture, and melt season dynamics. Third, for an individual melt season, filling and spilling of micro- and meso-depressions by snowmelt over frozen ground was the driver of hillslope connectivity and runoff delivery. Through a coupled analysis of trends, hierarchies and patterns, this research has advanced our understanding of runoff generation over seasonally-frozen ground. The long-term decrease in spring soil water recharge and snowmelt-runoff is a threat to dryland crop production and economic prosperity in farming. These findings have implications for modeling these threats by guiding new empirical frameworks for lumped hillslope runoff based on what we found in our long terms analysis and identifying what micro- and meso-scale features are important to now include in our process-based distributed snowmelt models.