Investigating the role of snow water equivalent on streamflow predictability during drought
Snowpack provides the majority of predictive information for water supply forecasts (WSF) in snow-dominated basins across the western US. Therefore, it is important to understand the utility of the snowpack-streamflow relationship during drought years when WSFs are critical. Reduced snowpack magnitude and increased streamflow variability in drought years have exacerbated forecast errors, challenging the assumptions of stationarity of WSFs. Here, we investigate the relationship between snow water equivalent (SWE) and April-July streamflow volume (AMJJ-V) during drought, using observations from 31 USGS streamflow gages and 54 SNOTEL stations. We withhold hydrological drought years and perform a set of idealized experiments to understand how the climatology of different training subsets impacts forecasts of extreme drought years. We generally report overprediction in drought years. However, training the forecast model on drier years minimizes residuals by 20% in withheld drought years relative to a baseline case for colder regions. We also propose an ‘adaptive sampling’ approach of dynamically selecting training years based on antecedent SWE conditions, showing error reductions of up to 20% in historical drought and wet years relative to the period of record. These alternate training protocols provide opportunities for addressing the challenges of future drought risk to water supply planning.
parthkumar.modi@colorado.edu
Civil, Environmental, and Architectural Engineering Graduate Student, CU Boulder