Comparing SNOTEL soil moisture pulse and Sentinel-1 estimates of snowmelt runoff timing across the western U.S.: implications for radar remote sensing

Abstract

Satellite-based synthetic aperture radar (SAR) has been used to assess and quantify snowmelt across large spatial and temporal scales. While there have been recent advancements in SAR-based snow water equivalent (SWE) retrieval methods, obtaining accurate estimates of SWE requires knowledge of the amount of liquid water content in the snowpack to minimize uncertainty. Recent studies have utilized Sentinel-1 SAR to identify snowmelt runoff onset in complex, high-elevation terrain based on the seasonal minimum backscatter received by the sensor, however, detailed investigations into the snowpack state before and after snowmelt runoff onset are lacking. In this study, I integrated repeat field measurements, SNOw TELemetry (SNOTEL) station data (n=260) across the western U.S., and paired Sentinel-1 SAR estimates of runoff generation to 1) assess how snowpack conditions evolved prior to and after Sentinel-1 SAR-derived runoff onset estimates, and 2) evaluate Sentinel-1 SAR estimates of runoff generation with SNOTEL-derived estimates of melt output via soil moisture "pulses". I found that SNOTEL soil moisture pulses preceded Sentinel-1 SAR estimates of snowmelt runoff onset by a median 3 days (± standard deviation 25.3 days) and post-dated peak SWE by a median of 3 days (± standard deviation 18.2 days). Soil moisture pulse dates occurred earliest in montane forests/prairie snowpacks and latest at SNOTEL stations in maritime snowpacks. Snow density and number of positive degree days on soil moisture pulse date increased with latitude and longitude and decreased with elevation. While satellite-based estimates of snowmelt runoff onset provide a promising methodology for improving spaceborne retrievals of SWE, I emphasize the importance and influence of local climatological conditions on runoff onset signal clarity for both in-situ and satellite-based estimates.