Assessment of Passive Microwave Satellite and Modeled Snow Water Equivalent Estimates in the Red River Basin of the North

The Red River basin (RRB) of the North is very vulnerable to spring snowmelt floods because of its flat terrain, low permeability soils and the presence of river ice jams resulting from the river’s northward flow direction. Accurate predictions of the onset and magnitude of major flood events in the RRB have been very difficult to forecast, in part due to limited field observations of the primary snow melt drivers such as the snow water equivalent (SWE). Coarse-resolution (25-km) passive microwave observations from satellite instruments are well suited for the monitoring of SWE because they are sensitive to the distribution of snow in the landscape. They also cover large areas with nearly daily and constant repeat periods and operate virtually under all weather conditions, thereby providing temporally consistent and observation-based continuous records of SWE. This study compared daily satellite SWE observations from 2003 to 2017 to spatially distributed, modeled output from SNODAS and GlobSnow-2 in the RRB. Our study employed operational satellite SWE products from past and current passive microwave satellite missions (e.g. AMSR2, AMSR-E, SSM/I) as well as SWE retrieved from brightness temperature (T_b) measurements using a physically-based Chang-type scattering approach screened for daily melt and cloud cover events. Cloud cover was determined using a temperature-insensitive polarization difference ratio; melt was obtained from calculating the diurnal SWE differences. Satellite and model output SWE values were summarized by aggregating the results spatially (i.e. to the NOAA North Central River Forecast Center forecast basins) and temporally (i.e. by obtaining daily screened and weekly unscreened maximum SWE composites) to investigate the value of analyzing daily satellite SWE observations relative to weekly maximums. We examined the ability of the satellite and model output observations to consistently capture the annual maximum SWE. Temporal accuracy of screened and unscreened satellite and model output SWE was evaluated by computing (1) the linear correlation coefficient (R) and (2) the Nash-Sutcliffe model efficiency index.