Evaluation of Gridded Snow Water Equivalent Products in British Columbia, Canada

Abstract:

Accurate regional scale estimates of snow water equivalent (SWE) are essential to skillful snow state initializations in hydrological models and improving streamflow predictions. This is especially true for the nival rivers of British Columbia, Canada. Several state-of-the-art gridded products are evaluated against in situ manual snow concourse surveys for their ability to estimate SWE over the province. These products span the range of reanalyses (ERA-Interim and MERRA), land data assimilation systems (GLDAS-1 and -2), observational products (CMC Snow Analysis), microwave emission products (NSIDC SWE Climatology), and hybrid observational products (GlobSnow). Interannual time series for the manual snow surveys and each gridded SWE product are constructed for each survey period and evaluated for amplitude and phase agreement. Comparisons of these timeseries demonstrate that the products in this study generally significantly underestimate SWE, with the most negative relative biases reported near the end of the snow season. Correlations suggest that relative interannual variations are best captured by most products earlier in the snow season, but significant geographic trends are not apparent. Skill scores also decline through the first few months of the season but show improvement as melt-off occurs in May and June. Peak SWE, which typically occurs on or near the beginning of April survey, is one of the most challenging times for accurate estimation, possibly associated with difficulties in estimating SWE for extremely large snowpacks.