Creation and Characterization of an Observation-based, Multi-dataset Snow Water Equivalent Product for Climate Model and Climate Forecast Validation

Northern Hemisphere snow water equivalent (SWE) is a key element of the Northern Hemisphere energy and water balance. It is expected to respond in a complex way to projected temperature and precipitation changes, decreasing at midlatitudes associated with warming and increasing at high latitudes associated with increases in precipitable water. Evaluation of past and projected SWE trends and initialization of seasonal to decadal forecasts requires a gridded observational SWE product. For snow cover extent several such data sets exist and their intercomparison has led to a reasonable understanding of the errors therein. We use a similar multi-data set approach for SWE, comparing an ensemble of daily gridded products in order to produce a combined product with quantitative measures of spread.

We use several daily, observation-based SWE products obtained from remote sensing, land surface assimilation systems, physical snow models and reanalyses: (1) GlobSnow analysis based on satellite-based passive microwave retrievals; (2) Global Land Data Assimilation System (GLDAS); (3) Crocus distributed physical snow model driven by ERA-Interim meteorology; (4) ERA-Interim/Land reanalysis; (5) Modern Era Retrospective Analysis for Research and Applications (MERRA reanalysis); and (6) Snow Model simulations forced by downscaled MERRA meteorology (available for the Arctic only). While the climatologies of the various SWE products differ by as much as 50%, their interannual variability and daily anomalies are comparable. The latter correlations are between 50%-90% on both interannual and intraseasonal time scales. In particular, we examine how inter-product spread, spatial correlation and temporal correlation of total Northern Hemisphere snow mass vary over the analysis period and among the various data products. Our analysis indicates that the reanalysis-derived products (3-6) are most similar to one another in comparison to GlobSnow (1) and GLDAS (2). Preliminary analysis also suggests that details of the driving meteorology rather than the particular snow or land model bear the largest influence on the resulting SWE anomalies in data products (2-6). Evaluations of historical simulations and seasonal forecasts from coupled climate model simulations using our combined data product will be presented.