PASR

A new product, the Pan-Arctic Snowfall Reconstruction (PASR) is developed to address the problem of cold season precipitation gauge biases for the 1940-1999 period. The method used to create the PASR is different from methods used in other large-scale precipitation data products and has not previously been employed for estimating pan-arctic snowfall. The NASA Interannual-to-Seasonal Prediction Project Catchment Land Surface Model is used to reconstruct solid precipitation from observed snow depth and surface air temperatures. The method is tested at four stations in the United States and Canada where results are examined in depth. Reconstructed snowfall at Dease Lake, British Columbia and Barrow, Alaska is higher than gauge observations. Reconstructed snowfall at Regina, Saskatchewan and Minot, North Dakota is lower than gauge observations, probably because snow is transported by wind out of the Prairie region and enters the hydrometeorological cycle elsewhere. These results are similar to gauge biases estimated by a water budget approach. Reconstructed snowfall is consistently higher than snowfall from ECMWF Reanalysis-40 but does not have a consistent relationship with snowfall derived from the WMO Solid Precipitation Intercomparison Project correction algorithms. Advantages of the PASR approach include (1) the assimilation of snow depth observations captures blowing snow where it is deposited and (2) the modeling approach takes into account physical snowpack evolution. These advantages suggest the PASR product could bea valuable alternative to statistical gauge corrections and that arctic ground-based solid precipitation observing networks might emphasize snow depth measurements over gauges.