My collaborators on this work include Bruno Tremblay (McGill Univ.), Marc Stieglitz (Georgia Tech.), Stephen Déry (UNBC), and Gavin Gong (Columbia Univ.).

The amount and distribution of snowfall in the Arctic has significant effects on global climate. However, measurements of snowfall from gauges are strongly biased. A new method is described for reconstructing snowfall from observed snow depth records, meteorological observations, and running the NASA Seasonal-to-Interannual Prediction Project Catchment Land Surface Model (NSIPP CLSM) in an inverse mode. This method is developed and tested with observations from Reynolds Creek Experimental Watershed. Results show snowfall can be accurately reconstructed on the basis of how much snow must have fallen to produce the observed snow depth. The mean cumulative error (bias) of the reconstructed precipitation for 11 snow seasons is 29 mm snow water equivalent (SWE) for the corrected gauge measurement compared to -77 mm SWE for the precipitation from the corrected snow gauges. This means the root-mean-square error of reconstructed solid precipitation is 30% less than that of gauge corrections. The intended application of this method is the pan-Arctic landmass, where estimates of snowfall are highly uncertain but where more than 60 years of historical snow depth and air temperature records exist.

This paper served as a proof-of-concept for development of the PASR product. It was published here:

Cherry, J., B. Tremblay, S. Déry, M. Stieglitz, “Solid precipitation reconstruction using   snow depth measurements and a land surface hydrology model”, Water Resour. Res, Vol. 41, No. 9, W09401, 10.1029/2005WR003965. PDF