We use cookies to ensure that we give you the best experience on our website. You can change your cookie settings at any time. Otherwise, we'll assume you're OK to continue.

Durham University

Research & business

View Profile

Publication details for Professor B. Huntley

Hancock, S., Huntley, B., Ellis, R. & Baxter, R. (2014). Biases in Reanalysis Snowfall Found by Comparing the JULES Land Surface Model to GlobSnow. Journal of Climate 27(2): 624-632.

Author(s) from Durham


Snow exerts a strong influence on weather and climate. Accurate representation of snow processes within models is needed to ensure accurate predictions. Snow processes are known to be a weakness of land surface models (LSMs), and studies suggest that more complex snow physics is needed to avoid early melt. In this study the European Space Agency (ESA)'s Global Snow Monitoring for Climate Research (GlobSnow) snow water equivalent and NASA's MOD10C1 snow cover products are used to assess the accuracy of snow processes within the Joint U.K. Land Environment Simulator (JULES). JULES is run offline from a general circulation model and so is driven by meteorological reanalysis datasets: Princeton, Water and Global Change-Global Precipitation Climatology Centre (WATCH-GPCC), and WATCH-Climatic Research Unit (CRU). This reveals that when the model achieves the correct peak accumulation, snow does not melt early. However, generally snow does melt early because peak accumulation is too low. Examination of the meteorological reanalysis data shows that not enough snow falls to achieve observed peak accumulations. Thus, the earlier studies' conclusions may be as a result of weaknesses in the driving data, rather than in model snow processes. These reanalysis products bias correct precipitation using observed gauge data with an undercatch correction, overriding the benefit of any other datasets used in their creation. This paper argues that using gauge data to bias-correct reanalysis data is not appropriate for snow-affected regions during winter and can lead to confusion when evaluating model processes.