An evaluation of a point snow model and a mesoscale model for regional climate simulations
Date
2006
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Delaware
Abstract
The ability of atmospheric and climate models to accurately predict snow cover and areal snow distribution is vital for climate simulations. However, due to computational efficiency, general circulation and mesoscale models employ simplified snow physics, limiting their ability to simulate the true nature of the snowpack. A one-dimensional point snow model SNTHERM and Mesoscale Model 5 (MM5) are compared to assess their utility in the evaluation of snow cover. MM5 was run over an upper Great Plains domain for six consecutive winters, producing one set of snowcover predictions from its simplified snow model. Atmospheric output from MM5 was used to drive SNTHERM at the locations of six first-order National Weather Service (NWS) stations, providing snowpack estimates from a more detailed snowpack model. Snowpack variables from each model were compared to observations at the same six weather stations. Results indicate that there is general agreement between the models on daily to monthly time scales, although their physics vary considerably. Snow physics comparisons demonstrate that NOAH LSM ground and subsurface temperatures are too cold because of the simplistic one-layer slab representation, causing snow accumulation (ablation) overestimations (underestimations). A modified version of SNTHERM was able to reasonably reproduce NOAH LSM snow accumulation and ablation rates by using average monthly NOAH LSM 0.05 cm soil temperature as a surrogate for ground temperature and NOAH LSM snow thermal conductivity. Results establish that not for a 400 kg m -3 maximum snow density cutoff, NOAH LSM snow thermal conductivities were too high, leading to a very dense, thermally conductive snowpack, further exacerbating slow ablation rates.