Seasonal Soil Freeze/Thaw Uncertainty Across North America via SEUP Ensemble Land Surface Modeling

Abstract:

Soil temperature and freeze-thaw state transitions exert dominant control on the surface and subsurface water and heat fluxes, carbon budget, and soil–atmosphere fluxes of trace gases. Although, accurate estimation of these metrics is of great importance, there are considerable uncertainties in observational datasets due to the lack of ground measurements, and coarse spatiotemporal resolution of current remote sensing techniques as well as their sensitivity to snow or vegetation. Recently, high-resolution land surface modeling (LSM) provides a unique opportunity to estimate soil freeze-thaw state on the continental scale. However, uncertainties associated with LSMs and meteorological forcing data need to be studied before LSMs are employed in a data assimilation system or future prediction framework. In this study, a nine member ensemble of models (Noah 2.7.1, Noah-MP, and JULES) and meteorological forcing data (MERRA2, GDAS, and ECMWF) with a 5 km spatial resolution, recently developed through the Snow Ensemble Uncertainty Project (SEUP), is employed to explore spatial and temporal variability of seasonal freeze-thaw cycles and winter soil temperature over North America during the 2009-2017 period. Also, the uncertainty of the ensemble is quantified by seasonal snow classes to provide insight regarding the role of seasonal snow characteristics on the variability of model-derived winter characteristics of the soil.