Analysis of uncertainty in simulated exchange of heat and moisture at the land-atmosphere interface

Abstract

Land surface models (LSMs) serve to describe the atmosphere-land surface exchange in numerical weather prediction models (NWPMs) and global circulation models (GCMs). The use of empirical soil and vegetation parameters in LSMs introduces uncertainty that propagates and affects predictions of the lower boundary conditions. To statistically assess that uncertainty in predicted evapotranspiration (water transport by direct evaporation from bare ground and canopy and transpiration by the canopy) and ground heat flux for natural ranges of atmospheric soil and vegetation conditions, the Gaussian Error Propagation method is utilized. The assessed uncertainties in direct and canopy water evaporation, transpiration and ground heat flux display prominent diurnal cycles. Prediction of evapotranspiration in desert areas is limited by the uncertainty in the evaporation of water collected on the canopy and transpiration. To improve predictions of evapotranspiration the maximal canopy storage and shielding factor should be determined with higher accuracy. It is found that uncertainty in ground heat flux is particularly great in dry and warm areas covered with sandy clay loam. A better prediction of ground heat flux requires a better parameterization of thermal conductivity and a higher degree of accuracy of the pore size distribution index.