The goal of this paper is to evaluate atmospheric water budget components from Eta model output. While radiosonde based atmospheric water budget estimates are, due to temporal and spatial sampling, too imprecise to be applied to the areas less than large-scale river basins, it is still not clear whether mesoscale models can provide us with accurate estimates of atmospheric water budget parameters for continental-scale areas. There have been no direct comparisons of observation-based and model-generated atmospheric water budget terms.

Our approach is to estimate atmospheric vapor flux convergence from 5+ years of Eta MOLTS subset with temporal and spatial limitations of radiosonde array. Comparison of these estimates to the benchmark radiosonde-based vapor flux convergence on the time scales from days to several years will answer the following questions: - are there any biases in Eta model vapor flux convergence estimates? - what are the errors of Eta model vapor flux convergence relative to the natural variability of the water budgets on different time scales? - how well can we close model water budget for the continental-scale area? - how different are derived water budget parameters - evapotranspiration and surface/subsurface storage - in Eta model and radiosondes?

Assuming that Eta model performance is the best during radiosonde data assimilation times (00Z and 12Z) and that errors grow for smaller basins, our estimates will put a lower limit of the error for the model-based water budget estimates.