Ensemble data assimilation to quantify land-atmosphere coupling-strength errors in numerical weather prediction model formulation

Experiments to quantify the convective daytime coupling between the atmosphere and the land surface via similarity theory-based parameterizations typical of mesoscale numerical weather prediction models are constructed and analyzed. We argue that ensemble data assimilation provides a context for unambiguous interpretation of model error by constraining model state elements to follow observations while adhering to model physics. The atmospheric surface layer is constrained with ensemble data assimilation of shelter and anemometer-height observations (2-m temperature, 2-m water vapor mixing ratio, and 10-m winds) in a single column model that uses the same physical parameterization schemes as the Weather Research and Forecast (WRF) mesoscale numerical weather prediction model.

Results show that thermal and momentum coupling between the atmosphere and the soil is systematically poor. Increments to model temperature, wind, and humidity from the assimilation are systematically small, indicating that the model is closely following the observations. The increments that remain reveal systematic model error, which is manifested by the assimilation acting to reduce vertical thermal gradients and strengthen vertical momentum gradients through the atmospheric surface layer. Systematic modification of the vertical surface-layer gradients acts to reduce sensible heat flux and increase momentum flux and drag. This is consistent with a trend away from free convection and with the hypothesis that Monin-Obukhov similarity theory is often invalid. Qualitatively similar results are found whether or not the soil is specified from observations, or it evolves according to a land surface model that may or may not be incremented by the data assimilation.