Using a data-driven statistical model to better evaluate surface turbulent heat fluxes in weather and climate numerical models: a demonstration study

Abstract. This study proposes the use of a data-driven statistical model to freeze the errors due to differences in environmental forcing when evaluating the surface turbulent heat fluxes from weather and climate numerical models with the observations. It takes advantage of continuous acquisition over approximately ten years of near-surface sensible and latent heat fluxes (H and LE respectively) together with ancillary parameters over the supersite "Météopole" of the French national research infrastructure ACTRIS-FR, located in Toulouse. The statistical model consists of several multi-layer perceptrons (MLPs) with the same architecture. Thirteen variables characterizing the environmental forcing in the surface layer at an hourly time scale are used as input parameters to estimate H and LE simultaneously. The MLPs are trained using 5-year observational data under a 5-fold cross-validation. The remaining data is used to test the estimates on unknown conditions. A case study is performed with data from a regional climate simulation. The performance of the statistical model ranges within the state-of-the-art surface parametrization schemes on hourly and seasonal time scales. It has also a good generalization ability, but hardly estimates negative H and large LE. The statistical model is used to evaluate the simulated fluxes under the simulated environment to better examine the flaws of their numerical formulation throughout the simulation. Comparison of simulated fluxes with observed and MLP-based fluxes show different results. According to MLP-based fluxes in the simulated environment, the land surface scheme of this climate model tends to underestimate large sensible heat flux. Thus, it incorrectly partitions between surface heating and evaporation during the late summer. Our innovative method provides insight into differently evaluating the simulated near-surface turbulent heat fluxes when a long period of comprehensive observations is available. It can usefully support ongoing efforts for improvements of surface parametrization schemes.