Multi-objective calibration of the land surface model SEWAB

A complex land surface model is generally characterized by a multitude of parameters, which are not exactly known a priori. Therefore a model calibration is needed. Manual calibrations are always subjective, depending on the experience of the modeler. Automatic calibration procedures overcome these shortcomings. The land surface model SEWAB is calibrated by use of the Multi-Objective Shuffled Complex Evolution Metropolis (MOSCEM) algorithm. This algorithm minimizes two objective functions, which compares the measured and simulated latent and sensible heat fluxes. For the calibration and validation period datasets of turbulent heat fluxes measured during the field campaign LITFASS2003 within the project EVA-GRIPS (Regional Evaporation at Grid/Pixel Scale over heterogeneous Land Surfaces) are used. Significant differences in latent and sensible heat fluxes of three types of land use during the main vegetation period will be shown and compared with the model output.