Physically consistent mesoscale model evaluation in complex terrain

This study introduces a novel evaluation technique for mesoscale atmospheric models in complex terrain, addressing challenges related to grid point (GP) selection, model resolution, and differences in the terrain and measurement heights between model and observation sites. The technique includes a pre-evaluation step designed to enhance model evaluation accuracy by correcting for systematic biases arising from discrepancies in sensor height between surface-layer diagnosed variables and actual measurement heights. Additionally, a lapse rate correction is proposed that takes into account the temporal evolution of the valley atmosphere. We use a GP for the evaluation with point measurements that represents the topographic characteristics at the measurement site better than the nearest GP. The analysis of a case study using a WRF simulation with different grid spacing is conducted for a synoptically undisturbed valley wind day over the Inn Valley, to study the effect of grid resolution during an entire diurnal cycle. The proposed method shows significant potential in reducing uncertainties in model performance assessments by selecting GPs that better represent in-situ conditions, including accurate land cover and orography characteristics. This approach is expected to facilitate more reliable inter-model comparisons and enhance the overall robustness of atmospheric model evaluations in complex terrains.