Use of Field Experiments in Improving the Land-surface Description in Atmospheric Models: Calibration, Aggregation and Scaling

Abstract

When the first integrated terrestrial experiments were designed in the early 1980s, land-surface processes in atmospheric models were poorly described, as compared to present-day parameterisations. There was also very little data available that was suitable for the local calibration of surface schemes or few datasets for the largescale implementation of these schemes within mesoscale Numerical Weather Prediction (NWP) or climate models. At that time, two main tasks faced the Community. First, to collect enough information from representative land uses to develop and calibrate advanced landsurface parameterisations of the water and energy budgets (in place of classical simple schemes inspired from the bucket model of Manabe, 1969). Second, to develop an upscaling strategy that would allow those surface schemes to be used at a grid scale of several tens of kilometres, while keeping complexity and Computing cost compatible with the other physical parameterisations of the atmosphere. The integrated terrestrial experiments HAPEX-MOBILHY, FIFE, EFEDA, NOPEX, HAPEX-Sahel and BOREAS opened the way for a more realistic parameterisation of land-surface changes in NWP and GCM models and thus solved some of the Problems outlined above. The successes and failures of this strategy are summarised briefly.