Regional climate simulation with a high resolution GCM: surface radiative fluxes

Abstract

The ability of a high resolution (T106) version of the ECHAM3 general circulation model to simulate regional scale surface radiative fluxes has been assessed using observations from a new compilation of worldwide instrumentally-measured surface fluxes (Global Energy Balance Archive, GEBA). The focus is on the European region where the highest density of observations is found, and their use for the validation of global and regional climate models is demonstrated. The available data allow a separate assessment of the simulated fluxes of surface shortwave, longwave, and net radiation for this region. In summer, the incoming shortwave radiation calculated by the ECHAM3/T106 model is overestimated by 45 W m⁻² over most of Europe, which implies a largely unrealistic forcing on the model surface scheme and excessive surface temperatures. In winter, too little incoming shortwave radiation reaches the model surface. Similar tendencies are found over large areas of the mid-latitudes. These biases are consistent with deficiencies in the simulation of cloud amount, relative humidity and clear sky radiative transfer. The incoming longwave radiation is underestimated at the European GEBA stations predominantly in summer. This largely compensates for the excessive shortwave flux, leading to annual mean net radiation values over Europe close to observations due to error cancellation, a feature already noted in the simulated global mean values in an earlier study. Furthermore, the annual cycle of the simulated surface net radiation is strongly affected by the deficiencies in the simulated incoming shortwave radiation. The high horizontal resolution of the GCM allows an assessment of orographically induced flux gradients based on observations from the European Alps. Although the model-calculated and observed flux fields substantially differ in their absolute values, several aspects of their gradients are realistically captured. The deficiencies identified in the model fields are generally consistent at most stations, indicating a high degree of representativeness of the measurements for their larger scale setting.