Regional-scale evaluation of a land surface scheme from atmospheric boundary layer observations

The study describes an evaluation of three configurations of the Interactions Soil-Biosphere-Atmosphere (ISBA) land surface scheme fully coupled with the Meso-NH mesoscale atmospheric model. The ability of the modeling system to correctly reproduce the screen-level variables as well as the boundary layer characteristics is examined for more than 30 clear convective days monitored during the CERES 2005 and 2007 field campaigns. For the horizontal resolution considered (8 km), this study shows that the boundary layer characteristics and the low-level variables are better simulated when the subgrid-scale surface process variability is simulated explicitly using the so-called "tiling method." An additional improvement is brought when the CO2 diurnal cycle is used interactively because of the physical link between the stomatal conductance used both for CO2 assimilation and plant transpiration. The parameterization of this link between CO2 and evaporation fluxes improves the simulation of the Bowen ratio and therefore of the atmospheric boundary layer. The last part of the paper discusses the realism of the simulated regional field of CO2 when the carbon configuration is activated in the mesoscale model. Large regional variability of CO2 within the atmospheric boundary layer is found in response to the spatial and seasonal variability of CO2 surface fluxes with respect to the three main land covers in the area: pine forest, extensive winter (wheat), and summer (maize) crops.