Bibliographie complète

Abstract A new GCM land surface scheme is introduced, incorporating three soil layers with physically based calculations of heat and moisture transfers at the surface and across the layer boundaries. Snow‐covered and snow‐free areas are treated separately. The energy balance equation is solved iteratively for the surface temperature; the surface infiltration rate is calculated using a simplified theoretical analysis allowing for surface ponding. Snow cover is modelled as a discrete ‘soil’ layer. The results generated by CLASS are compared with those of an older model incorporating the force‐restore method for the calculation of surface temperature and a bucket‐type formulation for the ground moisture. Several month‐long test runs are carried out in stand‐alone mode. It is shown that the surface temperature in the old scheme responds more slowly to diurnal forcing and more quickly to longer term forcing than that modelled by CLASS, while its one‐layer representation of soil moisture proves incapable of reproducing changes in the surface fluxes owing to surface variations of moisture content. Finally, the lumped treatment of snow and soil in the old scheme results in an extremely fast disappearance of the snow pack under certain conditions.