A new version of the land surface scheme has been completed and is now applied in comparative tests of version 2 of the Rossby Centre Regional Atmospheric Climate Model (RCA2) using analysed fields from the ECMWF reanalysis project (ERA). The scheme contains two soil layers and a vegetation layer. There are two prognostic temperatures, one covering the top soil layer plus vegetation and one for a second, deeper soil layer. There is also a third, bottom soil temperature relaxed to six-hourly ERA fields. For soil moisture there are two prognostic layers but no bottom relaxation is used. A hydrologically-based soil moisture model (beta model) is used to represent subgrid soil moisture variability. A hydrological snow model makes regard to subgrid temperature variability using a geographical database for variance of topography. There are equations for heat and moisture exchange between the two soil layers. Here the hydraulic and thermal properties depend on soil type and soil moisture. Transpiration flux transports moisture from both soil layers depending on a stomatal resistance of vegetation surfaces as function of daylight intensity, soil water deficit, fraction of frozen soil water, air temperature and water vapour pressure deficit in the air. A treatment of rainfall interception on vegetation is used, broadly following the ISBA model, with a vegetation layer storing intercepted water. Subgrid weighting of albedo, surface roughness and parameters for calculating surface resistance is made using a geographical database for area fraction of forest and open land. The leaf area index varies seasonally for short vegetation and for deciduous forest, but not for coniferous forest. A soil freezing/melting algorithm influencing soil temperature is used. Implicit methods are used for solving the equations of most surface variables. A summary of model results compared to observations, is given at the end of the report.