Developments in a land surface scheme used for regional climate simulations in SWECLIM

During the latest years a number of more or less complicated coupled atmosphere-hydrology land-surface schemes (LSSs) has been presented. The degree of complexity needed depends on the processes that must be described. The most important task for a surface scheme in a regional or global climate model is to correctly divide the net radiation available energy between sensible and latent heat fluxes. That means that it is mostly enough with 2-3 soil layers and 1-2 canopy layers.

It has been argued that one should be careful introducing sophisticated formulations into a parameterization scheme. The only reason for introducing new formulations should not just be an improved physical description but also because it can be justified from evaluations of the results from the model. This philosophy has been followed during the development of the Rossby Centre regional Atmospheric climate model (RCA), which is used within the Swedish Regional Climate Modelling Programme (SWECLIM).

The global climate models are too coarse in horizontal resolution to be used as tools for strategic decisions due to changing climate effects on a regional scale. Therefore, SWECLIM was initiated to study the future climate in the Nordic region for horizontal resolutions down to 22x22 km². RCA is based on the operational weather forecast model HIRLAM, used by several European countries. The LSS in RCA has been improved following the ISBA-scheme and experiences based on hydrological modelling with the HBV model.

At the moment, the LSS consists of a two-active-soil-layers model where the force-restore method is used for soil temperature and moisture. Vegetation acts as a moisture sink through transpiration from the deep layer and drainage and runoff are based on formulations in HBV. A freezing/thawing algorithm of soil moisture has been introduced to reduce, as otherwise observed, anomalously low temperatures in winter.

Snow plays an important role for the climate in the Nordic region. Model results indicate that sub-grid terrain variability has to be considered for snow melt. Therefore, based on hydrological experience, the snow melt is set dependent on sub-grid temperature variability which is a function of sub-grid orography. Below a threshold in snow layer depth the snow cover goes below 100%, which is important for correctly forecasting averaged grid-square albedo and air temperature. To avoid unrealistic snow melt for fractional snow cover conditions, the snow temperature is set different from the ground surface temperature.

The canopy layer has its own prognostic temperature and prognostic precipitation interception. In the region of interest, snow free canopy is often observed at the same time as the ground is snow covered. For this situation it is especially important with a canopy temperature different from the ground surface temperature to correctly forecast sensible and latent heat fluxes, respectively.