One problem with current land surface schemes (LSS) used in weather prediction and climate models is their inabilty to reproduce streamflow in river basins. This can be attributed to the weak representation of their upper (infiltration) and lower (baseflow) boundary conditions in their water balance / transport equations. Operational (traditional) hydrological models, which operate on the same spatial scale as a LSS, on the other hand, are able to reproduce streamflow time series. Their infiltration and baseflow equations are often empirically based and therefore have been neglected by the LSS community. It must be argued that we need to include a better representation of long time scales (as represented by groundwater and baseflow) into the current LSS to make valuable predictions of streamflow and water resources.

This paper concentrates on the lower boundary condition of water fluxes within LSS. It reviews briefly previous attempts to incorporate groundwater and more realistic lower boundary conditions into LSS and summarizes the effect on the runoff (baseflow) production time scales as compared to currently used lower boundary conditions in LSS. The Noah - LSM in the NLDAS (North American Data Assimilation System Project) setting is used to introduce a simplified groundwater model, based on the linearized Boussinesq equation. The Noah - LSM will be coupled to a linear routing model to investigate the effects of the new lower boundary condition on the water balance (in particular, streamflow) in small to medium sized catchments in the NLDAS domain. Model performance will be evaluated for parameters within the parameter space of the Noah model.