The present work is a step in evaluating the impact of running a surface scheme as a surface model separate from the atmospheric model. This study investigates the differences in the surface fluxes generated by a surface model when run at various resolutions but with the same meteorological forcing at a given (low) resolution. A first set of experiments shows that the results obtained with the surface model can reproduce quite closely those obtained by running the same surface scheme inside the driving atmospheric model if the meteorological forcing is provided with sufficient frequency to the surface model. It is found in particular that providing accumulated output of incoming solar radiation instead of only instantaneous fluxes permits a much lower frequency for the exchange of information from the atmospheric model to the surface model for the same degre of accuracy.
While preparing the geophysical fields for our experiments, we had at our disposal different sources of data as well as a choice of algorithms to aggregate them at the desired resolution. We found that these different inputs had a sizeable impact on the results of the surface model and realized that the proper specification of the various vegetation masks and soil types together with their translation into the parameters used by our surface model (for the moment a version of ISBA) is very important and probably more so than the resolution issue. This motivated us to study in more detail the sensitivity of the surface model to its raw and derived input parameters alongside with the sensitivity to the horizontal resolution at wich the surface model is run. We will present at the conference the results obtained at that date on the relative impact of the uncertainty in the input parameters and of the scale at which they are specified.
Supplementary URL: