Although treatment of land-surface interactions has received increased attention in recent years, only limited analyses have been conducted to validate land-surface models (LSMs), both to determine their shortcomings and to suggest methods of improvement. We use observed plant-available soil moisture for the top one meter of soil from Illinois (USA), China, Russia, and Mongolia to evaluate and improve land-surface modeling. Special focus is placed upon quantifying errors due to parameterization of the vegetation-soil system and accuracy of the meteorological forcing data.
<p>
As part of the Global Soil Wetness Project (GSWP), 10 different LSMs were forced with meteorological observations for a two year period (1987-88). The GSWP experiment used the meteorological forcing data and soil parameters from the International Satellite Land Surface Climatology Project (ISLSCP) CD. Because each model received the same meteorological forcing variables we were able to make meaningful comparisons among the 10 models. Further, by comparing the results from each of these models with soil moisture observations, we were able to assess how well the models performed in different geographic regions. In general, the models are successful in representing the seasonal cycle of soil moisture for China and Illinois, but unable to capture the full range of soil moisture values in Mongolia.
<p>
Three additional sets of modeled soil moisture allow us to explore issues associated with the performance of LSMs further. Two reanalysis projects carried out at the European Centre for Medium-Range Weather Forecasts (ECMWF) and the National Center for Environmental Prediction (NCEP) have created global soil moisture data sets that include the two-year period used in the GSWP experiment. Each reanalysis project uses identical soil parameters for the entire globe. A third set of results for this two-year period were obtained from off-line experiments using the Simplified Simple Biosphere (SSiB) model (including its default grassland or agricultural parameters) forced with locally observed meteorological data. Comparisons of the results from these varying approaches allow us to evaluate how well the meteorological forcing data from the ISLSCP CD captured important weather phenomenon, and enable us to draw some conclusions concerning how much land-surface modeling suffers from improper soil parameters, poor meteorological forcing data, or faulty LSMs.