Soil Science Challenges in Land Surface and Global Climate Modeling.

Land surface models simulate mass and energy fluxes at the regional, continental and global scale and, when coupled to atmospheric or climate models, are used to forecast weather and climate or to study climate change under different atmospheric composition scenarios.  In these models, soil physical characteristics are crucial for the correct computation of soil moisture profiles, water stress by the vegetation and the right partition of precipitation in evapotranspiration and runoff, as well as the correct partition of net radiation in sensible and latent heat fluxes. On the other hand, these partitions also depend on the plant physiology and soil chemical properties. We present two examples in which better parameterizations of soil physical and chemical properties lead to improved performance of land surface models. In the first example, a land surface model uses three different soil physical parameterizations: First, the soil water retention curve is determined using a global pedotransfer lookup table and a global soil texture dataset. Second, the same pedotransfer lookup table is used, but soil texture is measured in situ. Third, all soil retention curve parameters are measured locally. Model performance is evaluated against measured patterns of soil moisture, and improved significantly as more local information was used. In the second example, it was implemented a new parameterization for the physiological parameter Vcmax as a function of available soil phosphorous. This new parameterization allowed improvements in the simulations of spatial variability of vegetation biomass in Amazonia. These two examples demonstrate how better soil physical and chemical data improve the performance of land surface models, and may improve atmospheric and climate modeling as well. We conclude with a discussion of the data needs of the land surface community.