Regional modeling of grassland biogeochemistry using GIS

We used an ecosystem coupled to a Geographic Information System (GIS) to simulate spatial variability in storage and fluxes of C and N within grassland ecosystems. The GIS contained information on driving variables required to run the model. These were soil texture, monthly precipitation and monthly minimum and maximum temperatures. We overlayed polygon maps of the above variables to produce a driving variable map of our study region. The final map had 768 polygons in 160 unique classes. The ecosystem model was run to a steady state for each class and NPP, soil organic matter (SOM), net N mineralization and trace gas emission were mapped back into the GIS for display. Variation in all of the above propertiees occurred within the region. NPP was primarily controlled by climate and patterns followed spatial variation in precipitation closely. Soil organic matter, in contrast, was controlled largely by soil texture within this climatic range. Error associated with aggregation within the study area showed that spatial averages over the study area could be used to drive simulations of NPP, which is linearly related to rainfall. More spatial detail had to be preserved for accurate simulation of SOM, which is nonlinearly related to texture. Mechanistic regional models form a valuable link between process studies and global models.