Regional-scale forest ecosystem modeling: database development, model predictions and validation using a Geographic Information System

Regional-scale forest ecosystem models can provide insight into how forest hydrology and productivity differ in response to variations in climate and soil conditions. However, the databases necessary to define and validate these models are difficult to compile and utilize. The use of a Geographic Information System (GIS) can greatly simplify problems in database management. As an example of regional-scal e ecosystem model database development and model validation, the Arc-Info GIS and the PnET-IIS forest ecosystem model were used to predict and validate annual drainage and net primary productivity (NPP) on a 0.5° x 0.5° grid (approximately 50 x 75 km) for southern pine forests in the state of Georgia, USA. PnET-IIS is a lump-sum physiological model which used historic climate data from 1951 to 1984, along with soil water-holding capacity and species-specific vegetation charac- teristics. Annual predictions of drainage were well correlated with measured United States Geological Survey (USGS) drainage (r = 0.87, p < 0.0001), and predicted NPP was related with Forest Inventory Assessment (FIA) growth data collected across the state (r = 0.84, p < 0.0001). This study demonstrated the utility of a GIS in broad-scale ecosystem modeling and suggests that the need for model/GIS inter- facing in future research will continue to increase.