Linking a spatially-explicit model of acacias to GIS and remotely-sensed data

Spatially-explicit and landscape-related simulation models are increasingly used in ecology, but are often criticized because their parameterization h as high data requirements. A frequently suggested approach to overcome this difficulty is the linkage of spatially-explicit or landscape-related model s with GIS (geographic information system) and remote-sensing technology. G IS can provide data on relevant landscape features, such as topography, and satellite images can be used to identify spatial vegetation distribution. In this paper, we use these techniques for simple, cost-inexpensive (in bot h time and money) parameterization based on readily-available GIS and remot ely-sensed data. We use a previously developed, spatially-explicit model of the population d ynamics of an Acacia species in the Negev desert of Israel (SAM, spatial Ac acia model) to investigate if model initialization (measurement of current tree distribution) can be obtained from readily-available satellite images using a radiometric vegetation index (NDVI, normalized difference vegetatio n index). Furthermore, we investigate the applicability and the advantages of using an explicit consideration of landscape features in the model based on topographic data from a GIS. Using a DEM (digital elevation model), we compare the wadi topography to the current tree distribution observed in th e field. We found that the readily-available GIS and remotely-sensed data are not su fficient to significantly support the parameterization and further developm ent of the model. We conclude that despite the possible benefit of linking spatially-explicit models with other techniques the advantage compared to d ata sampling in the field is limited by a possible mismatch of scales and t he dominant role of stochasticity that may override the relevance of certai n spatially-explicit information.