Bayesian modeling for forest cover dynamics in Shimla District.

Decision making in land use planning needs understanding about the pattern of changes. The current study aims to analyse and predict the land use and land cover change, with the focus on forests, in Shimla District using Bayesian model. Population growth, agricultural-horticulture demands, tourism growth are putting pressure on the valuable forest ecosystem and natural resources of the district. In this study, land cover maps were prepared for the periods 1970s, 1980s and 1990s using remote sensing data. The actual positive changes (i.e., increase in forest) and negative changes (i.e., decrease in forest) derived from the time-series land cover maps were used as apriori evidence in the Bayesian model to derive the statistical weights for various environmental parameters. The environmental parameters were analysed under 4 major group of factors i.e., topographic, land use, landscape, land-water. The probabilistic contribution (i.e., weight) of each attribute under each map was utilised within the weighted summation model to derive spatial maps of potential positive and negative change. The accuracy of the model was validated using actual change maps. Accuracy of the model was 85% for the positive change and 80% for the negative change. The resultant predicted maps of positive and negative change were overlaid together and potential zones of conservation and afforestation were identified.