Desertification Assessment Using Bayesian Belief Networks

Desertification is a permanent loss of land productivity due to change in the ecological parameters and management conditions andcan take place in various climates. Appropriate management practices to prevent desertification should be applied and decision support tools can also assist managers to select the best management options. It is a difficult task for mangers to take decision when they are uncertain about the outcomes of various environmental conditions and their management practices and when there are uncertainties in understanding the ecosystem behavior under stress. Bayesian Belief Network (BBN) models provide an appropriate tool to accommodate uncertainty in the prediction of these systems. The BBN models graphically represent system variables and their cause-and-effect relationships. These models can be used for scenario and diagnostic analyses and can be applied in areas where empirical data are scarce through utilizing expert knowledge. In this paper, the capability of BBN models in desertification assessment and management is discussed using an example from a Chinese desert.