Land-use forecasting using a Bayesian network

Abstract

The article proposes a forecasting scenario to be used for improving the quality of land of one of the most developed industrial-agrarian regions of the European Russia – the Tula region. The purpose of the work is land-use forecasting using the Bayesian networks analysis. The networks were built by considering the most significant indicators of both the ecological state and the negative impacts on lands and, in addition, the potential indicators of the improved lands. As a result of the analysis, characteristic high rates of factor loadings were established determining the need for land management. One of the land management measures proposed within the project employed soy cultivation on 5% of the farmlands surface and of pennycress on 5% of the rest of lands for improving the ecological state and decreasing the need for undertaking land management measures from 91% to 83.5%.