Mapping Dieback Intensity Distribution in Zagros Oak Forests Using Geo-statistics and Artificial Neural Network

The first and most important issue in forest drought management is knowledge of the location and severity of forest decline. In this regard, we used geostatistics and artificial neural network methods to map the dieback intensity of oak forests in the  Ilam province, Iran. We used a systematic random sampling with a 250 × 200 m grid to establish 100 plots, each covering 1200 m2. The percentage of the declined trees in each plot was measured and recorded. Also, a composite soil sample was extracted from the center and the four corners of each plot in order to determine their physical and chemical properties. After examining the normality of the data, the dieback intensity map was made using interpolation methods and the artificial neural network. The results showed  that the best method for dieback intensity estimation was the artificial neural network with an accuracy of 85 %, by using the multilayer perceptron algorithm. Oak decline was found to be mainly related to the slope, soil moisture, soil organic content and soil bulk density.