Assessment of geostatistical and interpolation methods for mapping forest dieback intensity in Zagros forests

Abstract

During recent years, oak decline has been widely spread across Brant's oak (Quercus Brantii Lindl.) stands in the Zagros Mountains, Western Iran, which caused large-area forest dieback in several sites. Mapping the intensity and spatial distribution of forest dieback is essential for developing management and control strategies. This study evaluated a range of geostatistical and interpolation methods to explore the spatial structure and provide area-based maps of the intensity of forest dieback across a representative test site - Ilam Province - that was severely affected by Oak decline. The geostatistical analysis provided in-depth measures of the spatial structure amongst the selective sampling units (120 quadratic sample plots of 1200 m2), which eventually resulted in an area-based maps of dieback intensity. The accuracy of the applied methods was assessed by mean error percentage (%ME), root mean squared error percentage (%RMSE) and coefficient of determination (R2). Results showed moderate spatial structure within the sampling units. Moreover, cokriging (associated with soil humidity and aspect as independent variables) approach resulted in the highest accuracy, followed by two other methods of kriging and Radial Basis Function. Results suggested that cokriging can accurately estimate the intensity of dieback and its spatial distribution in the study area. According to this, an average dieback intensity of 18.12 % was estimated within the study area.