Making objective forest stand maps of mixed managed forest with spatial interpolation and multi-criteria decision analysis

Abstract

Abstract: The spatial interpolation and multi-criteria decision analysis (MCDA) capabilities of geographic information systems have the potential to create new approaches to forest management. In this study of heterogeneously structured stand maps, the potential use of the regularized spline with tension (RST) interpolation method and of ELECTRE TRI MCDA was investigated. For each species and diameter class, one map of the predicted volume per ha was produced with the RST method. The map used data from a total of 1050 circular sample plots. By repeating the same process for the eight species occurring in the study area, 31 volume maps were produced. The accuracy of these prediction maps was calculated at the pixel (20 x 20 m) level and at the area level (per ha). An accuracy of greater than or equal to 97% was achieved at the pixel level, whereas a minimum accuracy of 86% was achieved for the area-based calculations. In addition, these 31 volume maps were compared with the management report results obtained from the government institute responsible for defining management plans. These comparisons were performed for the total volume of all species with volume ratios greater than 1%. The comparisons showed 21, 14, 4, and 2 % accuracies for Calabrian pine, Oriental beech, black pine and oak species, respectively. Following interpolation, these 31 maps were geo-computed, and a volume-based stand map was produced. The 890 different mixture variations resulting from combinations of the volume of species composition and the stand diameter class in these maps were classified according to expert knowledge. In this classification process, ELECTRE TRI MCDA was used to benefit from the capabilities provided by geographic information systems. Finally, ELECTRE TRI was used to reduce the 890 different mixture combinations to 70 stand type classes.