Hierarchical Model of Landscape Mapping. Case Study: Multiscale Mapping on Natural Ecosystem Baluran National Park

Abstract

We investigated the minimum field size and sample density needed for collecting soil samples for geostatistical analyses of field-scale spatial variability of electrical conductivity of the saturation extract (ECe), in 3 salt-affected soil classes (percentage salt crusts: very severely = class 1; severely = class 2, and moderately = class 3). In areas of each class, 2 representative sites of 50 × 50 m2 were selected (n=6). At each site, 100 soil samples were collected using stratified, systematic, unaligned sampling, then we analyzed for ECe. These data were rearranged into 6 datasets representing different field sizes and/or sample density for semivariogram analysis and kriging interpolation. Through comparisons of the datasets, it was found that the field size should be ≥ 40 × 40 m2, and the sample density ≥ 1: 10 × 10 m2. This was particularly true for areas of classes 1 and 3; however, because of the extremely high variation of soil ECe in class 2 areas, accuracy of the relevant interpolated (kriged) maps was relatively poor, and further study is required to improve the method for areas of extreme variability.