Identifying Landscape Pattern Metrics for the Hani Terrace in Yunnan, China

Abstract: This study seeks to isolate a select group of landscape metrics particularly well-suited for describing the Hani Terrace in southwest of China. We examined the response of 47 landscape metrics to a large range of imagery grain sizes. Based on a correlation analysis, the original 47 metrics were placed into 21 groups such that all metrics within a group were strongly correlated with each other with a value of more than 0.9, and were represented by a single descriptor. Using these cross-sectional metrics in the context of principal components analysis, we found that five factors explained almost 93% of the total variation in the landscape pattern. The highest loadings for these five factors were the Splitting index (SPLIT), Patch area distribution (AREA_CV), Shannon's diversity index (SHDI), Euclidean nearest neighbor distance distribution (ENN_AM), and Total core area (TCA), respectively. Considering the real landscape, we added the Patch fractal dimension distribution (FRAC_MN) as the sixth land...