Responses of landscape metrics to altering grain size in the Three Gorges Reservoir landscape in China

Abstract

Quantifying the response of landscape metrics to an altering observation scale is crucial to understanding environmental changes and managing ecosystem services. Whereas the scaling behaviors of landscape metrics in spatial heterogeneity analysis have been well identified by previous research, there remains a need to examine these effects in areas undergoing rapid change. Here, we aim to reveal the landscape scale effect in the Three Gorges Reservoir (TGR) area, China, using a case study on Zigui County. We applied a suite of common landscape metrics (12 indices at the class level and 17 indices at the landscape level) to characterize the landscape pattern and examine the response of the metrics to altering grain size using a series of land-use/land-cover data with gradient resolutions. The results reveal that significant scale effects exist in most pattern metrics in the TGR landscape. In addition, the different responses to the altering grain size occurred with different landscape metrics and various land-use/land-cover types. With respect to changing grain size, all of the selected pattern metrics at the landscape level displayed high or medium sensitivity in response to changing grain size except the Fractal Dimension Index and the landscape-diversity indices. The behavior of the metrics in response to altering grain size can be grouped into four types (Type 1, Type 2, Type 3, and Type 4). The class-level metrics with high sensitivity were Mean Patch Size, the Contiguity Index, the Euclidean Nearest-Neighbor Distance, the Perimeter-Area Ratio, and Patch Density for all land-use/land-cover types, whereas low sensitivities were detected in the response of the Fractal Dimension Index and the Largest Patch Index. Based on the response to the altering resolution of input data, the class-level metrics could be grouped into three types (Type a, Type b, and Type c). Considering the scaling behavior of landscape metrics, we suggest using a set of suitable remote-sensing images to quantify the landscape pattern in the TGR landscape and similar areas.