Ecological Indicators (Jul 2021)
Multi-scale spatial correlation between vegetation index and terrain attributes in a small watershed of the upper Minjiang River
Abstract
The upstream of Minjiang River is an important ecological barrier and one of the main water sources of the Chengdu Plain and upstream Yangtze River. However the ecological balance is vulnerable in this area. Terrain undulation has a certain impact on vegetation coverage in mountain areas. However, it is not clear how the influence of topography on vegetation coverage changes with pixel scale. This study proposed a framework by combining Discrete Wavelet Transform (DWT) and Geographically Weighted Regression (GWR) to detect the relationships between topography and vegetation in a typical fragile ecological area. The satellite-derived Normalized Difference Vegetation Index (NDVI) and Digital Elevation Model (DEM) were combined to examine the terrain undulation impacts on vegetation coverage at different scales in a small watershed of the Upper Minjiang River. The study results showed that approximate components of NDVI had consistently strong spatial autocorrelation at multi-scale. Elevation and south-facing index of aspect were positively correlated with NDVI, whereas slope was negatively correlated with NDVI. The order of correlation between terrain attributes and NDVI was elevation > aspect > slope. The correlation of NDVI with elevation and slope increased as the decomposition scale increased, whereas the correlation between NDVI and south-facing index of aspect did not show obvious change rule with increasing scale. The spatial heterogeneity of impact degree of elevation decreased as the scale parameter increased, but slope and aspect did not show obviously smoothing effect as the scale parameter increased. The findings will help identifying the favorable topographic position for vegetation growth and its controlling mechanisms and provide a new insight for ecological restoration practices.
