Spatial and temporal evolutions of vegetation coverage in the Tarim River Basin and their responses to phenology

Abstract

Vegetation plays an important role in soil erosion, and its coverage as an important parameter appears in almost all soil erosion models, such as USLE (Universal Soil Loss Equation), RWEQ (Revised Wind Erosion Equation) and WEPS (Wind Erosion Prediction System) et al. In the past several decades, scientist have paid attention to photosynthetic vegetation coverage (PVC) based on Normalized Difference Vegetation Index (NDVI). However, in soil wind-eroded region, there is almost no PVC in the nongrowing season and is a large amount of nonphotosynthetic vegetation (NPVC) in the growing season. Because the effect of NPVC on soil erosion is the same as that of PVC, vegetation coverage (FVC) in soil erosion field would include PVC and NPVC. Based on field observation, NDVI and meteorological data et al., this paper studied temporal and spatial distribution of vegetation coverage in the Tarim River Basin, and its response to surface temperature (T) and rain (R). The main conclusions are as follows. (1) PVC based on NDVI, would result in an underestimation of FVC in the growing season with about 30%, which led to an overestimation of soil wind erosion with 12%, indicating great errors in the nongrowing season. (2) FVC had obvious region distributions. FVC with 10%∼40%, 40%∼70%, and>70% were distributed in the mountain of middle elevation and the edge of oasis, the cultivated of the oasis, and the alpine grass of the Kongque River Subbasin, respectively; FVC<10% was widely distributed. With increasing of FVC, its area proportion (y) decreased as an exponential function (y = 73.19e-0.038FVC). FVC in the Kongque River, Aksu River, and Weigan-Kuqa River subbasins was better than that in the Hotan River, Keriya River, Cheerchen River, and mainstream Tarim River subbasins. Since the restoration of the ecological environment in the Tarim River Basin in 2000, FVC of 0 ∼ 10% decreased, while that of 40%∼70% increased, and the other fluctuated. (3) There was a linear relationship (FVC = a1T + a2R + b, where a1, a2, and b were the fitting coefficients) among FVC, T, and R, although the coefficients would change in different subbasins or half-months.