Global environment changes rapidly alter regional hydrothermal conditions, which undoubtedly affects the spatiotemporal dynamics of vegetation, especially in arid and semi-arid areas. However, identifying and quantifying the dynamic evolution and driving factors of vegetation greenness under the changing environment are still a challenge. In this study, gradual trend analysis was applied to calculate the overall spatiotemporal trend of the normalized difference vegetation index (NDVI) time series of Xinjiang province in China, the abrupt change analysis was used to detect the timing of breakpoint and trend shift, and two machine learning methods (boosted regression tree and random forest) were used to quantify the key factors of vegetation change and their relative contribution rate. The results have shown that vegetation has experienced overall recovery over the past 20 years in Xinjiang, and greenness increased at a rate of 17.83 10-4 year-1. Cropland, grassland, and sparse vegetation were the main biome types where vegetation restoration is happening. Nearly 10% of the pixels (about 166000 km2) were detected to have breakpoints from 2004 to 2016 of the monthly NDVI, and most of the breakpoints were concentrated in the ecotone of various biomes. CO2 concentration was the most prevalent environmental factor to increase vegetation greenness, because continuous emission of CO2 greatly enhanced the fertilization effect, further promoted vegetation growth. Besides, cropland expansion and desertification control were the vital anthropogenic factors to vegetation turning "green" in Xinjiang, and most areas under anthropogenic were mainly in oasis areas. These findings provide new insights and measures for the regional response strategies and terrestrial ecosystem protection.
Keywords: Abrupt change; Environmental factors; Gradual change; Land-use management; Vegetation greenness; Xinjiang.
© 2021. The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.