Quantification of driving factors on NDVI in oasis-desert ecotone using geographical detector method

Abstract

Within oasis-desert ecotone regions, the normalized difference vegetation index (NDVI) is an important parameter for evaluating the growth of vegetation. An accurate quantitative study between NDVI and environmental and anthropogenic factors is critical for understand the driving factors of vegetation growth in oasis-desert ecotone. In 2016, four periods Landsat 8 OLI_TIRS images, relevant climatological parameters data (air temperature, air relative humidity, wind velocity and accumulated temperature), land cover type data and soil data were selected as proxies. In order to quantify the explanatory power for NDVI spatial and temporal distribution in the southern edge of Dunhuang City and northern side of the Mingsha Mountain, the geographical detector model was used to explain the potential influences of factors versus the spatial distribution of NDVI, and each explanatory variable’s relative importance can be calculated. The factor detector results disclose that the spatial distribution of NDVI is primarily dominated by land cover type. The risk detector results show that, high NDVI region is located within woodland. The mean value of NDVI displays an increase and then decrease trend with air temperature increase. With the increase of wind velocity and decrease of air relative humidity, the NDVI value shows a decrease trend. The interactive q values between the two factors are higher than any q value of separated factors. Results also indicate that the strongest interactive effects of NDVI are different in distinct seasons. Consequently, anthropogenic activity is more important than environmental factors on NDVI in oasis-desert ecotone. We also demonstrate that air relative humidity rather than air temperature have played a greater role in NDVI spatial distribution.