Understanding of the relationship between vegetation change and physical geographic factors based on Geographical Detector

摘要

In order to analyze the effect of physical geographic factors on vegetation change in arid and semi-arid ecosystems, assess the relative role of individual physical geographic factors and the interaction between factors on vegetation changes quantitatively, this study takes the Xinjiang area as an example, uses the GIS spatial analysis technology and Geographical Detector model based on the analysis of variance to analysis the influence of physical geographic factors on the vegetation quantitatively. First of all, the spatial-temporal variations of vegetation in Xinjiang area over the last 30 years were analyzed using 1982-2011 GIMMS NDVI3g data as the indicator of vegetation activity. Secondly, the effects of mean annual precipitation, mean annual temperature, sunshine duration, mean annual wind velocity, DEM, slope and aspect, soil type and vegetation type were selected as potential physical geographic factors. Finally, the influence of physical geographic factors on vegetation change in Xinjiang area was analyzed using the Geographical Detector model. The results show that: (1) the annual coverage of vegetation in Xinjiang area was gradually increasing in 1982-2011 years (linear rate 0.0017/a, P=0.000). (2) the area of vegetation improvement was greater than the area of vegetation degradation. The area of vegetation improvement was mainly distributed in the northern part of the Tianshan Mountains and the Tarim Watershed, the vegetation degradation region was mainly distributed in the southern and Northeast part of Xinjiang. (3) precipitation, soil and vegetation types had the greatest influence on NDVI, followed by temperature, sunshine duration and DEM, and the other factors had little effect. (4) DEM enhanced the effect of soil type on NDVI, and sunshine duration and DEM enhanced all the effect of temperature on NDVI. So, sunshine duration and DEM can be used as the auxiliary indicator in the vegetation growth monitoring. Our results brought new insights on the monitoring of vegetation dynamic and could provide a basic reference for the local inhabitants and policy-makers to restore degraded arid and semi-arid ecosystems.