Abstract: The analysis of spatiotemporal variations and driving forces of regional vegetation carbon sink has implications for improving the carbon sequestration capacity of vegetation and contributing to national carbon neutrality. Based on MODIS remote sensing data and soil microbial respiration equations, the spatiotemporal variations and main driving factors of vegetation carbon sink capacity at different spatial scales in Ningxia were explored by using random forest, GeoDetector, and spatial partial correlation analysis. The results would provide scientific basis for ecological protection and high-quality development of the Yellow River Basin in Ningxia. The results showed that the vegetation carbon sink capacity in Ningxia increased overall (with southern part > central part > northern part) but significantly decreased locally in urban centers such as Yinchuan City, Shizuishan City, and Shapotou District in the northern irrigation area from 2000 to 2020. The vegetation carbon sink capacity in Ningxia was mainly influenced by natural factors and the interactions among them, including vegetation, vegetation-climate, and vegetation-elevation, followed by human factors (gross domestic product (GDP) and land use type). The nonlinear enhanced interaction between vegetation normalized index (NDVI) and precipitation was the main driving factor of variations in vegetation carbon sink capacity (84.8%). In terms of different regions, vegetation, population density, vegetation-population density, and vegetation-climate were the main driving forces of vegetation carbon sink capacity in the northern irrigation area, while vegetation, climate, vegetation-climate and vegetation-GDP were the main driving forces in the central arid zone and the southern mountainous area.

Key words: vegetation carbon sink, net ecosystem productivity, random forest, GeoDetector, Ningxia