Spatial and temporal patterns of vegetation carbon sources/sinks and driving factors in southeastern Xizang from 2000 to 2020

Net ecosystem productivity (NEP) is a crucial measurement for understanding ecosystem function and carbon cycling. On the basis of Moderate Resolution Imaging Spectroradiometer (MODIS) data, meteorological data, and topographic data, combined with the enhanced Carnegie Ames Stanford Approach (CASA) model, we analyzed the spatiotemporal trends and changes in the vegetation NEP in southeastern Xizang from 2000 to 2020. Additionally, we employed the GeoDetector model to identify the driving factors influencing the vegetation NEP. The results indicated that: (1) from 2000 to 2020, the annual vegetation NEP in southeastern Xizang exhibited a fluctuating increasing trend. The multiyear average vegetation NEP was 519.06 gC m^-2 a^-1, ranging from 11.23 to 1,333.40 gC m^-2 a^-1. The minimum and maximum values occurred in 2010 and 2015, respectively. (2) The spatial distribution pattern of the vegetation NEP revealed an overall trend of higher values in southern areas and lower values in northern areas, with higher values in eastern areas compared to western areas. The average area of the carbon source regions (NEP < 0) was approximately 70,119 km², whereas the average area of the carbon sink regions (NEP > 0) was approximately 82,017 km². Overall, the region exhibited a carbon sink characteristic. (3) Altitude, precipitation, and temperature were the primary drivers influencing the vegetation NEP. In regions at lower elevations and in the southern and eastern parts of the study area-where thermal and moisture conditions are relatively favorable-NEP values were generally higher. In contrast, NEP was markedly lower in the northern and high-elevation areas characterized by low temperatures and limited water availability. The interactions between any two of these factors had a greater impact on the vegetation NEP than the independent effect of any single factor did, highlighting a synergistic dual-factor enhancement effect.