环境科学
气候变化
土壤碳
植被(病理学)
碳循环
滞后
生态系统
温带气候
大气科学
条件概率
水文学(农业)
全球变暖
喀斯特
固碳
气候学
地球大气中的二氧化碳
总有机碳
陆地生态系统
算法的概率分析
自行车
温室气体
荒漠化
生态水文学
铅(地质)
作者
Niu Shuai,Yecui Hu,Wang Yufan
标识
DOI:10.1186/s13021-026-00500-x
摘要
Global warming has intensified the frequency of droughts. Once droughts exceed a critical threshold, they pose a severe threat to vegetation growth and limit its carbon fixation capacity. However, current research still has an insufficient understanding of the probabilistic thresholds at which drought triggers the loss of vegetation carbon fixation (VCF) capacity under sustained greening, especially lacking a systematic comparison of the differences between karst areas (KA) and non-karst areas (NKA) under complex hydrogeological conditions. For this purpose, this study uses the Copula method and conditional probability to derive the VCF loss and its trigger thresholds under different drought scenarios from 1982 to 2022, and explores the key driving factors influencing these thresholds. The results show that: (1) Compared to non-karst areas (NKA), VCF in the KA of southern China is more sensitive to drought, with a higher response intensity and a shorter lag time. (2) The probability of VCF loss in KA is significantly higher than in NKA, and the drought trigger threshold is also higher. (3) The VCF thresholds in both KA and NKA are primarily driven by the “temperature (TEM)- moisture” coupling. However, KA show higher TEM sensitivity and dependence on deep soil water storage (SWC 3-4 ), while in NKA, the thresholds are influenced by multiple factors, including shallow soil water storage (SWC 1-2 ), atmospheric vapor pressure deficit (VPD), and soil organic carbon (SOC). From the perspective of probabilistic thresholds, the results reveal the hidden drought vulnerability of VCF capacity in humid greening regions, and can provide a quantitative basis for maintaining regional carbon sink stability, implementing zonal vegetation management, and promoting proactive drought risk prevention and control.
科研通智能强力驱动
Strongly Powered by AbleSci AI