Early warning of drought-induced vegetation stress using multiple satellite-based ecological indicators
Ying Wang, Yanan Chen, Jianguang Wen, Chaoyang Wu, Wei Zhou, Lei Han, Xuguang Tang
Abstract
• The evolution processes of two extreme droughts in 2011 and 2022 across Southwest China were analyzed. • We examined the early responses of four indicators including GOSIF, Φ f , NIRv and NDVI to droughts. • GOSIF was identified as the most sensitive and responded early to drought, followed by NIRv. • In spatial terms, GOSIF also exhibited the strongest correlations with drought intensity. Droughts have posed, and continue to pose, severe risks to terrestrial ecosystems. Particularly against the backdrop of global climate change, the intensity and frequency of extreme droughts are expected to further aggravate. However, a significant gap persists in early drought warning for vegetation monitoring. Therefore, this study examined the spatial and temporal dynamics of two summer drought events happened in Southwest China in 2011 and 2022, and analyzed the early responses of four ecological indicators including global Orbiting Carbon Observatory-2 (OCO-2) SIF dataset (GOSIF), the leaf-scale fluorescence yield ( Φ f ), the near-infrared reflectance of vegetation (NIRv) and the normalized difference vegetation index (NDVI) to drought extremes. All these indicators successfully captured the drought-induced vegetation stress, but as a proxy for vegetation photosynthesis, GOSIF was the most sensitive. Specifically, during the 2022 drought, GOSIF fell below the baseline year as early as day of year (DOY) 193, whereas NIRv and NDVI began at DOY 201, and Φ f lagged severely. Similar behaviour was also found in the drought period of 2011. Overall, compared to the baseline year, GOSIF, Φ f , NIRv and NDVI decreased by 96.93 %, 54.11 %, 43.92 % and 17.03 % in 2011, and reduced by 70.00 %, 42.01 %, 48.74 % and 19.53 % in 2022, respectively. During the past two decades, GOSIF exhibited the strongest correlation with drought intensity ( r = 0.880, p < 0.05), followed by NIRv ( r = 0.875, p < 0.05) and NDVI ( r = 0.871, p < 0.05), and Φ f was the weakest ( r = 0.432, p > 0.05). Spatially, the proportion of areas where the correlations exceeded 0.6 by GOSIF and NIRv were 42.39 % and 39.32 %, respectively. In summary, this study demonstrated that global re-constructed GOSIF possesses considerable potential as an early warning indicator for vegetation drought.