Capturing information about the nonlinear impact between droughts and vegetation dynamics based on nonlinear dynamical system theory

Understanding the response of vegetation dynamics to droughts at regional scales is of great importance to reveal the behavior mechanism of terrestrial ecosystems. Under the realm of statistical analysis, this study intends to detect the quantitative information about the impact from droughts to vegetation dynamics, from their respective time series. Two problems should be considered regarding the relationship between droughts and vegetation dynamics: the nonlinearity and the directionality. In the literatures of the relationship between droughts and vegetation dynamics, the mostly applied statistical technique is correlation-based analysis, which, however, is essentially a linear and directionless tool, and thus fails to unravel the nonlinear relationship or distinguish the direction of the impact. Therefore, conducted in Pearl River Basin (PRB), this study applies Convergent Cross Mapping (CCM) based on Nonlinear Dynamical System (NDS) theory to quantify the nonlinear impact from meteorological drought (SPI and SPEI), hydrological drought (SRI), and agricultural drought (SM) on vegetation dynamics (NDVI) respectively, and the reverse impacts from vegetation dynamics to different types of drought. Specifically, we found that both SPI and SPEI have a nonlinear impact on NDVI with similar strength around 0.60, while NDVI only has a reverse impact on SPEI around 0.20. Compared with SPI/SPEI, SRI has a weaker nonlinear impact on NDVI (0.44), but the reverse impact from on SRI is quite strong (0.76). SM also has a significant nonlinear impact on NDVI (0.62), and NDVI has a similar reverse impact on SM (0.58). In general, the outcomes of this study provide a nonlinear statistical perspective for the relationship between vegetation dynamics and droughts under the context of global warming, which helps to better understand their underlying physical mechanism and drought forecasting.