Causal Inferences on Drought Propagation Dynamics Using Extended Convergent Cross Mapping

Propagation of meteorological drought to agricultural drought depends on the magnitude and direction of causal relationship between precipitation and vegetation dynamics. However, most previous studies have relied on correlation analysis, which fails to capture directionality and may result in spurious links. This study questions the validity of using such correlations for understanding bidirectional drought propagation. To address this, we apply convergent cross mapping (CCM), a causal analysis method that can separate true causation from coincidence by checking whether the state of a driver variable can be reconstructed from the affected variable’s time series. CCM also allows for identifying time lags in the propagation process. Results showed that rainfall deficits have a stronger influence on vegetation than the reverse. Although rainfall generally affects vegetation across India, the central-eastern forest region (Chhattisgarh and Odisha) showed little to no causal link. Vegetation’s impact on rainfall was evident only in the Ganga basin. Moreover, vegetation typically responds to rainfall deficits after 3–6 months, whereas its influence on rainfall appears after a longer delay of 6–8 months.