Causal Pathways of Agricultural Drought Development in Semi-Arid Regions using NDVI-based Vegetation Stress Proxies

Abstract

Climate change is increasing the frequency and intensity of extreme events, including droughts, floods, and compound events. Agricultural drought remains difficult because it is shaped by multivariate interactions among climatic, hydrological, and vegetation-related variables. Although correlation and regression methods identify statistical associations, they have limited ability to distinguish directly from indirect relationships and conditional dependencies among multiple interacting time series. This study applies the Peter and Clark Momentary Conditional Independence plus (PCMCI+) causal discovery framework to identify the drivers, temporal dependencies, and lagged feedback mechanisms of agricultural drought in the semi-arid Bundelkhand region of India. ERA5-Land reanalysis rainfall data yielded a root mean square error of 42.367 mm, and a coefficient of determination of 0.864 against Indian Meteorological Department (IMD) rainfall observations for the Banda District, supporting its reliability for regional-scale analysis. ERA5-Land reanalysis data were thus to represent hydroclimatic and biophysical variables within PCMCI+. The causal results showed that temperature, humidity, evaporation, Normalized Difference Vegetation Index (NDVI), and Leaf Area Index exhibited stronger linkages with agricultural drought than precipitation and soil moisture, with most significant causal links occurring at short time lags of 1–3 months at p < 0.05. Across model configurations, precipitation and soil moisture showed weak or inconsistent causal links with vegetation indicators, suggesting partial hydroclimatic-vegetation decoupling at monthly scale. Temperature related variables emerged as dominant drivers of agricultural drought, highlighting the need to consider temperature stress alongside rainfall. Overall, this study reveals multi-lag climate vegetation interactions and land–atmosphere feedback governing agricultural drought in semi-arid regions.