Linking drought indicators and crop yields through causality and information transfer: a phenology-based analysis

Drought indicators, which are quantitative measurements of drought severity and duration, are used to monitor and predict the risk and effects of drought, particularly in relation to the sustainability of agriculture and water supplies. This research uses causal inference and information theory to discover the drought index, which is the most efficient indicator for agricultural productivity and a valuable metric in estimating and predicting crop yield. The causal connection between precipitation, maximum air temperature, drought indices and corn and soybean yield is ascertained by cross convergent mapping (CCM), while the information transfer between them is determined through transfer entropy (TE). This research is conducted on rainfed agricultural lands in Iowa, considering the phenological stages of crops. Based on the nonlinearity analysis conducted using S-map, it is determined that causality analysis could not be carried out using CCM due to the absence of nonlinearity in the soybean yield data. The results are intriguing as they uncover both the causal connection between corn yield and precipitation and maximum temperature indices. Based on the analysis, the drought indices with the strongest causal relationship to crop production are SPEI-9m and SPI-6m during the silking period, and SPI-9m and SPI-6m during the doughing period. Therefore, these indices may be considered as the most effective predictors in crop yield prediction models. The study highlights the need of considering phenological periods when estimating crop production, as the causal relationship between corn yield and drought indices differs for the two phenological periods.