A data-driven framework for assessing climatic impact drivers in the context of food security

Understanding how physical climate-related hazards affect food production requires transforming climate data into relevant information for regional risk assessment. Data-driven methods can bridge this gap; however, more development must be done to create interpretable models, emphasizing regions lacking data availability. The main objective of this article was to evaluate the impact of climate risks on food security. We adopted the climatic impact driver (CID) approach proposed by Working Group I (WGI) in the Sixth Assessment Report (AR6) of the Intergovernmental Panel on Climate Change (IPCC). In this study, we applied the CID framework using a random forest model in a bootstrapping experiment to identify the most influential indices driving crop yield losses. We also used SHapley Additive exPlanations (SHAP) with the random forest model for explanatory analysis, enabling us to pinpoint critical thresholds for these indices-thresholds that, when exceeded, significantly increase the probability of impact. Additionally, we investigated the effects of two CID types (heat and cold and wet and dry) represented by categories of climate extreme indices on crop yields, with a particular focus on maize and soybeans in key agricultural municipalities in Brazil. We found that mean precipitation is a highly relevant CID. However, there is a window in which crops are more vulnerable to a precipitation deficit. In many regions of Brazil, for example, soybeans face an increased risk of yield losses when precipitation falls below 100 mm per month in December, January and February - marking the end of the growing season in those areas. Nevertheless, including climate means remains highly relevant and recommended for studying the impact of climate risk on agriculture. Our findings contribute to a growing body of knowledge critical for informed decision-making, policy development and adaptive strategies in response to climate change and its impact on agriculture.