Análise dos efeitos do ENOS sobre o clima e a produção agrícola no Brasil: um estudo de caso dos estados de Goiás e Rio Grande do Sul

Abstract

Understanding the effects of the El Niño-Southern Oscillation (ENSO) is crucial for Brazilian agriculture, given its influence on production risks. This study aimed to evaluate the individual importance of climatic variables and municipalities in model performance, establishing evidence of the phenomenon’s effect on the high-yield class based on SHAP values. Fixed agricultural production cycles were analyzed, covering the period between planting and harvesting, lasting ten months in the state of Rio Grande do Sul and eight months in Goiás, for rice, bean, and soybean crops. Cycles from 1961 to 2019 were grouped according to El Niño, La Niña, and Neutral phases using the functional k-means method. The study focused on the period from 1990 to 2019, totaling 29 agricultural cycles, and tested the hypothesis that the three phases of the phenomenon produce similar effects through Functional Analysis of Variance (FANOVA) at a 5% significance level. This analysis considered the variables of maximum temperature, minimum temperature, and accumulated weekly precipitation. The results showed significant evidence of climatic influence only in Rio Grande do Sul, which motivated the performance of multiple tests to identify which phases differ from each other. The impact of the phenomenon on crop productivity was also evaluated using the Random Forest algorithm, applying yield classification categorized by classical k-means. The main results indicate that in Rio Grande do Sul, the influence of the El Niño-Southern Oscillation on climatic variables is marked, reflecting differences among the analyzed phases. SHAP values showed that municipal factors predominate in model performance, while climatic variables have lesser but still relevant contribution. It is concluded that the phenomenon exerts significant climatic and productive impacts in Rio Grande do Sul and that, despite its complexity, municipal regional aspects play a fundamental role in forecasting agricultural yield. These results contribute to the development of adaptive strategies in response to climate variability associated with the El Niño-Southern Oscillation.