Avaliação das variáveis climáticas: precipitação, temperatura e umidade relativa do ar da cidade de Franca-SP por meio de análise de séries temporais

Abstract

Understanding how climatic conditions interfere with agriculture is fundamental, especially for coffee cultivation, one of the main economic activities in the city of Franca-SP, recognized for its relevance in coffee production in Brazil. In this context, the objective of this study was to adjust time series models capable of predicting climatic variables such as monthly precipitation, average temperature, and relative air humidity, contributing to assisting in the decision-making of coffee growers and rural producers. To this end, SARIMA-type models were adjusted using profitable monetary data between January 2011 and May 2025, planned for the period from January to May 2025. The model selection criteria were based on the lowest values of the Akaike Information Criterion (AIC) and the Bayesian Schwarz Criterion (BIC). Predictive per formance was evaluated using the metrics RMSE (Root Mean Squared Error), MAE (Mean Absolute Error), MASE (Mean Absolute Scaled Error), and ME (Mean Error). The results in dicated distinct performances for each climatic variation. The total monthly exception showed reasonable performance with the SARIMA (0,0,1)(0,1,1)12 model, possibly due to the high variability characteristic of this variable. The average monthly temperature showed the best performance among the variables tested with the SARIMA (1,0,1)(0,1,1)12 model, demonstra ting high precision in the isolated variables. Relative air humidity showed superior results with the SARIMA (0,0,1)(0,1,1)12 model, although with superior performance compared to tempe rature. Overall, the results indicate that models from the SARIMA family can be useful tools for forecasting climate variations, potentially assisting in planning and decision-making related to coffee crop management.