Assessing the suitability of different modeling techniques for meteorological forecasting on Chickpea wilt

Abstract

The daily climate data collected for Hisar district between November 1, 1977 and April 30, 2022, has been analyzed and presented in this study. The data set was divided into two parts: training and testing data. This study presents the results of ARIMA, state space, and seasonal Holt-Winters models fitted for maximum temperature, minimum temperature, relative humidity (M), relative humidity (E), bright sunshine hours, and rainfall. The models were trained on data spanning from November 1977 to April 2013. The top selected ARIMA models were chosen based on evaluation criteria, such as the Akaike information criterion, root mean squared error, mean absolute error, mean absolute percentage error, in-sample MSE, and the maximum number of significant coefficients. The state space models were selected based on minimum values of the Akaike information criterion (AIC), Bayesian information criterion (BIC), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), in-sample Mean Squared Error (MSE), and Mean Absolute PercentageError (MAPE). The seasonal Holt-Winters models were fitted with additive specifications and a period of 365. Global chickpea production is highly dependent on various biotic and abiotic stresses. One of the critical biotic stresses, Fusarium wilt, significantly limits chickpea productivity causing economic losses ranging from 10 to 40% in many countries and escalates to 100% when temperature and humidity are favourable. Weather forecasting is crucial in plant disease management as it helps to predict disease outbreaks by analyzing how weather conditions influence pathogen development and spread, allowing farmers to take timely preventative measures.