Seasonal ARIMA approach for forecasting whitefly population

Abstract

The whitefly (Bemisia tabaci Gennadius), a polyphagous pest, has significant economic importance globally due to its adverse impact on pulse production, particularly as the primary vector of yellow mosaic disease in Vigna crops. In the present investigation, historical data were used to consider pest population forecasting, aimed at mitigating losses due to pest infestations and facilitating enhanced crop yields. By incorporating a weekly seasonality parameter, the Seasonal Autoregressive Integrated Moving Average (SARIMA) time-series model was employed to model and forecast the whitefly pest population. The SARIMA model encompasses components of seasonal autoregression, seasonal differencing, and seasonal moving averages, in addition to the standard ARIMA components. The autocorrelation function (ACF) and partial autocorrelation function (PACF) were calculated, which facilitated the identification and development of Seasonal ARIMA models, aimed at elucidating the time series and assisting in future forecasts of the whitefly population. The model SARIMA (2,1,0) (0,0,2) [52] was found to be the most appropriate, based on the lowest values of σ² and the Akaike Information Criterion (AIC). Advance forecasting of pest outbreaks will help farmers and researchers take necessary preventive measures to manage and reduce pest infestation.