- English

Climate and rainfall are extremely non-linear and complicated phenomena, which require numerical modelling to simulate
for accurate prediction. We obtained local historical rainfall data for 12 meteorological stations in the Vietnamese
Mekong Delta (VMD) for the 45-year period 1978–2022, to predict annual rainfall trends. A statistical time series predicting
technique was used based on the autoregressive integrated moving average (ARIMA) model. We utilized the seasonal
ARIMA process of the form (p,1,q)(P,1,Q) for our study area. The best seasonal autoregressive integrated moving average
(SARIMA) models were then selected based on the autocorrelation function (ACF) and partial autocorrelation function
(PACF), the minimum values of Akaike Information Criterion (AIC) and the Schwarz Bayesian Information (SBC). The seasonal
autoregressive integrated moving average model with external regressors (SARIMAX) was discovered, and a series
of SARIMA models of various orders were estimated and diagnosed. To evaluate model fitting, we used the Nash–Sutcliffe
coefficient (Nash) and the root-mean-square error (RMSE). The study has shown that the SARIMA (1, 1, 1)(2, 1, 1)11
and SARIMA (1, 1, 1)(2, 1, 1)12 model were appropriate for analyzing and forecasting future rainfall patterns at particular
meteorological station in the VMD. The results showed the SARIMA model is more reliable and provides more accurate
projections than other commonly used statistical methods, notably interval forecasts. We found that interpretable and
reliable near-term location-specific rainfall predicts can be provided by the SARIMA-based statistical predicting model.
- English