Analysis and prediction of rainfall trends over Bangladesh using Mann–Kendall, Spearman’s rho tests and ARIMA model

Abstract

In this study, 60-year monthly rainfall data of Bangladesh were analysed to detect trends. Modified Mann–Kendall, Spearman’s rho tests and Sen’s slope estimators were applied to find the long-term annual, dry season and monthly trends. Sequential Mann–Kendall analysis was applied to detect the potential trend turning points. Spatial variations of the trends were examined using inverse distance weighting (IDW) interpolation. AutoRegressive integrated moving average (ARIMA) model was used for the country mean rainfall and for other two stations data which depicted the highest and the lowest trend in the Mann–Kendall and Spearman’s rho tests. Results showed that there is no significant trend in annual rainfall pattern except increasing trends for Cox’s Bazar, Khulna, Satkhira and decreasing trend for Srimagal areas. For the dry season, only Bogra area represented significant decreasing trend. Long-term monthly trends demonstrated a mixed pattern; both negative and positive changes were found from February to September. Comilla area showed a significant decreasing trend for consecutive 3 months while Rangpur and Khulna stations confirmed the significant rising trends for three different months in month-wise trends analysis. Rangpur station data gave a maximum increasing trend in April whereas a maximum decreasing trend was found in August for Comilla station. ARIMA models predict +3.26, +8.6 and −2.30 mm rainfall per year for the country, Cox’s Bazar and Srimangal areas, respectively. However, all the test results and predictions revealed a good agreement among them in the study.