Modeling the Changes in Water Balance Components of Highly Irrigated Western Part of Bangladesh
Abstract. The objectives of the study are to explore the changes in water balance components (WBC) by co-utilizing discrete wavelet transformation (DWT) and different forms of Mann–Kendal (MK) test; and to develop wavelet autoregressive moving average (ARIMA) models for forecasting the WBC. Trend test results reveal that the most of the trends (about 73 %) identified in potential evapotranspiration (PET) show decreasing tendency during the hydrological year 1981–82 to 2012–13 in the western part of Bangladesh, however most of the changes (about 82 %) are insignificant at 5 % significant level. Actual evapotranspiration (AET), annual deficit and annual surplus also show the almost similar tendency. Rainfall and temperature show increasing trends, but WBC show inverse of this tendency and suggesting that traditional concept of changes in PET associated with changes in temperature cannot explain the changes in WBC. Moreover, it is found that generally 8-years (D3) to 16-years (D4) periodic components are the effective components and are responsible for trends found in original data of WBC in western part of Bangladesh. Wavelet denoising of WBC time series has been done to improve the performance of models as actual data affected by noise and show unsatisfactory performances. The quality of denoised data has been ensured by relevant statistical analysis. Performance of wavelet ARIMA models have been assessed by Nash–Sutcliffe Efficiency (NSE) coefficient and coefficient of determination (R2). The obtained results indicate that performances of wavelet ARIMA models of WBC are acceptable to very good and clearly demonstrate the advantages of denoising over actual data. The models validation results reveal that the forecasted values are very close to actual values with acceptable mean percentage error and residuals also follow normally distribution. Performances and validation results indicate that models can be used for short term forecasting of WBC. Further studies on different combinations of wavelet analysis would be facilitated to develop better models for WBC in context of climate change and findings of study can be used to improve water resources management in highly irrigated western part of Bangladesh.