Combining BPANN and wavelet analysis to simulate hydro-climatic processes-a case study of the Kaidu River, North-west China

Using the hydrological and meteorological data in the Kaidu River Basin during 1957-2008, we simulated the hydro-climatic process by back-propagation artificial neural network (BPANN) based on wavelet analysis (WA), and then compared the simulated results with those from a multiple linear regression (MLR). The results show that the variation of runoff responded to regional climate change. The annual runoff (AR) was mainly affected by annual average temperature (AAT) and annual precipitation (AP), which revealed different variation patterns at five time scales. At the time scale of 32-years, AR presented a monotonically increasing trend with the similar trend of AAT and AP. But at the 2-year, 4-year, 8-year, and 16-year time-scale, AR presented nonlinear variation with fluctuations of AAT and AP. Both MLR and BPANN successfully simulated the hydroclimatic process based on WA at each time scale, but the simulated effect from BPANN is better than that from MLR.