A GCMs-based mathematic model for droughts prediction in the Haihe Basin, China: Multi-GCM Divide-Integration

Abstract. Recently, the skilful prediction of climate change has drawn high attention from the scientific community. Evidence has been reported the skill of prediction is not satisfactory for the magnitude of inter-annual precipitation and extreme precipitation, and at a smaller spatial scale as well. Based on observational data sets and outputs from the Global Climate Models (GCMs), this study aims at achieving a mathematical model, named multi-GCM divide-integration model (MGDI). The MGDI model is developed by hybridizing finer spatial scale and multi-linear regression model (MLRM) on five state-of-art of GCMs to improve the skills of five GCMs, which is applied to the second level of water resources regionalization in China. It is found that the performance after MGDI model correction has been improved significantly over that of individual GCMs. The errors between observation and simulation after correction (1.6 % ~ 4.4%) are within the margin of error (smaller than 5 %) and all of the varying trends in each second level of water resources regionalization were same. Furthermore, this study also used the MGDI model to predict the variation of precipitation and droughts at different spatial scale, including second level of water resources regionalization of China and the whole HHB, for the next 40 years. Predictions indicate the climate will gradually change from drying to wetting over the HHB wherein the trend of annual rainfall is 9.3 mm/10 a. The frequency of drought events will be decreasing as time goes on. The occurrence of mild and severe drought in the Luan River and Jidong Coastal, Tuhai majia River are higher than that in other regions, 9 and 8 respectively. These findings would provide scientific support for current water resources management and future drought-resisting planning of districts in China.