基于CEEMD和BP神经网络的鄱阳湖流域旱涝长期预测模型研究
2017-10-12万智巍蒋梅鑫贾玉连洪祎君
万智巍,蒋梅鑫*,贾玉连,洪祎君,章 鸣
(1.江西师范大学 地理与环境学院 鄱阳湖湿地与流域研究教育部重点实验室,江西 南昌 330022;2.中国科学院 地理科学与资源研究所 陆地表层格局与模拟重点实验室,北京 100101)
基于CEEMD和BP神经网络的鄱阳湖流域旱涝长期预测模型研究
万智巍1,蒋梅鑫1*,贾玉连1,洪祎君2,章 鸣1
(1.江西师范大学 地理与环境学院 鄱阳湖湿地与流域研究教育部重点实验室,江西 南昌 330022;2.中国科学院 地理科学与资源研究所 陆地表层格局与模拟重点实验室,北京 100101)
基于鄱阳湖流域1470~2014年的原始旱涝等级序列,利用最新的完备集合经验模态分解(Complete Ensemble Empirical Mode Decomposition, CEEMD)和BP神经网络(Back Propagation Neural Network),构建了鄱阳湖流域旱涝的长期预测CEEMD-BP模型。结果表明:与EEMD相比,CEEMD对原始数据进行平稳化处理的效果更好,能更有效地提取原始数据中隐含的周期信号和长期趋势;原始数据经CEEMD分解后得到若干个本征模函数(Intrinsic Mode Function, IMF)序列,BP神经网络可以较好地拟合或预测这些IMF序列;CEEMD-BP模型对鄱阳湖流域1985~2014年旱涝等级序列的预测精度优于单一BP神经网络的。CEEMD-BP模型对2015~2064年的长期预测显示,未来50年鄱阳湖流域的旱涝指数有先上升后下降的趋势。
旱涝;长期预测模型;CEEMD;BP神经网络;鄱阳湖流域
Abstract: According to the original drought and flood grade sequences of the Poyang Lake basin from 1470 to 2014, the author constructed an improved long-term prediction model (CEEMD-BP model) for the drought and flood in this basin by using the latest Complete Ensemble Empirical Mode Decomposition (CEEMD) and Back Propagation (BP) neural network. The results showed that CEEMD was more effective than EEMD in the smooth processing of original data, and it could more effectively extract the implied periodic signal and long-term trend from the original data. Several IMF (Intrinsic Mode Function) sequences were obtained from the decomposition of original data by CEEMD, and these IMF sequences could be better simulated or predicted by BP neural network. The predictive accuracy of CEEMD-BP model for the drought and flood grade sequences of the Poyang Lake basin from 1985 to 2014 was better than that of single BP neural network. The CEEMD-BP model predicts that the drought and flood index in the Poyang Lake basin will increase first and then decrease during 2015~2064.
Keywords: Drought and flood; Long-term prediction model; CEEMD; BP neural network; Poyang Lake basin
近年来全球范围内的极端气候水文事件屡有发生,IPCC第5次评估报告指出极端气候水文事件发生的范围、强度、频率已有显著改变[1];并且由于全球气候变暖加剧,未来或将有更频繁更严重的洪水、干旱等极端事件发生[2-3]。旱涝灾害带来的不利影响直接制约了社会、经济发展和人民生命安全,已引起各国政府和国际组织的高度重视,成为当前学术界研究的热点领域[4-5]。……
