收稿：2018-01-23，
录用：2018-4-24，
纸质出版：2019-09
移动端阅览
湖泊流域汇水径流过程的模拟预测是一种复杂系统中的时间序列分析问题。模型选择上，现有的机理模型法与辨识模型法各有利弊。同时，现有的模型多采用静态数据驱动模拟，不能有效利用传感网实时观测数据来改善模拟不确定性的问题。本文基于深度循环神经网络技术，提出一种适应动态数据驱动的模式，可融合遥感数据与原位传感器站点数据的DTSM(Dynamic Data Driven Time Series Model)时序模拟预测模型，并在观测值与数值模拟之间建立了一种能动态反馈、自适应调整的模拟框架，解决了传统辨识模型法对时序信息挖掘较弱导致模拟精度较低的问题。通过在鄱阳湖多个子流域入湖径流的案例中验证，显示静态数据驱动模式下，以不同数据源作为输入模拟时，本文DTSM模型的纳希效率系数
E
ns
精度比机理模型提高10个百分点以上；相比静态模式，动态数据驱动模式的模拟精度有进一步提高，尤其是对于静态模式精度较低的流域，提高更为明显。
Performing rainfall-runoff simulation and prediction in a lake basin is a time-series analysis problem in complex systems. Existing mechanistic and identification methods for model selection have advantages and disadvantages. Mechanism model has a clear physical explanation
but it requires professional data for support and the model-solving process is complicated. Identification model is flexible and its solution is simple
but it has difficulty building a universal model and the accuracy of the model is low. Existing simulation models are also based mostly on static data
which cannot effectively use real-time observation data from sensor networks to improve simulation uncertainty. This study aims to improve the problem of traditional identification models being incapable of effectively using timing information
which results in low simulation accuracy. Moreover
it establishes a simulation and simulation framework for dynamic feedback and adaptive adjustment between observation and numerical simulation. This research proposes a dynamic data-driven model based on deep recurrent neural network
named dynamic data-driven time sequence model
which consists of a multilayered long short-term memory loop body and a fully connected layer. The proposed model incorporates runoff remote sensing rainfall data and ground station observation rainfall data as static data input and recent ground station actual runoff observation data as dynamic data input to simulate catchment runoff process. Several cases of multiple sub-river runoff into Poyang Lake indicate that in static data-driven mode
with TRMM_3B42_V7 precipitation as input
the ENS accuracy of DTSM is 10 percentage points higher or more than that of the mechanism model. The cases also indicate that in static data-driven mode
with the fusion precipitation from TRMM_3B42_V7 and ground station as input
the ENS accuracy of DTSM is 29 percentage points higher or more than that of the mechanism model. Lastly
the dynamic data-driven model can further improve the accuracy of simulation compared with the static-driven model
and the improvement is substantial in the basin with lower accuracy in static data-driven mode. The model based on deep recurrent neural network can effectively extract the timing information from the data. The dynamic data-driven model can make adaptive adjustments to improve simulation uncertainty. Based on the above two reasons
the DTSM proposed in this paper can achieve the same or a better simulation accuracy than existing representative mechanism models. At the same time
DTSM is flexible and the solving process is simple.
Carlson R F, MacCormick A J A and Watts D G. 1970. Application of linear random models to four annual streamflow series. Water Resources Research, 6(4): 1070–1078
程甜甜, 李赛, 张兴刚, 张永涛. 2016. 山东药乡小流域降雨径流关系研究. 水土保持学报, 30(2): 34–37, 43
Cheng T T, Li S, Zhang X G and Zhang Y T. 2016. Study on the relationship between rainfall and runoff in Yaoxiang small watershed of Shandong province. Journal of Soil and Water Conservation, 30(2): 34–37, 43
费明哲. 2015. 新一代TRMM V7降水产品在鄱阳湖流域精度分析及应用研究. 南京: 南京邮电大学
Fei M Z. 2015. Accuracy Analysis and Application of New Generation TRMM V7 Precipitation Product in Poyang Lake Basin. Nanjing: Nanjing University of Posts and Telecommunications
Gers F A, Schraudolph N N and Schmidhuber J. 2002. Learning precise timing with LSTM recurrent networks. Journal of Machine Learning Research, 3(1): 115–143
Google. 2017. Tensorflow Chinese community[EB/OL]. http://www.tensorfly.cn/.2018.10
Graves A, Fernández S and Schmidhuber J. 2005. Bidirectional LSTM networks for improved phoneme classification and recognition//Proceedings of the 15th International Conference Artificial Neural Networks: Formal MODELS and Their Applications - ICANN 2005. Warsaw, Poland: Springer: 799-804
Graves A. 2012. Sequence transduction with recurrent neural networks. Computer Science, 58(3): 235–242
郭俊, 周建中, 张勇传, 宋利祥, 刘强. 2010. 基于改进支持向量机回归的日径流预测模型. 水力发电, 36(3): 12–15
Guo J, Zhou J Z, Zhang Y C, Song L X and Liu Q. 2010. Daily runoff forecast based on improved support vector machine regression model. Water Power, 36(3): 12–15
Hammer C L and Small G W. 2000. Artificial neural networks for the automated detection of trichloroethylene by passive fourier transform infrared spectromrtry. Analytical Chemistry, 72(7): 1680–1689
Hochreiter S and Schmidhuber J. 1997. Long short-term memory. Neural Computation, 9(8): 1735–1780
胡平. 2004. 降雨径流组合预测理论及其应用研究. 武汉: 华中科技大学
Hu P. 2004. Application Research on Rainfall-Runoff Combinatorial Forecasting Theory. Wuhan: Huazhong University of Science and Technology
胡庆芳. 2013. 基于多源信息的降水空间估计及其水文应用研究. 北京: 清华大学
Hu Q F. 2013. Rainfall Spatial Estimation Using Multi-source Information and Its Hydrological Application. Beijing: Tsinghua University
黄钰瀚, 张增信, 费明哲, 金秋. 2016. TRMM 3B42卫星降水数据在赣江流域径流模拟中的应用. 长江流域资源与环境, 25(10): 1618–1625
Huang Y H, Zhang Z X, Fei M Z and Jin Q. 2016. Hydrological evaluation of the TMPA multi-satellite precipitation estimates over the Ganjiang Basin. Resources and Environment in the Yangtze Basin, 25(10): 1618–1625
江西水文信息网.2017. http://www.jxsl.gov.cn/slxxhw/jhsq/index.html
Jiangxi Hydrology Information Network. 2017. http://www.jxsl.gov.cn/slxxhw/jhsq/index.html.2018-10-1
李云良, 张奇, 李相虎. 2013. 鄱阳湖流域分布式水文模型的多目标参数率定. 长江流域资源与环境, 22(5): 565–572
Li Y L, Zhang Q and Li X H. 2013. Multi-objectives model calibration for distributed hydrological model in the Poyang lake watershed. Resources and Environment in the Yangtze Basin, 22(5): 565–572
刘佳凯, 张振明, 鄢郭馨, 余新晓. 2016. 潮白河流域径流对降雨的多尺度响应. 中国水土保持科学, 14(4): 50–59
Liu J K, Zhang Z M, Yan G X and Yu X X. 2016. Multi-scale analysis on precipitation-runoff relationship in Chaobaihe basin. Science of Soil and Water Conservation, 14(4): 50–59
刘硕. 2017. TRMM卫星与地面雨量站网的降水数据融合及其水文模拟应用. 武汉: 武汉大学
Liu S. 2017. Precipitation Data Fusion and Its Application in Hydrological Simulation of TRMM Satellite and Ground Rainfall Network. Wuhan: Wuhan University
吕玉娟, 彭新华, 高磊, 张中彬. 2014. 红壤丘陵岗地区坡地产流产沙特征及影响因素研究. 水土保持学报, 28(6): 19–23, 51
Lv Y J, Peng X H, Gao L and Zhang Z B. 2014. Characteristics of runoff and soil loss and their influential factors on sloping land in red soil hilly region. Journal of Soil and Water Conservation, 28(6): 19–23, 51
Machado F, Mine M, Kaviski E and Fill H. 2011. Monthly rainfall–runoff modelling using artificial neural networks. Hydrological Sciences Journal, 56(3): 349–361
NASA Earth Data. 2017. [EB/OL]. https://mirador.gsfc.nasa.gov. 2018-10-1
国家气象信息中心. 2017. [EB/OL].
National Meteorological Information Center. 2017. [EB/OL]. http://data.cma.cn
唐国强, 李哲, 薛显武, 胡庆芳, 雍斌, 洪阳. 2015. 赣江流域TRMM遥感降水对地面站点观测的可替代性. 水科学进展, 26(3): 340–346
Tang G Q, Li Z, Xue X W, Hu Q F, Yong B and Hong Y. 2015. A study of substitutability of TRMM remote sensing precipitation for gauge-based observation in Ganjiang River basin. Advances in Water Science, 26(3): 340–346
汪丽娜, 李粤安, 陈晓宏. 2009. 基于支持向量机的降雨—径流预测研究. 水文, 29(1): 13–16
Wang L N, Li Y A and Chen X H. 2009. Prediction of rainfall-runoff based on support vector machine method. Journal of China Hydrology, 29(1): 13–16
Wu C L, Chau K W and Fan C. 2010. Prediction of rainfall time series using modular artificial neural networks coupled with data-preprocessing techniques. Journal of Hydrology, 389(1/2): 146–167
杨正瓴, 张军, 陈曦, 张惊雷, 陈红新. 2007. 复杂系统行为预测的" 机理+辨识”策略. 中国科技论文在线[2006-09-29]. http://www.paper.edu.cn/releasepaper/content/200609-432
Yang Z L, Zhang J, Chen X, Zhang J L and Chen H X. 2007. " Mechanism model + identification model” strategy for Prediction of complex systems’ behavior Online[EB/OL]. [2006-09-29]. http://www.paper.edu.cn/releasepaper/content/200609-432
王野乔, 龚建雅, 夏军, 林辉, 戴星照, 方朝阳. 2016. 鄱阳湖流域生态安全及其监控. 北京: 科学出版社
Wang Y Q, Gong J Y, Xia J, Lin H, Dai X Z and Fang C Y. 2006. Ecological security and monitoring in Poyang Lake basin. Beijing: Science Press
赵雪花, 陈旭, 袁旭琦. 2014. 基于EMD的数据驱动模型在径流预测中的应用. 系统工程, (9): 150-154
Zhao X H, Chen X and Yuan X Q. 2014. Application of data-driven model based on empirical mode decomposition for runoff forecasting. Systems Engineering, (9): 150-154
周志华. 2016. 机器学习. 北京: 清华大学出版社
Zhou Z H. 2016. Machine Learning. Beijing: Tsinghua University Press
左岗岗. 2017. 基于深度学习的渭河流域径流预测系统研究.西安理工大学
Zuo G G. 2017. The Research of Wei River Runoff Prediction System Based on Machine Learning. Xi’an University of Technology.
相关作者
相关机构
京公网安备11010802024621
