Application of combined correction method based on Ensemble Kalman Filter and LSTM neural network in hydrological modelling

LV Yonglin; LIU Xinbo; WANG Pengxiang; ZHAI Shaojun; CAO Hui; QI Hongwei

doi:10.14188/j.1671-8844.2025.0095

您当前的位置：

首页 >

文章列表页 >

Application of combined correction method based on Ensemble Kalman Filter and LSTM neural network in hydrological modelling

更新时间：2026-10-15

- Application of combined correction method based on Ensemble Kalman Filter and LSTM neural network in hydrological modelling
- Engineering Journal of Wuhan University Pages: 1-12(2026)
- 作者机构：
  
  1.中国长江电力股份有限公司，湖北宜昌 443002
  2.武汉大学水利水电学院，湖北武汉 430072
- 作者简介：
- 基金信息：
- DOI：10.14188/j.1671-8844.2025.0095
  CLC： P333
- Online First：15 October 2026，
- 稿件说明：
移动端阅览
LV Yonglin, LIU Xinbo, WANG Pengxiang, et al. Application of combined correction method based on Ensemble Kalman Filter and LSTM neural network in hydrological modelling[J/OL]. Engineering Journal of Wuhan University, 2026, 1-12.
DOI：

LV Yonglin, LIU Xinbo, WANG Pengxiang, et al. Application of combined correction method based on Ensemble Kalman Filter and LSTM neural network in hydrological modelling[J/OL]. Engineering Journal of Wuhan University, 2026, 1-12. DOI： 10.14188/j.1671-8844.2025.0095.

摘要

准确获取径流信息可为水资源评价、水资源调度和水资源管理与利用提供科学依据和数据支撑。本文基于三水源新安江模型，构建了一种集合卡尔曼滤波（EnKF）与长短期记忆神经网络（LSTM）的联合校正方法（EnKF-LSTM）。以岷沱江流域2010-2019年日尺度径流模拟为研究对象，综合评估了EnKF-LSTM、EnKF和LSTM的校正效果。研究结果表明，与校正前的初始三水源新安江模型结果对比来看，三种校正方法均可提升日径流过程模拟效果，其中EnKF-LSTM联合校正方法效果最为显著，LSTM次之，EnKF相对较差。相对于校正前，EnKF-LSTM可使岷江流域径流模拟NSE、R

和KGE分别提升25.37%、22.86%和13.92%；沱江流域径流模拟NSE、R

和KGE分别提升18.57%、22.85%和11.54%。从三种校正方法对丰水期与枯水期校正效果来看，EnKF-LSTM联合校正方法也均优于EnKF与LSTM。本文构建的EnKF-LSTM联合校正方法能够有效地提高径流模拟的精度，可为流域的径流模拟提供有效的技术支撑。

Abstract

Accurate runoff information is an important basis for the study of hydrological processes

which can provide scientific basis and data support for water resources management and utilization. In this study

we develop an integrated correction method (EnKF-LSTM) of ensemble Kalman filter (EnKF) and long short-term memory neural network (LSTM) based on the Xin'anjiang model. The Mintuo River Basin is selected as a case study area to comprehensively evaluate the correction effects of EnKF-LSTM

EnKF and LSTM. The results show that all the three calibration methods can improve the accuracy of daily runoff modelling

among which the EnKF-LSTM shows the best performance

followed by LSTM and EnKF. Compared with the pre-corrected scenario

the NSE

and KGE given by EnKF-LSTM increases by 25.37%

22.86% and 13.92% in the Minjiang River Basin

respectively

and 18.57%

22.85% and 11.54% in the Tuojiang River Basin

respectively. Moreover

the EnKF-LSTM outperforms both EnKF and LSTM in terms of both the flood and dry seasons. Therefore

the developed EnKF-LSTM in this study can effectively improve the accuracy of runoff simulations and provide effective technical support for runoff simulations in the basin.

关键词

Keywords

references

Chen S , Huang J , Huang J C . Improving daily streamflow simulations for data-scarce watersheds using the coupled SWAT-LSTM approach [J ] . JOURNAL OF HYDROLOGY , 2023 , 622 : 129734 .

Feng Z kai , Shi P fei , Yang T , et al . Parallel cooperation search algorithm and artificial intelligence method for streamflow time series forecasting [J ] . JOURNAL OF HYDROLOGY , 2022 , 606 : 127434 .

Tursun A , Xie X , Wang Y , et al . Enhancing streamflow simulation in large and human-regulated basins: Long short-term memory with multiscale attributes [J ] . JOURNAL OF HYDROLOGY , 2024 , 630 : 130771 .

苟娇娇 , 缪驰远 , 段青云 . 水文模型参数敏感性分析——优化——区域化方法研究进展 [J ] . 地理科学进展 , 2022 , 41 ( 7 ): 1338 - 1348 .

GOU jiaojiao , MIAO Chiyuan , DUAN Qingyun . Progress in parameter sensitivity analysis-optimization-regionalization methods for hydrological models [J ] . Progress in Geography , 2022 , 41 ( 7 ): 1338 - 1348

McMillan H , Jackson B , Clark M , et al . Rainfall uncertainty in hydrological modelling: An evaluation of multiplicative error models [J ] . JOURNAL OF HYDROLOGY , 2011 , 400 ( 1-2 ): 83 - 94 .

Tang G , Clark M P , Knoben W J M , et al . The Impact of Meteorological Forcing Uncertainty on Hydrological Modeling: A Global Analysis of Cryosphere Basins [J ] . WATER RESOURCES RESEARCH , 2023 , 59 ( 6 ): e2022WR033767 .

张旭旻 , 瞿思敏 , 李倩 , 等 . 基于协整理论的淮河流域上游洪水预报实时校正方法 [J ] . 水资源保护 , 2022 , 38 ( 6 ): 88 - 95+145 .

ZHANG Xumin , QU Simin , LI Qian , et al . Real-time correction method of flood forecasting for the upper Huaihe River Basin based on cointegration theory [J ] . Water Resources Protection , 2022 , 38 ( 6 ): 88 - 95+145 .

刘开磊 , 李致家 , 姚成 , 等 . 水文学与水力学方法在淮河中游的应用研究 [J ] . 水力发电学报 , 2013 , 32 ( 6 ): 5 - 10 .

LIU Kailei , Li Zhijia , Yao Cheng , et al . Study on hydrology and hydraulic methods applied to the middle Huai River basin [J ] . Journal of Hydroelectric Engineering , 2013 , 32 ( 6 )P: 5 - 10

孙文宇 , 姚成 , 刘志雨 , 等 . 秦淮河水文水动力模型及实时校正 [J ] . 河海大学学报(自然科学版) , 2022 , 50 ( 4 ): 26 - 33 .

SUN Wenyu , YAO Cheng , LIU Zhiyu , et al . Hydrological and hydrodynamic model of Qinhuai River and real-time correction [J ] . Journal of Hohai University (Natural Sciences) , 2022 , 50 ( 4 ): 26 - 33 .

刘昱辰 , 刘佳 , 刘录三 , 等 . 基于LSTM实时校正的WRF/WRF-Hydro耦合径流预报 [J ] . 水利学报 , 2023 , 54 ( 11 ): 1334 - 1346 .

LIU Yuchen , LIU Jia , LI Lusan , et al . WRF/WRF-Hydro coupled streamflow forecasting based on real-time updating using LSTM [J ] . Journal of Hydraulic Engineering , 2023 , 54 ( 11 ): 1334 - 1346 .

陈鑫 , 刘艳丽 , 张建云 , 等 . 数据挖掘技术在洪水预报实时校正中的应用 [J ] . 水力发电学报 , 2022 , 41 ( 8 ): 54 - 62 .

CHEN Xin , LIU Yanli , ZHANG Jianyun , et al . Application of data mining techniques in real-time correction of flood forecasts [J ] , Journal of Hydroelectric Engineering , 2022 , 41 ( 8 ): 54 - 62 .

包为民 , 孙逸群 , 周俊伟 , 等 . 基于总体最小二乘法的系统响应修正方法 [J ] . 水利学报 , 2017 , 48 ( 5 ): 560 - 567 .

BAO Weimin , SUN Yiqun , ZHOU Junwei , et al . A new version of system response method for error correction based on total least squares [J ] . Journal of Hydraulic Engineering , 2017 , 48 ( 5 ): 560 - 567 .

Gong J , Weerts A H , Yao C , et al . State updating in a distributed hydrological model by ensemble Kalman filtering with error estimation [J ] . JOURNAL OF HYDROLOGY , 2023 , 620 : 129450 .

司伟 , 包为民 , 瞿思敏 . 洪水预报产流误差的动态系统响应曲线修正方法 [J ] . 水科学进展 , 2013 , 24 ( 4 ): 497 - 503 .

SI Wei , BAO Weimin , QU Simin . Runoff error correction in real-time flood forecasting based on dynamic system response curve [J ] . Advances in Water Science , 2013 , 24 ( 4 ): 497 - 503 .

张小琴 , 刘可新 , 包为民 , 等 . 产流误差比例系数的系统响应修正方法 [J ] . 水科学进展 , 2014 , 25 ( 6 ): 789 - 796 .

ZHANG Xiaoqin , LIU Kexin , BAO Weimin , et al . Runoff error proportionality coefficient correction method based on system response [J ] . Advances in Water Science , 2014 , 25 ( 6 ): 789 - 796 .

Zhang X , Liu P , Cheng L , et al . A back-fitting algorithm to improve real-time flood forecasting [J ] . JOURNAL OF HYDROLOGY , 2018 , 562 : 140 - 150 .

Wang J , Bao W , Xiao Z , 等 . Multi-model integrated error correction for streamflow simulation based on Bayesian model averaging and dynamic system response curve [J ] . JOURNAL OF HYDROLOGY , 2022 , 607 : 127518 .

Liu Y , Ji Y , Liu D , et al . A new method for runoff prediction error correction based on LS-SVM and a 4D copula joint distribution [J ] . JOURNAL OF HYDROLOGY , 2021 , 598 : 126223 .

张建云 . 中国水文预报技术发展的回顾与思考 [J ] . 水科学进展 , 2010 , 21 ( 4 ): 435 - 443 .

ZHANG Jianyun . Review and reflection on China's hydrological forecasting techniques [J ] . Advances in Water Science , 2010 , 21 ( 4 ): 435 - 443 .

梁忠民 , 赵建飞 , 段雅楠 , 等 . 微分形式新安江模型 [J ] . 水科学进展 , 2024 , 35 ( 3 ): 374 - 386 .

LIANG Zhongmin , ZHAO Jianfei , DUAN Yanan , et al . Differential-form Xin′anjiang model [J ] . Advances in Water Science , 2024 , 35 ( 3 ): 374 - 386 .

刘松 , 佘敦先 , 张利平 , 等 . 基于Morris和Sobol的水文模型参数敏感性分析 [J ] . 长江流域资源与环境 , 2019 , 28 ( 6 ): 1296 - 1303 .

LIU Song , SHE Dunxian , ZHANG Liping , et al . Global Sensitivity Analysis of Hydrological Model Parameters Based on Morris and Sobol Methods [J ] . Resources and Environment in the Yangtze Basin , 2019 , 28 ( 6 ): 1296 - 1303 .

Sobol I M . Sensitivity estimates for nonlinear mathematical models .[J ] . Math.model.comput.exp , 1993 , 1 ( 1 ): 112 - 118 .

Piazzi G , Thirel G , Perrin C , et al . Sequential Data Assimilation for Streamflow Forecasting: Assessing the Sensitivity to Uncertainties and Updated Variables of a Conceptual Hydrological Model at Basin Scale [J ] . WATER RESOURCES RESEARCH , 2021 , 57 ( 4 ): e2020WR028390 .

Cho K , Kim Y . Improving streamflow prediction in the WRF-Hydro model with LSTM networks [J ] . JOURNAL OF HYDROLOGY , 2022 , 605 : 127297 .

李志超 , 邬强 , 胡彩虹 , 等 . 基于LSTM误差校正蓄满-超渗兼容模型的山洪预报研究 [J ] . 武汉大学学报(工学版) , 2023 , 56 ( 10 ): 1161 - 1171 .

LI Zhichao , WU Qiang , HU Caihong , et al . Flash flood prediction based on LSTM error correction saturation-excess and infiltration-excess compatible model [J ] . Engineering Journal of Wuhan University , 2023 , 56 ( 10 ): 1161 - 1171 .

Cho K , Kim Y . Improving streamflow prediction in the WRF-Hydro model with LSTM networks [J ] . JOURNAL OF HYDROLOGY , 2022 , 605 : 127297 .

田辽西 , 覃华清 . 基于多变量LSTM神经网络的地下水位预测方法研究 [J ] . 干旱区资源与环境 , 2024 , 38 ( 9 ): 138 - 146 .

TIAN Liaoxi , QIN Huaqing . Multivariable LSTM model-based groundwater level prediction [J ] . Journal of Arid Land Resources and Environment , 2024 , 38 ( 9 ): 138 - 146 .

Moradkhani H , Hsu K L , Gupta H , et al . Uncertainty assessment of hydrologic model states and parameters: Sequential data assimilation using the particle filter [J ] . Water Resources Research , 2005 , 41 ( 5 )[ 2025-04-13 ] .

陈华 , 郭家力 , 郭生练 , 等 . 统计降尺度方法及其评价指标比较研究 [J ] . 水利学报 , 2012 , 43 ( 8 ): 891 - 897 .

CHEN Hua , GUO Jiali , GUO Shenglian , et al . Comparison of different statistical downscaling methods and evaluation indicators in climate change impact on runoff [J ] . Journal of Hydraulic Engineering , 2012 , 43 ( 8 ): 891 - 897 .

Liu D . A rational performance criterion for hydrological model [J ] . JOURNAL OF HYDROLOGY , 2020 , 590 : 125488 .

Patil A , Ramsankaran R . Improving streamflow simulations and forecasting performance of SWAT model by assimilating remotely sensed soil moisture observations [J ] . Journal of Hydrology , 2017 , 555 : 683 - 696 .

Frame J M , Kratzert F , Raney A , et al . Post-Processing the National Water Model with Long Short-Term Memory Networks for Streamflow Predictions and Model Diagnostics [J ] . JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION , 2021 , 57 ( 6 ): 885 - 905 .

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Applicability of catchment hydrologic models in arid and semi-arid regions

Study of real-time prediction of soil nitrate transport and transformation based on ensemble Kalman filter

Application of ensemble Kalman filter to phreatic water flow

Study on tracking inversion of rockfill dams with prior information supervision

Multi-dialect speech recognition based on lightweight Conformer and dialect embedding

Related Author

Song LIU

Liping ZHANG

Dunxian SHE

Qiang WANG

Chen HU

Jun XIA

Longfei LI

Jiaqi ZHANG

Related Institution

State Key Laboratory of Water Resources and Hydropower Engineering Science, Wuhan University

AI问答

⁰