A Novel Deep Learning Approach for Data Assimilation of Complex Hydrological Systems

Abstract In hydrological research, data assimilation (DA) is widely used to fuse the information contained in process‐based models and observational data to reduce simulation uncertainty. However, many popular DA methods are limited by low computational efficiency or their reliance on the Gaussian a...

Full description

Saved in:
Bibliographic Details
Main Authors: Jiangjiang Zhang, Chenglong Cao, Tongchao Nan, Lei Ju, Hongxiang Zhou, Lingzao Zeng
Format: Article
Language:English
Published: Wiley 2024-02-01
Series:Water Resources Research
Online Access:https://doi.org/10.1029/2023WR035389
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849687163251195904
author Jiangjiang Zhang
Chenglong Cao
Tongchao Nan
Lei Ju
Hongxiang Zhou
Lingzao Zeng
author_facet Jiangjiang Zhang
Chenglong Cao
Tongchao Nan
Lei Ju
Hongxiang Zhou
Lingzao Zeng
author_sort Jiangjiang Zhang
collection DOAJ
description Abstract In hydrological research, data assimilation (DA) is widely used to fuse the information contained in process‐based models and observational data to reduce simulation uncertainty. However, many popular DA methods are limited by low computational efficiency or their reliance on the Gaussian assumption. To address these limitations, we propose a novel DA method called DA(DL), which leverages the capabilities of DL to model non‐linear relationships and recognize complex patterns. DA(DL) first generates a large volume of training data from the prior ensemble, and then trains a DL model to update the system knowledge (e.g., model parameters in this study) from multiple predictors. For highly non‐linear models, an iterative form of DA(DL) can be implemented. Additionally, strategies of data augmentation and local updating are proposed to enhance DA(DL) for problems involving small ensemble size and the equifinality issue, respectively. In two hydrological DA cases involving Gaussian and non‐Gaussian distributions, DA(DL) shows promising performance compared to two ensemble smoother methods, that is, ES(K) and ES(DL), which respectively apply the Kalman‐ and DL‐based updates. Potential improvements to DA(DL) can be made by designing better DL model architectures, imposing physical constraints to the training of the DL model, and further updating other important variables like model states, forcings and error terms.
format Article
id doaj-art-05ee1b01b03b4b939085fe82ea548aa0
institution DOAJ
issn 0043-1397
1944-7973
language English
publishDate 2024-02-01
publisher Wiley
record_format Article
series Water Resources Research
spelling doaj-art-05ee1b01b03b4b939085fe82ea548aa02025-08-20T03:22:26ZengWileyWater Resources Research0043-13971944-79732024-02-01602n/an/a10.1029/2023WR035389A Novel Deep Learning Approach for Data Assimilation of Complex Hydrological SystemsJiangjiang Zhang0Chenglong Cao1Tongchao Nan2Lei Ju3Hongxiang Zhou4Lingzao Zeng5Yangtze Institute for Conservation and Development, Hohai University Nanjing ChinaYangtze Institute for Conservation and Development, Hohai University Nanjing ChinaYangtze Institute for Conservation and Development, Hohai University Nanjing ChinaNational Demonstration Center for Environment and Planning, College of Geography and Environmental Science, Henan University Kaifeng ChinaCollege of Metrology and Measurement Engineering, China Jiliang University Hangzhou ChinaZhejiang Provincial Key Laboratory of Agricultural Resources and Environment College of Environmental and Resource Sciences, Zhejiang University Hangzhou ChinaAbstract In hydrological research, data assimilation (DA) is widely used to fuse the information contained in process‐based models and observational data to reduce simulation uncertainty. However, many popular DA methods are limited by low computational efficiency or their reliance on the Gaussian assumption. To address these limitations, we propose a novel DA method called DA(DL), which leverages the capabilities of DL to model non‐linear relationships and recognize complex patterns. DA(DL) first generates a large volume of training data from the prior ensemble, and then trains a DL model to update the system knowledge (e.g., model parameters in this study) from multiple predictors. For highly non‐linear models, an iterative form of DA(DL) can be implemented. Additionally, strategies of data augmentation and local updating are proposed to enhance DA(DL) for problems involving small ensemble size and the equifinality issue, respectively. In two hydrological DA cases involving Gaussian and non‐Gaussian distributions, DA(DL) shows promising performance compared to two ensemble smoother methods, that is, ES(K) and ES(DL), which respectively apply the Kalman‐ and DL‐based updates. Potential improvements to DA(DL) can be made by designing better DL model architectures, imposing physical constraints to the training of the DL model, and further updating other important variables like model states, forcings and error terms.https://doi.org/10.1029/2023WR035389
spellingShingle Jiangjiang Zhang
Chenglong Cao
Tongchao Nan
Lei Ju
Hongxiang Zhou
Lingzao Zeng
A Novel Deep Learning Approach for Data Assimilation of Complex Hydrological Systems
Water Resources Research
title A Novel Deep Learning Approach for Data Assimilation of Complex Hydrological Systems
title_full A Novel Deep Learning Approach for Data Assimilation of Complex Hydrological Systems
title_fullStr A Novel Deep Learning Approach for Data Assimilation of Complex Hydrological Systems
title_full_unstemmed A Novel Deep Learning Approach for Data Assimilation of Complex Hydrological Systems
title_short A Novel Deep Learning Approach for Data Assimilation of Complex Hydrological Systems
title_sort novel deep learning approach for data assimilation of complex hydrological systems
url https://doi.org/10.1029/2023WR035389
work_keys_str_mv AT jiangjiangzhang anoveldeeplearningapproachfordataassimilationofcomplexhydrologicalsystems
AT chenglongcao anoveldeeplearningapproachfordataassimilationofcomplexhydrologicalsystems
AT tongchaonan anoveldeeplearningapproachfordataassimilationofcomplexhydrologicalsystems
AT leiju anoveldeeplearningapproachfordataassimilationofcomplexhydrologicalsystems
AT hongxiangzhou anoveldeeplearningapproachfordataassimilationofcomplexhydrologicalsystems
AT lingzaozeng anoveldeeplearningapproachfordataassimilationofcomplexhydrologicalsystems
AT jiangjiangzhang noveldeeplearningapproachfordataassimilationofcomplexhydrologicalsystems
AT chenglongcao noveldeeplearningapproachfordataassimilationofcomplexhydrologicalsystems
AT tongchaonan noveldeeplearningapproachfordataassimilationofcomplexhydrologicalsystems
AT leiju noveldeeplearningapproachfordataassimilationofcomplexhydrologicalsystems
AT hongxiangzhou noveldeeplearningapproachfordataassimilationofcomplexhydrologicalsystems
AT lingzaozeng noveldeeplearningapproachfordataassimilationofcomplexhydrologicalsystems