Simulation of Spring Discharge Using Deep Learning, Considering the Spatiotemporal Variability of Precipitation

Abstract Sparse precipitation data in karst catchments challenge hydrologic models to accurately capture the spatial and temporal relationships between precipitation and karst spring discharge, hindering robust predictions. This study addresses this issue by employing a coupled deep learning model t...

Full description

Saved in:
Bibliographic Details
Main Authors: Chunmei Ma, Haoyu Jiao, Yonghong Hao, Tian‐Chyi Jim Yeh, Junfeng Zhu, Huiqing Hao, Jiahui Lu, Jiankang Dong
Format: Article
Language:English
Published: Wiley 2025-04-01
Series:Water Resources Research
Subjects:
Online Access:https://doi.org/10.1029/2024WR037449
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850211981567459328
author Chunmei Ma
Haoyu Jiao
Yonghong Hao
Tian‐Chyi Jim Yeh
Junfeng Zhu
Huiqing Hao
Jiahui Lu
Jiankang Dong
author_facet Chunmei Ma
Haoyu Jiao
Yonghong Hao
Tian‐Chyi Jim Yeh
Junfeng Zhu
Huiqing Hao
Jiahui Lu
Jiankang Dong
author_sort Chunmei Ma
collection DOAJ
description Abstract Sparse precipitation data in karst catchments challenge hydrologic models to accurately capture the spatial and temporal relationships between precipitation and karst spring discharge, hindering robust predictions. This study addresses this issue by employing a coupled deep learning model that integrates a variation autoencoder (VAE) for augmenting precipitation and a long short‐term memory (LSTM) network for karst spring discharge prediction. The VAE contributes by generating synthetic precipitation data through an encoding‐decoding process. This process generalizes the observed precipitation data by deriving joint latent distributions with improved preservation of temporal and spatial correlations of the data. The combined VAE‐generated precipitation and observation data are used to train and test the LSTM to predict spring discharge. Applied to the Niangziguan spring catchment in northern China, the average performance of NSE, root mean square error, mean absolute error, mean absolute percentage error, and log NSE of our coupled VAE/LSTM model reached 0.93, 0.26, 0.15, 1.8, and 0.92, respectively, yielding 145%, 52%, 63%, 70% and 149% higher than an LSTM model using only observations. We also explored temporal and spatial correlations in the observed data and the impact of different ratios of VAE‐generated precipitation data to actual data on model performances. This study also evaluated the effectiveness of VAE‐augmented data on various deep‐learning models and compared VAE with other data augmentation techniques. We demonstrate that the VAE offers a novel approach to address data scarcity and uncertainty, improving learning generalization and predictive capability of various hydrological models. However, we recognize that innovations to address hydrologic problems at different scales remain to be explored.
format Article
id doaj-art-8bfd9add428b4565ae3523a82fdc4479
institution OA Journals
issn 0043-1397
1944-7973
language English
publishDate 2025-04-01
publisher Wiley
record_format Article
series Water Resources Research
spelling doaj-art-8bfd9add428b4565ae3523a82fdc44792025-08-20T02:09:26ZengWileyWater Resources Research0043-13971944-79732025-04-01614n/an/a10.1029/2024WR037449Simulation of Spring Discharge Using Deep Learning, Considering the Spatiotemporal Variability of PrecipitationChunmei Ma0Haoyu Jiao1Yonghong Hao2Tian‐Chyi Jim Yeh3Junfeng Zhu4Huiqing Hao5Jiahui Lu6Jiankang Dong7School of Computer and Information Engineering Tianjin Normal University Tianjin ChinaSchool of Computer and Information Engineering Tianjin Normal University Tianjin ChinaTianjin Key Laboratory of Water Resources and Environment Tianjin Normal University Tianjin ChinaTianjin Key Laboratory of Water Resources and Environment Tianjin Normal University Tianjin ChinaKentucky Geological Survey University of Kentucky Lexington KY USATianjin Key Laboratory of Water Resources and Environment Tianjin Normal University Tianjin ChinaSchool of Computer and Information Engineering Tianjin Normal University Tianjin ChinaTianjin Key Laboratory of Water Resources and Environment Tianjin Normal University Tianjin ChinaAbstract Sparse precipitation data in karst catchments challenge hydrologic models to accurately capture the spatial and temporal relationships between precipitation and karst spring discharge, hindering robust predictions. This study addresses this issue by employing a coupled deep learning model that integrates a variation autoencoder (VAE) for augmenting precipitation and a long short‐term memory (LSTM) network for karst spring discharge prediction. The VAE contributes by generating synthetic precipitation data through an encoding‐decoding process. This process generalizes the observed precipitation data by deriving joint latent distributions with improved preservation of temporal and spatial correlations of the data. The combined VAE‐generated precipitation and observation data are used to train and test the LSTM to predict spring discharge. Applied to the Niangziguan spring catchment in northern China, the average performance of NSE, root mean square error, mean absolute error, mean absolute percentage error, and log NSE of our coupled VAE/LSTM model reached 0.93, 0.26, 0.15, 1.8, and 0.92, respectively, yielding 145%, 52%, 63%, 70% and 149% higher than an LSTM model using only observations. We also explored temporal and spatial correlations in the observed data and the impact of different ratios of VAE‐generated precipitation data to actual data on model performances. This study also evaluated the effectiveness of VAE‐augmented data on various deep‐learning models and compared VAE with other data augmentation techniques. We demonstrate that the VAE offers a novel approach to address data scarcity and uncertainty, improving learning generalization and predictive capability of various hydrological models. However, we recognize that innovations to address hydrologic problems at different scales remain to be explored.https://doi.org/10.1029/2024WR037449machine learningdata augmentationgroundwaterLSTMhydraulic tomographyspring discharge
spellingShingle Chunmei Ma
Haoyu Jiao
Yonghong Hao
Tian‐Chyi Jim Yeh
Junfeng Zhu
Huiqing Hao
Jiahui Lu
Jiankang Dong
Simulation of Spring Discharge Using Deep Learning, Considering the Spatiotemporal Variability of Precipitation
Water Resources Research
machine learning
data augmentation
groundwater
LSTM
hydraulic tomography
spring discharge
title Simulation of Spring Discharge Using Deep Learning, Considering the Spatiotemporal Variability of Precipitation
title_full Simulation of Spring Discharge Using Deep Learning, Considering the Spatiotemporal Variability of Precipitation
title_fullStr Simulation of Spring Discharge Using Deep Learning, Considering the Spatiotemporal Variability of Precipitation
title_full_unstemmed Simulation of Spring Discharge Using Deep Learning, Considering the Spatiotemporal Variability of Precipitation
title_short Simulation of Spring Discharge Using Deep Learning, Considering the Spatiotemporal Variability of Precipitation
title_sort simulation of spring discharge using deep learning considering the spatiotemporal variability of precipitation
topic machine learning
data augmentation
groundwater
LSTM
hydraulic tomography
spring discharge
url https://doi.org/10.1029/2024WR037449
work_keys_str_mv AT chunmeima simulationofspringdischargeusingdeeplearningconsideringthespatiotemporalvariabilityofprecipitation
AT haoyujiao simulationofspringdischargeusingdeeplearningconsideringthespatiotemporalvariabilityofprecipitation
AT yonghonghao simulationofspringdischargeusingdeeplearningconsideringthespatiotemporalvariabilityofprecipitation
AT tianchyijimyeh simulationofspringdischargeusingdeeplearningconsideringthespatiotemporalvariabilityofprecipitation
AT junfengzhu simulationofspringdischargeusingdeeplearningconsideringthespatiotemporalvariabilityofprecipitation
AT huiqinghao simulationofspringdischargeusingdeeplearningconsideringthespatiotemporalvariabilityofprecipitation
AT jiahuilu simulationofspringdischargeusingdeeplearningconsideringthespatiotemporalvariabilityofprecipitation
AT jiankangdong simulationofspringdischargeusingdeeplearningconsideringthespatiotemporalvariabilityofprecipitation