Related Experiment Video
Updated: Jul 16, 2025

Continuous Hydrologic and Water Quality Monitoring of Vernal Ponds
Published on: November 13, 2017
A deep learning model for reconstructing centenary water storage changes in the Yangtze River Basin.
Jielong Wang1, Yunzhong Shen1, Joseph L Awange2
1College of Surveying and Geo-informatics, Tongji University, Shanghai 200092, PR China.
A new deep learning model, RecNet, reconstructs a century of water storage changes in the Yangtze River Basin, revealing vulnerability to extreme climate events and informing water management strategies.
Area of Science:
- Hydrology and Climate Science
- Earth Observation and Remote Sensing
- Artificial Intelligence in Environmental Science
Background:
- Accurate total water storage anomaly (TWSA) observations are crucial for understanding hydrological variability.
- Limited historical GRACE (Gravity Recovery and Climate Experiment) data in the Yangtze River Basin (YRB) hinder long-term variability studies.
- Understanding long-term TWSA is vital for effective water resource management and climate change adaptation.
Purpose of the Study:
- To develop and validate a deep learning model (RecNet) for reconstructing climate-driven TWSA in the YRB from 1923 to 2022.
- To assess the frequency of extreme hydrological conditions and their teleconnections with major climate patterns.
- To provide valuable data for long-term climate variability studies and future drought/flood projections in the YRB.
Main Methods:
- Developed RecNet, a deep learning model trained on precipitation, temperature, and GRACE observations using a weighted mean square error (WMSE) loss function.
- Validated RecNet performance against GRACE data, water budget estimates, hydrological models, drought indices, and existing reconstruction datasets.
- Employed independent component analysis and wavelet coherence analysis to investigate climate pattern influences on TWSA.
Main Results:
- RecNet successfully reconstructed historical water storage changes in the YRB, outperforming previous methods.
- The reconstructed data revealed recurrent extreme dry/wet conditions in the YRB over the past century.
- Significant coherence was found between major climate patterns (ENSO, IOD, PDO, NAO) and TWSA across the YRB.
Conclusions:
- The RecNet model provides a robust method for reconstructing long-term TWSA, overcoming data limitations.
- The YRB demonstrates significant vulnerability to climate variability, with extreme hydrological events occurring frequently.
- Reconstructed datasets offer critical insights for water resource management and climate change adaptation in the YRB.
Related Concept Videos
Design Example: Creating a Hydraulic Model of a Dam Spillway
Typical Model Studies
Design Example: Analyzing Capacity Contours for Flood Risk Assessment
Modeling and Similitude
Responses to Drought and Flooding
Regulation of Water Output

