A Multi-Task Learning Based Runoff Forecasting Model for Multi-Scale Chaotic Hydrological Time Series

Hui Zuo; Gaowei Yan; Ruochen Lu; Rong Li; Shuyi Xiao; Yusong Pang

doi:10.1007/s11269-023-03681-z

A Multi-Task Learning Based Runoff Forecasting Model for Multi-Scale Chaotic Hydrological Time Series

Hui Zuo, Gaowei Yan^*, Ruochen Lu, Rong Li, Shuyi Xiao, Yusong Pang

^*Corresponding author for this work

Transport Engineering and Logistics

Research output: Contribution to journal › Article › Scientific › peer-review

1 Citation (Scopus)

14 Downloads (Pure)

Abstract

Accurately predicting runoff is crucial for managing water resources, preventing and mitigating floods, scheduling hydropower plant operations, and protecting the environment. The hydrological dynamic composite system that forms runoff is complex and random, and seemingly random behavior may be caused by nonlinear variables in a simple deterministic system, which poses a challenge to runoff prediction. In this paper, we construct parallel and multi-timescale reservoirs from a chaotic theory perspective to simulate the stochasticity of chaotic systems. We propose a multi-task-based "Decomposition-Integration-Prediction" (Multi-SDIPC) model for runoff prediction. To validate our research results, we use the Catchment Attributes and Meteorology for Large-Sample Studies (CAMELS) dataset and compare our proposed model with 10 baseline models. The results show that our model has an average NSE metric of 0.83 and exhibits higher accuracy, better generalization, and greater stability than the other models in multi-step forecasting. Based on our findings, we recommend wider application of the Multi-SDIPC model in different regions of the world for medium or long-term runoff prediction.

Original language	English
Pages (from-to)	481-503
Number of pages	23
Journal	Water Resources Management
Volume	38 (2024)
Issue number	2
DOIs	https://doi.org/10.1007/s11269-023-03681-z
Publication status	Published - 2023

Bibliographical note

Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care
Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.

Funding

Funding Information:
This work was supported by the National Natural Science Foundation of China (61973226, 62003233), the Shanxi Provincial Department of Water Resources 2023 Water Technology Research and Promotion Subsidy Projects(2023GM17), the Shanxi Province Major Special Program of Science and Technology “Unveiling and Commanding“ Project(202201090301013), the Natural Science Foundation of Shanxi Province(202203021222101).

Keywords

Chaos theory
Convolutional neural network
Multi-task learning
Reservoir computing
Runoff prediction

Access to Document

10.1007/s11269-023-03681-z

s11269-023-03681-zFinal published version, 4.76 MB

Cite this

@article{06b29c3e6ea64636866fa8bb294d6127,

title = "A Multi-Task Learning Based Runoff Forecasting Model for Multi-Scale Chaotic Hydrological Time Series",

abstract = "Accurately predicting runoff is crucial for managing water resources, preventing and mitigating floods, scheduling hydropower plant operations, and protecting the environment. The hydrological dynamic composite system that forms runoff is complex and random, and seemingly random behavior may be caused by nonlinear variables in a simple deterministic system, which poses a challenge to runoff prediction. In this paper, we construct parallel and multi-timescale reservoirs from a chaotic theory perspective to simulate the stochasticity of chaotic systems. We propose a multi-task-based {"}Decomposition-Integration-Prediction{"} (Multi-SDIPC) model for runoff prediction. To validate our research results, we use the Catchment Attributes and Meteorology for Large-Sample Studies (CAMELS) dataset and compare our proposed model with 10 baseline models. The results show that our model has an average NSE metric of 0.83 and exhibits higher accuracy, better generalization, and greater stability than the other models in multi-step forecasting. Based on our findings, we recommend wider application of the Multi-SDIPC model in different regions of the world for medium or long-term runoff prediction.",

keywords = "Chaos theory, Convolutional neural network, Multi-task learning, Reservoir computing, Runoff prediction",

author = "Hui Zuo and Gaowei Yan and Ruochen Lu and Rong Li and Shuyi Xiao and Yusong Pang",

note = "Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.",

year = "2023",

doi = "10.1007/s11269-023-03681-z",

language = "English",

volume = "38 (2024)",

pages = "481--503",

journal = "Water Resources Management",

issn = "0920-4741",

publisher = "Springer",

number = "2",

}

TY - JOUR

T1 - A Multi-Task Learning Based Runoff Forecasting Model for Multi-Scale Chaotic Hydrological Time Series

AU - Zuo, Hui

AU - Yan, Gaowei

AU - Lu, Ruochen

AU - Li, Rong

AU - Xiao, Shuyi

AU - Pang, Yusong

N1 - Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.

PY - 2023

Y1 - 2023

N2 - Accurately predicting runoff is crucial for managing water resources, preventing and mitigating floods, scheduling hydropower plant operations, and protecting the environment. The hydrological dynamic composite system that forms runoff is complex and random, and seemingly random behavior may be caused by nonlinear variables in a simple deterministic system, which poses a challenge to runoff prediction. In this paper, we construct parallel and multi-timescale reservoirs from a chaotic theory perspective to simulate the stochasticity of chaotic systems. We propose a multi-task-based "Decomposition-Integration-Prediction" (Multi-SDIPC) model for runoff prediction. To validate our research results, we use the Catchment Attributes and Meteorology for Large-Sample Studies (CAMELS) dataset and compare our proposed model with 10 baseline models. The results show that our model has an average NSE metric of 0.83 and exhibits higher accuracy, better generalization, and greater stability than the other models in multi-step forecasting. Based on our findings, we recommend wider application of the Multi-SDIPC model in different regions of the world for medium or long-term runoff prediction.

AB - Accurately predicting runoff is crucial for managing water resources, preventing and mitigating floods, scheduling hydropower plant operations, and protecting the environment. The hydrological dynamic composite system that forms runoff is complex and random, and seemingly random behavior may be caused by nonlinear variables in a simple deterministic system, which poses a challenge to runoff prediction. In this paper, we construct parallel and multi-timescale reservoirs from a chaotic theory perspective to simulate the stochasticity of chaotic systems. We propose a multi-task-based "Decomposition-Integration-Prediction" (Multi-SDIPC) model for runoff prediction. To validate our research results, we use the Catchment Attributes and Meteorology for Large-Sample Studies (CAMELS) dataset and compare our proposed model with 10 baseline models. The results show that our model has an average NSE metric of 0.83 and exhibits higher accuracy, better generalization, and greater stability than the other models in multi-step forecasting. Based on our findings, we recommend wider application of the Multi-SDIPC model in different regions of the world for medium or long-term runoff prediction.

KW - Chaos theory

KW - Convolutional neural network

KW - Multi-task learning

KW - Reservoir computing

KW - Runoff prediction

UR - http://www.scopus.com/inward/record.url?scp=85178244842&partnerID=8YFLogxK

U2 - 10.1007/s11269-023-03681-z

DO - 10.1007/s11269-023-03681-z

M3 - Article

AN - SCOPUS:85178244842

SN - 0920-4741

VL - 38 (2024)

SP - 481

EP - 503

JO - Water Resources Management

JF - Water Resources Management

IS - 2

ER -

A Multi-Task Learning Based Runoff Forecasting Model for Multi-Scale Chaotic Hydrological Time Series

Abstract

Bibliographical note

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this