Surface and sub-surface flow estimation at high temporal resolution using deep neural networks

Ather Abbas; Sangsoo Baek; Minjeong Kim; Mayzonee Ligaray; Olivier Ribolzi; Norbert Silvera; Joong Hyuk Min; Laurie Boithias; Kyung Hwa Cho

doi:10.1016/j.jhydrol.2020.125370

Surface and sub-surface flow estimation at high temporal resolution using deep neural networks

Ather Abbas, Sangsoo Baek, Minjeong Kim, Mayzonee Ligaray, Olivier Ribolzi, Norbert Silvera, Joong Hyuk Min, Laurie Boithias^*, Kyung Hwa Cho

^*Corresponding author for this work

Physical Sciences and Engineering

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

17 Scopus citations

Abstract

Recent intensification in climate change have resulted in the rise of hydrological extreme events. This demands modeling of hydrological processes at high temporal resolution to better understand flow patterns in catchments. To model surface and sub-surface flows in a catchment we utilized a physically based model called Hydrological Simulated Program-FORTRAN and two deep learning-based models. One deep learning model consisted of only one long short-term memory (simple LSTM), whereas the other model simulated processes in each hydrological response unit (HRU) by defining one separate LSTM for each HRU (HRU-based LSTM). The models use environmental time-series data and two-dimensional spatial data to predict surface and sub-surface flows at 6-minute time step simultaneously. We tested our models in a tropical humid headwater catchment in northern Lao PDR and compared their performances. Our results showed that the simple LSTM model outperformed the other models on surface runoff prediction with the lowest MSE (7.4e−5 m³ s⁻¹), whereas HRU-based LSTM model better predicted patterns and slopes in sub-surface flow in comparison with the other models by having the smallest MSE value (3.2e−4 m³ s⁻¹). This study demonstrated the performance of a deep learning model when simulating hydrological cycle with high temporal resolution.

Original language	English (US)
Article number	125370
Journal	Journal of Hydrology
Volume	590
DOIs	https://doi.org/10.1016/j.jhydrol.2020.125370
State	Published - Nov 2020

Bibliographical note

Funding Information:
This work was supported by Korea Environment Industry & Technology Institute (KEITI) through “The Chemical Accident Prevention Technology Development Project” funded by Korea Ministry of Environment (MOE) (No. 2016001970001). The authors sincerely thank the Lao Department of Agricultural Land Management (DALaM) for its support, including granting the permission for field access, and the M-TROPICS Critical Zone Observatory (https://mtropics.obs-mip.fr/), which belongs to the French Research Infrastructure OZCAR (http://www.ozcar-ri.org/), for data access.

Funding Information:
This work was supported by Korea Environment Industry & Technology Institute ( KEITI ) through “The Chemical Accident Prevention Technology Development Project” funded by Korea Ministry of Environment ( MOE ) (No. 2016001970001 ). The authors sincerely thank the Lao Department of Agricultural Land Management (DALaM) for its support, including granting the permission for field access, and the M-TROPICS Critical Zone Observatory (https://mtropics.obs-mip.fr/), which belongs to the French Research Infrastructure OZCAR (http://www.ozcar-ri.org/), for data access.

Publisher Copyright:
© 2020 Elsevier B.V.

Keywords

Deep learning model
Hydrological Simulated Program-FORTRAN
Long short-term memory (LSTM)
Sub-surface flow
Surface runoff

ASJC Scopus subject areas

Water Science and Technology

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1016/j.jhydrol.2020.125370

Cite this

@article{1dc9db7556284418a8db5a1cb1ad8052,

title = "Surface and sub-surface flow estimation at high temporal resolution using deep neural networks",

abstract = "Recent intensification in climate change have resulted in the rise of hydrological extreme events. This demands modeling of hydrological processes at high temporal resolution to better understand flow patterns in catchments. To model surface and sub-surface flows in a catchment we utilized a physically based model called Hydrological Simulated Program-FORTRAN and two deep learning-based models. One deep learning model consisted of only one long short-term memory (simple LSTM), whereas the other model simulated processes in each hydrological response unit (HRU) by defining one separate LSTM for each HRU (HRU-based LSTM). The models use environmental time-series data and two-dimensional spatial data to predict surface and sub-surface flows at 6-minute time step simultaneously. We tested our models in a tropical humid headwater catchment in northern Lao PDR and compared their performances. Our results showed that the simple LSTM model outperformed the other models on surface runoff prediction with the lowest MSE (7.4e−5 m3 s−1), whereas HRU-based LSTM model better predicted patterns and slopes in sub-surface flow in comparison with the other models by having the smallest MSE value (3.2e−4 m3 s−1). This study demonstrated the performance of a deep learning model when simulating hydrological cycle with high temporal resolution.",

keywords = "Deep learning model, Hydrological Simulated Program-FORTRAN, Long short-term memory (LSTM), Sub-surface flow, Surface runoff",

author = "Ather Abbas and Sangsoo Baek and Minjeong Kim and Mayzonee Ligaray and Olivier Ribolzi and Norbert Silvera and Min, {Joong Hyuk} and Laurie Boithias and Cho, {Kyung Hwa}",

note = "Funding Information: This work was supported by Korea Environment Industry & Technology Institute (KEITI) through “The Chemical Accident Prevention Technology Development Project” funded by Korea Ministry of Environment (MOE) (No. 2016001970001). The authors sincerely thank the Lao Department of Agricultural Land Management (DALaM) for its support, including granting the permission for field access, and the M-TROPICS Critical Zone Observatory (https://mtropics.obs-mip.fr/), which belongs to the French Research Infrastructure OZCAR (http://www.ozcar-ri.org/), for data access. Funding Information: This work was supported by Korea Environment Industry & Technology Institute ( KEITI ) through “The Chemical Accident Prevention Technology Development Project” funded by Korea Ministry of Environment ( MOE ) (No. 2016001970001 ). The authors sincerely thank the Lao Department of Agricultural Land Management (DALaM) for its support, including granting the permission for field access, and the M-TROPICS Critical Zone Observatory (https://mtropics.obs-mip.fr/), which belongs to the French Research Infrastructure OZCAR (http://www.ozcar-ri.org/), for data access. Publisher Copyright: {\textcopyright} 2020 Elsevier B.V.",

year = "2020",

month = nov,

doi = "10.1016/j.jhydrol.2020.125370",

language = "English (US)",

volume = "590",

journal = "Journal of Hydrology",

issn = "0022-1694",

publisher = "Elsevier",

}

TY - JOUR

T1 - Surface and sub-surface flow estimation at high temporal resolution using deep neural networks

AU - Abbas, Ather

AU - Baek, Sangsoo

AU - Kim, Minjeong

AU - Ligaray, Mayzonee

AU - Ribolzi, Olivier

AU - Silvera, Norbert

AU - Min, Joong Hyuk

AU - Boithias, Laurie

AU - Cho, Kyung Hwa

N1 - Funding Information: This work was supported by Korea Environment Industry & Technology Institute (KEITI) through “The Chemical Accident Prevention Technology Development Project” funded by Korea Ministry of Environment (MOE) (No. 2016001970001). The authors sincerely thank the Lao Department of Agricultural Land Management (DALaM) for its support, including granting the permission for field access, and the M-TROPICS Critical Zone Observatory (https://mtropics.obs-mip.fr/), which belongs to the French Research Infrastructure OZCAR (http://www.ozcar-ri.org/), for data access. Funding Information: This work was supported by Korea Environment Industry & Technology Institute ( KEITI ) through “The Chemical Accident Prevention Technology Development Project” funded by Korea Ministry of Environment ( MOE ) (No. 2016001970001 ). The authors sincerely thank the Lao Department of Agricultural Land Management (DALaM) for its support, including granting the permission for field access, and the M-TROPICS Critical Zone Observatory (https://mtropics.obs-mip.fr/), which belongs to the French Research Infrastructure OZCAR (http://www.ozcar-ri.org/), for data access. Publisher Copyright: © 2020 Elsevier B.V.

PY - 2020/11

Y1 - 2020/11

N2 - Recent intensification in climate change have resulted in the rise of hydrological extreme events. This demands modeling of hydrological processes at high temporal resolution to better understand flow patterns in catchments. To model surface and sub-surface flows in a catchment we utilized a physically based model called Hydrological Simulated Program-FORTRAN and two deep learning-based models. One deep learning model consisted of only one long short-term memory (simple LSTM), whereas the other model simulated processes in each hydrological response unit (HRU) by defining one separate LSTM for each HRU (HRU-based LSTM). The models use environmental time-series data and two-dimensional spatial data to predict surface and sub-surface flows at 6-minute time step simultaneously. We tested our models in a tropical humid headwater catchment in northern Lao PDR and compared their performances. Our results showed that the simple LSTM model outperformed the other models on surface runoff prediction with the lowest MSE (7.4e−5 m3 s−1), whereas HRU-based LSTM model better predicted patterns and slopes in sub-surface flow in comparison with the other models by having the smallest MSE value (3.2e−4 m3 s−1). This study demonstrated the performance of a deep learning model when simulating hydrological cycle with high temporal resolution.

AB - Recent intensification in climate change have resulted in the rise of hydrological extreme events. This demands modeling of hydrological processes at high temporal resolution to better understand flow patterns in catchments. To model surface and sub-surface flows in a catchment we utilized a physically based model called Hydrological Simulated Program-FORTRAN and two deep learning-based models. One deep learning model consisted of only one long short-term memory (simple LSTM), whereas the other model simulated processes in each hydrological response unit (HRU) by defining one separate LSTM for each HRU (HRU-based LSTM). The models use environmental time-series data and two-dimensional spatial data to predict surface and sub-surface flows at 6-minute time step simultaneously. We tested our models in a tropical humid headwater catchment in northern Lao PDR and compared their performances. Our results showed that the simple LSTM model outperformed the other models on surface runoff prediction with the lowest MSE (7.4e−5 m3 s−1), whereas HRU-based LSTM model better predicted patterns and slopes in sub-surface flow in comparison with the other models by having the smallest MSE value (3.2e−4 m3 s−1). This study demonstrated the performance of a deep learning model when simulating hydrological cycle with high temporal resolution.

KW - Deep learning model

KW - Hydrological Simulated Program-FORTRAN

KW - Long short-term memory (LSTM)

KW - Sub-surface flow

KW - Surface runoff

UR - http://www.scopus.com/inward/record.url?scp=85089223073&partnerID=8YFLogxK

U2 - 10.1016/j.jhydrol.2020.125370

DO - 10.1016/j.jhydrol.2020.125370

M3 - Article

AN - SCOPUS:85089223073

SN - 0022-1694

VL - 590

JO - Journal of Hydrology

JF - Journal of Hydrology

M1 - 125370

ER -

Surface and sub-surface flow estimation at high temporal resolution using deep neural networks

Abstract

Bibliographical note

Keywords

ASJC Scopus subject areas

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this