Iterative integration of deep learning in hybrid Earth surface system modelling
Publikation: Bidrag til tidsskrift › Tidsskriftartikel › Forskning › fagfællebedømt
Standard
Iterative integration of deep learning in hybrid Earth surface system modelling. / Chen, Min; Qian, Zhen; Boers, Niklas; Jakeman, Anthony J.; Kettner, Albert J.; Brandt, Martin; Kwan, Mei Po; Batty, Michael; Li, Wenwen; Zhu, Rui; Luo, Wei; Ames, Daniel P.; Barton, C. Michael; Cuddy, Susan M.; Koirala, Sujan; Zhang, Fan; Ratti, Carlo; Liu, Jian; Zhong, Teng; Liu, Junzhi; Wen, Yongning; Yue, Songshan; Zhu, Zhiyi; Zhang, Zhixin; Sun, Zhuo; Lin, Jian; Ma, Zaiyang; He, Yuanqing; Xu, Kai; Zhang, Chunxiao; Lin, Hui; Lü, Guonian.
I: Nature Reviews Earth & Environment, Bind 4, 2023, s. 568–581.Publikation: Bidrag til tidsskrift › Tidsskriftartikel › Forskning › fagfællebedømt
Harvard
APA
Vancouver
Author
Bibtex
}
RIS
TY - JOUR
T1 - Iterative integration of deep learning in hybrid Earth surface system modelling
AU - Chen, Min
AU - Qian, Zhen
AU - Boers, Niklas
AU - Jakeman, Anthony J.
AU - Kettner, Albert J.
AU - Brandt, Martin
AU - Kwan, Mei Po
AU - Batty, Michael
AU - Li, Wenwen
AU - Zhu, Rui
AU - Luo, Wei
AU - Ames, Daniel P.
AU - Barton, C. Michael
AU - Cuddy, Susan M.
AU - Koirala, Sujan
AU - Zhang, Fan
AU - Ratti, Carlo
AU - Liu, Jian
AU - Zhong, Teng
AU - Liu, Junzhi
AU - Wen, Yongning
AU - Yue, Songshan
AU - Zhu, Zhiyi
AU - Zhang, Zhixin
AU - Sun, Zhuo
AU - Lin, Jian
AU - Ma, Zaiyang
AU - He, Yuanqing
AU - Xu, Kai
AU - Zhang, Chunxiao
AU - Lin, Hui
AU - Lü, Guonian
N1 - Publisher Copyright: © 2023, Springer Nature Limited.
PY - 2023
Y1 - 2023
N2 - Earth system modelling (ESM) is essential for understanding past, present and future Earth processes. Deep learning (DL), with the data-driven strength of neural networks, has promise for improving ESM by exploiting information from Big Data. Yet existing hybrid ESMs largely have deep neural networks incorporated only during the initial stage of model development. In this Perspective, we examine progress in hybrid ESM, focusing on the Earth surface system, and propose a framework that integrates neural networks into ESM throughout the modelling lifecycle. In this framework, DL computing systems and ESM-related knowledge repositories are set up in a homogeneous computational environment. DL can infer unknown or missing information, feeding it back into the knowledge repositories, while the ESM-related knowledge can constrain inference results of the DL. By fostering collaboration between ESM-related knowledge and DL systems, adaptive guidance plans can be generated through question-answering mechanisms and recommendation functions. As users interact iteratively, the hybrid system deepens its understanding of their preferences, resulting in increasingly customized, scalable and accurate guidance plans for modelling Earth processes. The advancement of this framework necessitates interdisciplinary collaboration, focusing on explainable DL and maintaining observational data to ensure the reliability of simulations.
AB - Earth system modelling (ESM) is essential for understanding past, present and future Earth processes. Deep learning (DL), with the data-driven strength of neural networks, has promise for improving ESM by exploiting information from Big Data. Yet existing hybrid ESMs largely have deep neural networks incorporated only during the initial stage of model development. In this Perspective, we examine progress in hybrid ESM, focusing on the Earth surface system, and propose a framework that integrates neural networks into ESM throughout the modelling lifecycle. In this framework, DL computing systems and ESM-related knowledge repositories are set up in a homogeneous computational environment. DL can infer unknown or missing information, feeding it back into the knowledge repositories, while the ESM-related knowledge can constrain inference results of the DL. By fostering collaboration between ESM-related knowledge and DL systems, adaptive guidance plans can be generated through question-answering mechanisms and recommendation functions. As users interact iteratively, the hybrid system deepens its understanding of their preferences, resulting in increasingly customized, scalable and accurate guidance plans for modelling Earth processes. The advancement of this framework necessitates interdisciplinary collaboration, focusing on explainable DL and maintaining observational data to ensure the reliability of simulations.
U2 - 10.1038/s43017-023-00452-7
DO - 10.1038/s43017-023-00452-7
M3 - Journal article
AN - SCOPUS:85164681104
VL - 4
SP - 568
EP - 581
JO - Nature Reviews Earth & Environment
JF - Nature Reviews Earth & Environment
SN - 2662-138X
ER -
ID: 362060515