Iterative integration of deep learning in hybrid Earth surface system modelling

Iterative integration of deep learning in hybrid Earth surface system modelling

Nature Reviews Earth & Environment

10.1038/s43017-023-00452-7

https://doi.org/10.1038/s43017-023-00452-7

Min Chen, Zhen Qian, Niklas Boers, Anthony J. Jakeman, Albert J. Kettner, Martin Brandt, Mei-Po Kwan, Michael Batty, Wenwen Li, Rui Zhu, Wei Luo, Daniel P. Ames, C. Michael Barton, Susan M. Cuddy, Sujan Koirala, Fan Zhang, Carlo Ratti, Jian Liu, Teng Zhong, Junzhi Liu, Yongning Wen, Songshan Yue, Zhiyi Zhu, Zhixin Zhang, Zhuo Sun, Jian Lin, Zaiyang Ma, Yuanqing He, Kai Xu, Chunxiao Zhang, Hui Lin, Guonian Lü

11 July 2023

Chen, M., Qian, Z., Boers, N., Jakeman, A. J., Kettner, A. J., Brandt, M., Kwan, M.-P., Batty, M., Li, W., Zhu, R., Luo, W., Ames, D. P., Barton, C. M., Cuddy, S. M., Koirala, S., Zhang, F., Ratti, C., Liu, J., Zhong, T., … Lü, G. (2023). Iterative integration of deep learning in hybrid Earth surface system modelling. Nature Reviews Earth & Environment, 4(8), 568–581. https://doi.org/10.1038/s43017-023-00452-7

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.

Publisher Copyright

There was no specific funding for the research done

This is a post-peer-review, pre-copyedit version of an article published in Nature Reviews Earth & Environment. The final authenticated version is available online at: http://dx.doi.org/10.1038/s43017-023-00452-7

https://oar.a-star.edu.sg/communities-collections/articles/22539

2662-138X

Institute of High Performance Computing