Journal article · 2025 · Open access

A mathematical comparison of data assimilation and machine learning in earth system state estimation from a Bayesian inference viewpoint

Xin Li, Feng Liu

Information Geography, 1(1), 100001 · Published 19 February 2025 · DOI 10.1016/j.infgeo.2025.100001

Read the free full textCite this paper

In brief

“A mathematical comparison of data assimilation and machine learning in earth system state estimation from a Bayesian inference viewpoint” is a 2025 journal article by Xin Li and Feng Liu, published in Information Geography. OpenAlex counts 12 citations as of 16 November 2026, which puts it in the top 3% of works published in 2025 by citations. It is open access under a CC BY-NC-ND licence: the full text is free to read at Information Geography.

Citations
12
Most cited in
2026 (7)
Field-weighted impact
3.68
References
28
Authors
2

Abstract

Data Assimilation (DA) and Machine Learning (ML) possess complementary strengths that are vital to improving state estimation and prediction in Earth system science. However, comprehensive comparisons of their methodologies are rare and complex. This study explores DA and ML through a mathematical framework based on Bayesian inference. We establish a unified framework to address state estimation and prediction, deriving cost functions of DA and ML using maximum a posteriori estimation. While DA and ML share similarities in cost function structures, they diverge in principles and applications. DA is rooted in physical models, while ML excels with large datasets. Despite their similarities in motivations, distinctions arise in equations, algorithms, cost functions, parameters, uncertainty handling, and data requirements. Integrating DA's theoretical robustness with ML's data leveraging capabilities is essential for advancing state estimation and prediction. DA can bolster ML's interpretability, while ML can boost DA's operational efficiency. This paper contributes to the discourse on how best to integrate these methodologies for improving state estimation and prediction in Earth system science. • The state estimation of DA and ML is reformulated by maximizing posterior probability conditioned by prior observations. • DA and ML differ in their methodologies; DA incorporates Earth system dynamics while ML learns from dense data. • DA can enhance ML's interpretability and ML can increase DA's operationability.

Cite this paper

Formatted from the catalogue record. Check names, capitals and page numbers against the paper before you submit.

APA 7

Li, X., & Liu, F. (2025). A mathematical comparison of data assimilation and machine learning in earth system state estimation from a Bayesian inference viewpoint. Information Geography, 1(1), 100001. https://doi.org/10.1016/j.infgeo.2025.100001

MLA 9

Li, Xin, and Feng Liu. “A mathematical comparison of data assimilation and machine learning in earth system state estimation from a Bayesian inference viewpoint.” Information Geography, vol. 1, no. 1, 2025, p. 100001. https://doi.org/10.1016/j.infgeo.2025.100001.

Chicago

Li, Xin and Feng Liu. 2025. “A mathematical comparison of data assimilation and machine learning in earth system state estimation from a Bayesian inference viewpoint.” Information Geography 1 (1): 100001. https://doi.org/10.1016/j.infgeo.2025.100001.

Harvard

Li, X. and Liu, F. (2025) ‘A mathematical comparison of data assimilation and machine learning in earth system state estimation from a Bayesian inference viewpoint’, Information Geography, 1(1), p. 100001. doi:10.1016/j.infgeo.2025.100001.

IEEE

X. Li and F. Liu, “A mathematical comparison of data assimilation and machine learning in earth system state estimation from a Bayesian inference viewpoint,” Information Geography, vol. 1, no. 1, p. 100001, 2025, doi: 10.1016/j.infgeo.2025.100001.

BibTeX

@article{li2025mathematical,
  title     = {{A mathematical comparison of data assimilation and machine learning in earth system state estimation from a Bayesian inference viewpoint}},
  author    = {Li, Xin and Liu, Feng},
  journal   = {Information Geography},
  year      = {2025},
  volume    = {1},
  number    = {1},
  pages     = {100001},
  doi       = {10.1016/j.infgeo.2025.100001},
  url       = {https://doi.org/10.1016/j.infgeo.2025.100001}
}

Download for a reference manager: BibTeX (.bib) · RIS for Zotero, Mendeley and EndNote (.ris)

Topic and keywords

Physical Sciences › Earth and Planetary Sciences › Atmospheric Science › Meteorological Phenomena and Simulations

  • data assimilation
  • machine learning
  • Bayesian inference
  • maximum a posteriori estimation
  • cost function
  • uncertainty quantification
  • Earth system prediction
  • physical model

Work with this paper

Most-cited papers in Earth and Planetary Sciences

All highly cited Earth and Planetary Sciences papers

Questions about this paper

How do I cite this paper in APA style?

Li, X., & Liu, F. (2025). A mathematical comparison of data assimilation and machine learning in earth system state estimation from a Bayesian inference viewpoint. Information Geography, 1(1), 100001. https://doi.org/10.1016/j.infgeo.2025.100001

How many times has this paper been cited?

12 times, according to OpenAlex on 16 November 2026. Google Scholar and Scopus count different sets of sources, so their numbers differ.

Is this paper free to read?

Yes. It is open access: read the free full text at Information Geography.

Who wrote this paper?

Xin Li, Feng Liu.

What is the DOI of this paper?

10.1016/j.infgeo.2025.100001. The permanent link is https://doi.org/10.1016/j.infgeo.2025.100001.

Bibliographic data from OpenAlex (CC0), last checked 16 November 2026. ResearchGuru does not host this paper; the links go to the publisher or the repository that does.