AI-powered spatiotemporal imputation and prediction of chlorophyll-a concentration in coastal ecosystems

Fan Zhang, Hiusuet Kung, Fa Zhang, Can Yang () and Jianping Gan ()
Additional contact information
Fan Zhang: The Hong Kong University of Science and Technology
Hiusuet Kung: The Hong Kong University of Science and Technology
Fa Zhang: The Hong Kong University of Science and Technology
Can Yang: The Hong Kong University of Science and Technology
Jianping Gan: The Hong Kong University of Science and Technology

Nature Communications, 2025, vol. 16, issue 1, 1-13

Abstract: Abstract Predicting spatiotemporal Chlorophyll-a (Chl_a) distributions is essential for diagnosing and analysing productivity and ecosystem health of coastal oceans. Yet, current tools remain inadequate for prognosing marine ecosystems through predicting spatiotemporal Chl_a distributions, particularly in the dynamic coastal ocean. Coupled physics-biogeochemical models struggle to resolve complex trophic interactions, while data-driven approaches are limited by incomplete satellite observations. We developed an advanced AI-powered spatiotemporal imputation and prediction (STIMP) model for predicting Chl_a in coastal ocean. STIMP adopts a novel paradigm that first imputes and subsequently predicts Chl_a across a broad spatiotemporal scale, resolving difficulties arising from incompletion, nonstationary temporal variations, and spatial heterogeneity of data through integrating specially designed modules. We demonstrated the STIMP’s robust imputation and prediction of Chl_a in four representative global coastal oceans. STIMP reduced the imputation mean absolute error (MAE) by 45.90–81.39% compared with the data interpolating empirical orthogonal function method in geoscience and by 8.92–43.04% against leading AI methods. With accurate imputation, STIMP demonstrated superior predictive accuracy, achieving MAE reductions of 58.99% over biogeophysical models and 6.54–13.68% over AI benchmarks. STIMP offers a new approach for predicting oceans’ Chl_a that typically have spatiotemporally limited data.

Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
https://www.nature.com/articles/s41467-025-62901-9 Abstract (text/html)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:nat:natcom:v:16:y:2025:i:1:d:10.1038_s41467-025-62901-9

Ordering information: This journal article can be ordered from
https://www.nature.com/ncomms/

DOI: 10.1038/s41467-025-62901-9

Access Statistics for this article

Nature Communications is currently edited by Nathalie Le Bot, Enda Bergin and Fiona Gillespie

More articles in Nature Communications from Nature
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().