A method for estimating particulate organic carbon at the sea surface based on geodetector and machine learning

Aug 5, 2026

Particulate organic carbon (POC) plays a key role in the marine carbon pump, and estimating its spatial patterns is essential for understanding the ocean carbon cycle. The study analyzes the spatial heterogeneity of 30 environmental factors and POC concentrations using geodetector to capture nonlinear relationships. Twenty factors—such as sea surface temperature, sea surface salinity, and chlorophyll-a—are selected and modeled with six machine-learning algorithms. Random forest achieves the best overall accuracy, enabling global sea-surface POC estimates for 2007–2016 and improved performance in coastal regions compared with other products.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages17

Reading Time43 min

AccessFree to download

Introduction

Background and significance of POC

Remote sensing and the challenge of direct retrieval

Prior estimation approaches using optical properties

Empirical relationships between POC and particle backscattering

TYPE Original Research PUBLISHED 28 December 2023 DOI 10.3389/fmars.2023.1295874 OPEN ACCESS EDITED BY Haiyong Zheng, Ocean University of China, China REVIEWED BY Qibin Lao, Guangdong Ocean University, China Hui Zheng, Henan University, China *CORRESPONDENCE Long Cui [z22160008@s.upc.edu.cn](z22160008@s.upc.edu.cn) RECEIVED 17 September 2023 ACCEPTED 11 December 2023 PUBLISHED 28 December 2023 CITATION Wu H, Cui L, Wang L, Sun R and Zheng Z (2023) A method for estimating particulate organic carbon at the sea surface based on geodetector and machine learning. Front. Mar. Sci. 10:1295874 . doi: 10.3389/fmars.2023.1295874 COPYRIGHT © 2023 Wu, Cui, Wang, Sun and Zheng. This isan open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. A method for estimating particulate organic carbon atthe sea surface based on geodetector and machine learning Huisheng Wu, Long Cui*, Lejie Wang, Ruixue Sun and Zhi Zheng College of Oceanography and Space Informatics, China University of Petroleum (East China), Qingdao, Shandong, China Particulate organic carbon (POC) is an essential component of the carbon pump within marine organisms. Exploring estimation methods for POC holds substantial signiﬁcance for understanding the marine carbon cycle. In this study, we investigated the spatial heterogeneity of 30 factors and POC concentrations using geodetector to account for nonlinearity, diversity, and complexity. Ultimately, 20 factors including sea surface temperature, sea surface salinity, and chlorophyll-a were selected as modeling variables. Six machine learning models—backpropagation neural network, convolutional neural network, attention-based neural network, random forest (RF), adaptive boosting, and extreme gradient boosting were used to compare their performance. The results indicate that among the six machine learning algorithms, RF exhibits the strongest performance, with a root mean square error of 0 . 11 [log(mg/m3)] and an average percentage deviation of 2. 73% . Global annual average sea surface POC concentrations were estimated for 2007 and compared to NASA’s POC product. The outcomes indicate that the RF model-based estimation method displays enhanced accuracy in estimating POC concentrations within intricate coastal environments, while the backpropagation neural network performed better in estimating POC concentrations in open ocean areas. Leveraging the RF model, global sea surface POC concentrations were estimated for the years 2007 through 2016, enabling a spatiotemporal analysis. The analysis unveils heightened POC concentrations in coastal regions and lower levels in open ocean areas. Furthermore, POC concentrations were greater in high-latitude regions compared to mid and low latitude counterparts. In conclusion, the global sea surface POC product in this study exhibits heightened spatial resolution and improved data completeness in contrast to other products. It enhances the accuracy of conventional POC estimation methods, particularly within coastal regions. KEYWORDS particulate organic carbon, machine learning, geodetector, ocean remote sensing, random forest Frontiers in Marine Science 01 [frontiersin.org](frontiersin.org) 1 Introduction Marine particulate organic carbon (POC) refers to the organic particles in the ocean that are generated through the metabolic processes of marine organisms, resuspension of sediments, and input from land sources. These particles include phytoplankton cells, bacteria, and organic debris, among other substances (Brewin et al., 2021) . POC accounts for approximatel

FAQs

Why is estimating particulate organic carbon (POC) important for marine science?

POC is a crucial component of the ocean carbon pump and strongly influences carbon turnover and flux. Understanding its spatiotemporal variations helps clarify the marine carbon cycle.

How does the study handle nonlinear and complex relationships between environmental factors and POC?

Geodetector is used to analyze spatial heterogeneity across 30 factors and POC concentrations, helping select 20 modeling variables such as sea surface temperature, salinity, and chlorophyll-a.

Which machine-learning model performs best and what does it enable?

Random forest shows the strongest performance overall and supports global sea-surface POC estimation for 2007–2016, enabling spatiotemporal analysis with improved accuracy in coastal environments.

Aug 5, 2026

Upload by Olivia Brown