Evaluation of machine learning methods and multi-source remote sensing data combinations to construct forest above-ground biomass models

Aug 4, 2026

Rapid and accurate estimation of forest above-ground biomass (AGB) is crucial for sustainable forest management. Because field surveys are costly and difficult, multi-source remote sensing is used to improve modeled AGB predictions. Four machine learning approaches—Random Forest, Gradient Boosting Decision Tree, Classification and Regression Trees, and Minimum Distance—were trained with single and multi-source remote sensing variables on the Google Earth Engine platform for the Taiyue Mountain forest in Shanxi, China. Results indicate GBDT with spectral indices yields the highest predictive performance across coniferous, broadleaved, and mixed forests.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages46

Reading Time1 h 56 min

AccessFree to download

Abstract

Introduction

Forest AGB estimation and remote sensing rationale

Machine learning for improved large-area prediction

Materials and Methods

Study area and data sources

Google Earth Engine workflow

Machine learning models (RF, GBDT, CART, MD)

Results

Model performance across forest types

Effects of single vs multi-source inputs

Effects of bivariate variable combinations

Discussion

Implications for mixed-species forest modeling

Conclusions

PR IFYS GOL BANGOR / BANGOR Evaluation of machine learning methods and multi-source remote sensing data combinations to construct forest above-ground biomass models Yan, Xingguang; Li, Jing; Smith, Andy; Yang, Di; Ma, Tianyue; Su, Yiting; Shao, Jiahao International Journal of Digital Earth DOI: 10.1080/17538947.2023.2270459 Published: 01/11/2023 Peer reviewed version Cyswllt i'r cyhoeddiad / Link to publication Dyfyniad o'r fersiwn a gyhoeddwyd / Citation for published version (APA): Yan, X. , Li, J. , Smith, A. , Yang, D. , Ma, T. , Su, Y. , & Shao, J. (2023) . Evaluation of machine learning methods and multi-source remote sensing data combinations to construct forest aboveground biomass models. International Journal of Digital Earth, 16(2), Article 4471-4491. [https://doi.org/10.1080/17538947.2023.2270459](https://doi.org/10.1080/17538947.2023.2270459) Hawliau Cyffredinol / General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal ? Take down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. 1 Evaluation of machine learning methods and multi-source remote 2 sensing data combinations to construct forest above-ground biomass 3 models 4 5 Xingguang Yana,b,c, Jing Lia*, Andrew R. Smithb,c, Di Yangd, Tianyue 6 Maa, YiTing Sua and Jiahao Shaoa 7 a College of Geoscience and Surveying Engineering, China University of Mining and 8 Technology-Beijing, Beijing 100083, China; 9 bSchool of Natural Sciences, Bangor University, Bangor, Gwynedd, LL57 2UW, UK 10 cEnvironment Centre Wales, Bangor University, Bangor, Gwynedd, LL57 2UW, UK 11 d Wyoming Geographic Information Science Center, University of Wyoming, WY 12 82070, USA 13 14 *Author for Correspondence: Jing Li, Email: [lijing](lijing@cumtb.edu.cn)[@cumtb.edu.cn](lijing@cumtb.edu.cn) 15 ORCID 16 Xingguang Yan [https://orcid.org/0009-0001-8280-4568](https://orcid.org/0009-0001-8280-4568) 17 Andrew R. Smith [http://orchid.org/0000-0001-8580-278X](http://orchid.org/0000-0001-8580-278X) 18 Di Yang [http://orchid.org/0000-0002-4010-6163](http://orchid.org/0000-0002-4010-6163) 19 20 21 Abstract: Rapid and accurate estimation of forest biomass is essential to drive 22 sustainable management of forests. Field-based measurements of forest above- 23 ground biomass (AGB) can be costly and difficult to conduct. Multi-source 24 remote sensing data offers potential to improve the accuracy of modelled AGB 25 predictions. Here, four machine learning methods: Random Forest (RF), 26 Gradient Boosting Decision Tree (GBDT), Classification and Regression Trees 27 (CART) and Minimum Distance (MD) were used to construct forest AGB 28 models ofTaiyue Mountain forest, Shanxi Province, China using single and 29 multi-sourced remote sensing data and the Google Earth Engine platform. 30 Results showed that the machine learning method that most accurately 31 predicted AGB was GBDT and spectral index for coniferous (R2=0 .99; 32 RMSE=65 .52 Mg/ha), broadleaved (R2=0 .97; RMSE=29 . 14 Mg/ha), and 33 mixed species (R2=0 .97; RMSE=81 . 12 Mg/ha) forest types. Models 34 constructed using bivariate variable combinations that included the spectral 35 index improved the AGB estimation accuracy of mixed species (R2=0.99; 36 RMSE=59.52 Mg/ha) forest types

FAQs

Which machine learning method performed best for predicting forest AGB?

Gradient Boosting Decision Tree (GBDT) achieved the most accurate AGB predictions across coniferous, broadleaved, and mixed forest types.

How did multi-source remote sensing data affect AGB model accuracy?

Parameterizing machine learning algorithms with multi-source remote sensing variables improved prediction accuracy, particularly for mixed species forests.

What role did spectral index variables play in the models?

Models including bivariate variable combinations with spectral indices improved AGB estimation accuracy for mixed species forests and had smaller impacts for coniferous and broadleaved forests.

Aug 4, 2026

Upload by Lucas Martin