Classification and Carbon-Stock Estimation of Mangroves in Dongzhaigang Based on Multi-Source Remote Sensing Data Using Google Earth Engine
Changes in carbon stock are a key indicator for assessing the carbon-pool function and the impact of regional carbon cycling on climate. Mangroves, as an essential component of coastal ecosystems, play a critical role in carbon sequestration. However, traditional carbon-sink assessments often overlo...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-03-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/17/6/964 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Changes in carbon stock are a key indicator for assessing the carbon-pool function and the impact of regional carbon cycling on climate. Mangroves, as an essential component of coastal ecosystems, play a critical role in carbon sequestration. However, traditional carbon-sink assessments often overlook biophysical differences between mangrove species and depend on field measurements, which are constrained by the challenging geographical conditions of the intertidal zone. This leads to biases and operational difficulties in estimating carbon stock. To address these challenges, this study proposes a fine-scale method for inter-species carbon-stock assessment, integrating mangrove three-dimensional structural information and spectral characteristics through Google Earth Engine (GEE). By combining GEDI data and Sentinel-2 imagery, this approach incorporates both vertical structure and spectral characteristics, overcoming the limitations of traditional models that neglect inter-species differences and vertical structural information. As a result, the accuracy of carbon-stock estimation is significantly improved. Unlike previous studies, this paper achieves a fully remote sensing-based partial carbon-stock assessment for mangrove species and quantifies carbon stock using the InVEST model, addressing the limitations of previous carbon-sink models. Specifically, on the GEE platform, Sentinel-2 imagery is used for inter-species classification through a random forest (RF) model, while a relationship model between canopy height and biomass is established using GEDI data to estimate biomass. To optimize feature selection, this study introduces a forward feature selection (FFS) approach, which incrementally selects the most predictive features, enhancing the stability and accuracy of the model. By combining biomass and classification results, a remote sensing-based carbon-sink assessment for mangroves is achieved. The study quantifies and visualizes the carbon stock of different mangrove species in Dongzhaigang, revealing that the region’s annual carbon stock totals 302,558.77 t. This validates the superiority and accuracy of the proposed method. |
|---|---|
| ISSN: | 2072-4292 |