What question did this study set out to answer?

The study aims to analyze land use/land cover dynamics and assess blue carbon storage in the Mahakam River Delta.

January 24, 2026Open Access

Remote sensing-based analysis of land use/land cover dynamics and blue carbon assessment in Mahakam River Delta, Indonesia

Puntos clave

The study aims to analyze land use/land cover dynamics and assess blue carbon storage in the Mahakam River Delta.
Utilized Google Earth Engine for land use/land cover analysis.
Employed Sentinel-2A satellite imagery and machine learning techniques for classification.
Applied InVEST methodology to estimate different carbon components including aboveground and soil organic carbon.
Assessed machine learning models like Random Forest and CatBoost Regression for better accuracy in carbon estimation.
Achieved an average Kappa coefficient of 0.8766 and overall accuracy of 91.07% using Random Forest.
Identified soil organic carbon (SOC) as the largest component of total carbon.
Mangrove forests were determined as the primary contributors to blue carbon storage.
CatBoost Regression outperformed other methods for estimating total carbon, while Support Vector Regression excelled in belowground and dead organic carbon estimations.

Resumen

Abstract Climate change represents the greatest global challenge confronting the entire planet. Increasing temperatures and greenhouse gas emissions jeopardize human survival and ecological equilibrium. Coastal habitats possess significant potential for carbon absorption and storage, with a capacity up to 100 times greater than that of terrestrial ecosystems. Nonetheless, human-induced pressures and extensive land conversion have resulted in the destruction of coastal habitats, including those in the Mahakam River Delta. This has led to a reduction in blue carbon storage capacity, jeopardizing its role in climate change mitigation. This study seeks to examine land use/land cover (LULC) dynamics utilizing Google Earth Engine (GEE) and to assess carbon sequestration employing the Integrated Valuation of Ecosystem Services and Trade-offs (InVEST) methodology across four components: aboveground carbon (AGC), belowground carbon (BGC), soil organic carbon (SOC), and dead organic matter carbon (DOC). The LULC categorization was performed utilizing Sentinel-2A Multi-Spectral Instrument (MSI) satellite imagery and machine learning techniques, demonstrating that Random Forest (RF) achieving the best performance, yielding an average Kappa coefficient of 0.8766 and an average overall accuracy (OA) of 91.07%, respectively. Estimates of carbon storage using InVEST indicated that SOC is the predominant component of total carbon, with mangrove forests being the primary contributor. Estimation methods based on machine learning indicated that CatBoost Regression (CBR) was superior for estimating TC, AGC, and SOC, and Support Vector Regression (SVR) excelled in estimating BGC and DOC. This study's results are anticipated to facilitate data-driven decision-making in coastal ecosystem conservation and climate change mitigation initiatives. Graphical Abstract

Leer artículo completoexternamente

Preguntar a la IA

Me gusta

Guardar

Ver artículo completo