Mangrove forests play a crucial role in ecosystems by offering essential environmental services that benefit nearby communities. They act as cultural resources, offering ecotourism, education, and preserving cultural heritage. Moreover, mangroves safeguard the coastlines from waves and surges by absorbing their force and preventing erosion. Furthermore, they hold economic significance due to their ability to store blue carbon. Restoration initiatives can generate revenue through carbon markets, while conservation prevents potential carbon loss. Given the potential ecosystem services that mangrove forests can provide, the availability of reliable biomass information is crucial. Not only to support protection and conservation efforts but also to maintain the potential economic value of mangrove forest ecosystems. The remote sensing approach, especially with active systems such as Synthetic Aperture Radar (SAR), is an attractive option due to its ability to capture the Earth's surface regardless of the atmospheric condition. Nevertheless, the typical backscatter method loses its reliability as biomass levels increase beyond a certain threshold, resulting in incomplete data due to excessive signal strength that overwhelms the receiver. This study emphasized the potential utilization of scattering mechanism parameters from polarimetric decomposition on Sentinel-1 to estimate mangrove biomass in West Kalimantan Province, Indonesia, in 2020. This study integrated the decomposition parameters of entropy, anisotropy, and alpha angle from H/A/α decomposition with backscatter parameters as dependent variables. Along with biomass information derived from the Global Ecosystem Dynamics Investigation (GEDI) as independent variables. Utilizing machine learning approach provided by H2O autoML, correlations were established between SAR parameters and GEDI-derived biomass to develop a predictive model. Then, various configurations of SAR parameters (both individually and in groups) were assessed using statistical metrics, including R2, RMSE, MSE, and MAE. Additionally, comparisons were conducted with biomass prediction products from other projects to assess the model`s predictive accuracy. The information obtained will be useful for large-scale mangrove biomass mapping with an acceptable level of accuracy, which will be useful in mangrove forest protection and conservation activities.

Mangrove forests play a crucial role in ecosystems by offering essential environmental services that benefit nearby communities. They act as cultural resources, offering ecotourism, education, and preserving cultural heritage. Moreover, mangroves safeguard the coastlines from waves and surges by absorbing their force and preventing erosion. Furthermore, they hold economic significance due to their ability to store blue carbon. Restoration initiatives can generate revenue through carbon markets, while conservation prevents potential carbon loss. Given the potential ecosystem services that mangrove forests can provide, the availability of reliable biomass information is crucial. Not only to support protection and conservation efforts but also to maintain the potential economic value of mangrove forest ecosystems. The remote sensing approach, especially with active systems such as Synthetic Aperture Radar (SAR), is an attractive option due to its ability to capture the Earth's surface regardless of the atmospheric condition. Nevertheless, the typical backscatter method loses its reliability as biomass levels increase beyond a certain threshold, resulting in incomplete data due to excessive signal strength that overwhelms the receiver. This study emphasized the potential utilization of scattering mechanism parameters from polarimetric decomposition on Sentinel-1 to estimate mangrove biomass in West Kalimantan Province, Indonesia, in 2020. This study integrated the decomposition parameters of entropy, anisotropy, and alpha angle from H/A/α decomposition with backscatter parameters as dependent variables. Along with biomass information derived from the Global Ecosystem Dynamics Investigation (GEDI) as independent variables. Utilizing machine learning approach provided by H2O autoML, correlations were established between SAR parameters and GEDI-derived biomass to develop a predictive model. Then, various configurations of SAR parameters (both individually and in groups) were assessed using statistical metrics, including R2, RMSE, MSE, and MAE. Additionally, comparisons were conducted with biomass prediction products from other projects to assess the model`s predictive accuracy. The information obtained will be useful for large-scale mangrove biomass mapping with an acceptable level of accuracy, which will be useful in mangrove forest protection and conservation activities.

Integrating Sentinel-1 SAR Polarimetry for Mangrove Forest Biomass Estimation: A Case Study in West Kalimantan Province, Indonesia

GHIVARRY, GIUSTI
2023/2024

Abstract

Mangrove forests play a crucial role in ecosystems by offering essential environmental services that benefit nearby communities. They act as cultural resources, offering ecotourism, education, and preserving cultural heritage. Moreover, mangroves safeguard the coastlines from waves and surges by absorbing their force and preventing erosion. Furthermore, they hold economic significance due to their ability to store blue carbon. Restoration initiatives can generate revenue through carbon markets, while conservation prevents potential carbon loss. Given the potential ecosystem services that mangrove forests can provide, the availability of reliable biomass information is crucial. Not only to support protection and conservation efforts but also to maintain the potential economic value of mangrove forest ecosystems. The remote sensing approach, especially with active systems such as Synthetic Aperture Radar (SAR), is an attractive option due to its ability to capture the Earth's surface regardless of the atmospheric condition. Nevertheless, the typical backscatter method loses its reliability as biomass levels increase beyond a certain threshold, resulting in incomplete data due to excessive signal strength that overwhelms the receiver. This study emphasized the potential utilization of scattering mechanism parameters from polarimetric decomposition on Sentinel-1 to estimate mangrove biomass in West Kalimantan Province, Indonesia, in 2020. This study integrated the decomposition parameters of entropy, anisotropy, and alpha angle from H/A/α decomposition with backscatter parameters as dependent variables. Along with biomass information derived from the Global Ecosystem Dynamics Investigation (GEDI) as independent variables. Utilizing machine learning approach provided by H2O autoML, correlations were established between SAR parameters and GEDI-derived biomass to develop a predictive model. Then, various configurations of SAR parameters (both individually and in groups) were assessed using statistical metrics, including R2, RMSE, MSE, and MAE. Additionally, comparisons were conducted with biomass prediction products from other projects to assess the model`s predictive accuracy. The information obtained will be useful for large-scale mangrove biomass mapping with an acceptable level of accuracy, which will be useful in mangrove forest protection and conservation activities.
2023
Integrating Sentinel-1 SAR Polarimetry for Mangrove Forest Biomass Estimation: A Case Study in West Kalimantan Province, Indonesia
Mangrove forests play a crucial role in ecosystems by offering essential environmental services that benefit nearby communities. They act as cultural resources, offering ecotourism, education, and preserving cultural heritage. Moreover, mangroves safeguard the coastlines from waves and surges by absorbing their force and preventing erosion. Furthermore, they hold economic significance due to their ability to store blue carbon. Restoration initiatives can generate revenue through carbon markets, while conservation prevents potential carbon loss. Given the potential ecosystem services that mangrove forests can provide, the availability of reliable biomass information is crucial. Not only to support protection and conservation efforts but also to maintain the potential economic value of mangrove forest ecosystems. The remote sensing approach, especially with active systems such as Synthetic Aperture Radar (SAR), is an attractive option due to its ability to capture the Earth's surface regardless of the atmospheric condition. Nevertheless, the typical backscatter method loses its reliability as biomass levels increase beyond a certain threshold, resulting in incomplete data due to excessive signal strength that overwhelms the receiver. This study emphasized the potential utilization of scattering mechanism parameters from polarimetric decomposition on Sentinel-1 to estimate mangrove biomass in West Kalimantan Province, Indonesia, in 2020. This study integrated the decomposition parameters of entropy, anisotropy, and alpha angle from H/A/α decomposition with backscatter parameters as dependent variables. Along with biomass information derived from the Global Ecosystem Dynamics Investigation (GEDI) as independent variables. Utilizing machine learning approach provided by H2O autoML, correlations were established between SAR parameters and GEDI-derived biomass to develop a predictive model. Then, various configurations of SAR parameters (both individually and in groups) were assessed using statistical metrics, including R2, RMSE, MSE, and MAE. Additionally, comparisons were conducted with biomass prediction products from other projects to assess the model`s predictive accuracy. The information obtained will be useful for large-scale mangrove biomass mapping with an acceptable level of accuracy, which will be useful in mangrove forest protection and conservation activities.
PolSAR
Sentinel-1 SAR
GEDI
Random Forest
Mangrove
File in questo prodotto:
File Dimensione Formato  
GHIVARRY_GIUSTI.pdf

accesso aperto

Dimensione 3.44 MB
Formato Adobe PDF
3.44 MB Adobe PDF Visualizza/Apri

The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.12608/67492