TY - JOUR
T1 - Monitoring the Distribution of Mangrove Area using Synthetic Aperture Radar (SAR) and Optic Remote Sensing Data Fusion based on Deep Learning in Kotabaru Regency, Indonesia
AU - Sudiana, Dodi
AU - Nisa, Jamilatun
AU - Rizkinia, Mia
AU - Dimyati, Ratih Dewanti
AU - Anggraini, Nanin
AU - Riyanto, Indra
AU - Prabuwono, Anton Satria
AU - Sumantyo, Josaphat Tetuko Sri
N1 - Publisher Copyright:
© 2024 Joint Journal of Novel Carbon Resource Sciences and Green Asia Strategy. All rights reserved.
PY - 2024/3
Y1 - 2024/3
N2 - Mangrove plants are crucial in providing ecosystem benefits that align with the Sustainable Development Goals (SDGs), particularly regarding climate regulation (SDG 13), due to their efficient carbon storage capabilities. Ensuring the preservation and effective management of this valuable natural resource requires precise mapping and monitoring systems. While mapping techniques using optical and radar remote sensing data have been utilized for monitoring mangrove areas, traditional methods for detecting mangrove damage face limitations concerning accuracy, efficiency, and automation. This research records a novel approach to mapping and monitoring mangrove ecosystems in Kotabaru Regency, South Kalimantan, from 2017 to 2021, using a fusion of Sentinel-1 and Sentinel-2 remote sensing satellite imagery data. The study demonstrates that the optimal combination of single-date Sentinel-1 and Sentinel-2 image inputs for mangrove identification in the deep learning model involves fusing Sentinel-2's original five-band data (Red, Green, Blue, NIR, and SWIR-1), four multispectral indices (FDI, MDI, MNDWI, and WFI), and Sentinel-1 SAR data (VV and VH). This fusion yields impressive performance metrics, including Overall Accuracy (OA) of 95.60%, Intersection over Union (IoU) of 93.09%, and F1-Score of 96.42%. Furthermore, the proposed optimal combination is utilized in this study to analyze the spatial-temporal dynamics of mangrove habitats in the study area every year from 2017 to 2021. The results reveal that the largest mangrove area in the study region was recorded as 8,240.06 hectares in 2019, while the smallest area was 7,069.68 hectares in 2020. This study demonstrates the potential of the proposed method as a valuable tool for accurate and efficient mangrove monitoring, providing critical information for effective conservation and management efforts.
AB - Mangrove plants are crucial in providing ecosystem benefits that align with the Sustainable Development Goals (SDGs), particularly regarding climate regulation (SDG 13), due to their efficient carbon storage capabilities. Ensuring the preservation and effective management of this valuable natural resource requires precise mapping and monitoring systems. While mapping techniques using optical and radar remote sensing data have been utilized for monitoring mangrove areas, traditional methods for detecting mangrove damage face limitations concerning accuracy, efficiency, and automation. This research records a novel approach to mapping and monitoring mangrove ecosystems in Kotabaru Regency, South Kalimantan, from 2017 to 2021, using a fusion of Sentinel-1 and Sentinel-2 remote sensing satellite imagery data. The study demonstrates that the optimal combination of single-date Sentinel-1 and Sentinel-2 image inputs for mangrove identification in the deep learning model involves fusing Sentinel-2's original five-band data (Red, Green, Blue, NIR, and SWIR-1), four multispectral indices (FDI, MDI, MNDWI, and WFI), and Sentinel-1 SAR data (VV and VH). This fusion yields impressive performance metrics, including Overall Accuracy (OA) of 95.60%, Intersection over Union (IoU) of 93.09%, and F1-Score of 96.42%. Furthermore, the proposed optimal combination is utilized in this study to analyze the spatial-temporal dynamics of mangrove habitats in the study area every year from 2017 to 2021. The results reveal that the largest mangrove area in the study region was recorded as 8,240.06 hectares in 2019, while the smallest area was 7,069.68 hectares in 2020. This study demonstrates the potential of the proposed method as a valuable tool for accurate and efficient mangrove monitoring, providing critical information for effective conservation and management efforts.
KW - deep learning
KW - image fusion
KW - mangrove monitoring
KW - Sentinel-1
KW - Sentinel-2
KW - Synthetic Aperture Radar (SAR)
UR - http://www.scopus.com/inward/record.url?scp=85189155021&partnerID=8YFLogxK
M3 - Article
AN - SCOPUS:85189155021
SN - 2189-0420
VL - 11
SP - 536
EP - 546
JO - Evergreen
JF - Evergreen
IS - 1
ER -