Remote sensing and machine learning based approaches in mapping coastal ecosystem and quantifying carbon stock

Blue carbon (BC) ecosystems involve mangroves, sea grasses and saltmarshes which are all influential coastal resources providing diverse environmental goods and services. These ecosystems play a pivotal role in the mitigating climate change impacts and global carbon cycle. On a large scale, it is challenging to monitor these ecosystems, which include time-consuming field measurements, but a remote sensing tool makes this monitoring more efficient by offering faster and broader coverage. This review focuses on how remote sensing utilized for mapping and monitoring BC ecosystems. Particularly multispectral and hyperspectral data, proves to be the most common method for mapping, Landsat time-series data are widely utilized for monitoring changes on larger scales. Despite the effectiveness of remote sensing, also challenges that persisted, including cloud coverage, spectral limitations and errors in microwave SAR data. Recent advances in multispectral imagery, SAR imagery, LiDAR data and pilotless aircraft, coupled with image analysis techniques, enhance the ability to quantify BC stocks at larger scales. However, challenges such as removal of atmospheric effect, water related issues and limitations in training samples hinder accurate estimation. This article gives an overview of use of remote sensing data to monitor BC ecosystem and quantify carbon stocks in those ecosystems. Despite challenges, the integration of multi satellite data fusion with machine learning techniques holds promise for advancing the accurate quantification of BC stocks.