Modelling biophysical parameters of Indian Sundarbans mangroves using fully polarimetric L-band Synthetic Aperture Radar data and ground observations

• Published in: Journal of coastal conservation 2023-12, Vol.27 (6), p.62, Article 62
• Main Authors: Kumar, Tanumi, Das, Prabir Kumar, Dey, Mansa, Bandyopadhyay, Soumya, Chandrasekar, K., Das, Anup Kumar
• Format: Article
• Language: English
• Subjects: Accuracy; Backscatter; Backscattering; Biomass; Coastal ecosystems; Coastal Sciences; Coefficients; Datasets; Earth and Environmental Science; Forests; Geography; Independent variables; Mangrove swamps; Mangroves; Mathematical models; Nature Conservation; Oceanography; Parameters; Phased arrays; Polarimetry; Radar; Radar arrays; Radar data; Remote sensing systems; Remote Sensing/Photogrammetry; Root-mean-square errors; SAR (radar); Satellite observation; Sustainability management; Synthetic aperture radar
• ISSN: 1400-0350; 1874-7841
• DOI: 10.1007/s11852-023-00994-4

Mangroves are one of the important coastal ecosystems in the world and Sundarbans is one of the major blocks of mangrove ecosystem. Indian covers 40% of the total Sundarbans mangrove area, while the remaining 60% is located in Bangladesh. For mangrove forest assessment and their sustainable management, there is need to estimate the forest biophysical/ structural attributes. The present study investigates some of the biophysical properties of mangrove vegetation of south-western part of Indian Sundarbans using L-band fully polarimetric Advanced Land Observation Satellite-2 (ALOS-2) Phased Array L-band Synthetic Aperture Radar-2 (PALSAR-2) data and field observations. The backscattering coefficients for all mangrove communities of the study area were computed where the cross polarizations (HV and VH) showed better potentiality for discrimination of the communities. Biophysical parameters of mangrove forest trees were modelled using multiple linear regressions. Synthetic Aperture Radar backscattering coefficients and some of the biophysical parameters were used as independent variables for basal area (BA) and aboveground biomass (AGB) estimations with the highest model R square value of 0.632 and 0.717, respectively. The validation root mean square error (RMSE) achieved in case of BA was 3.66 m 2 /ha, while the lowest validation RMSE attained for AGB was 46.10 Mg/ha. The methodology presented in the work may be applied for the entire Sundarbans (including both the remaining Indian part and the Bangladesh mangroves).