H/A/α Polarimetric Decomposition Of Dual Polarized Alos Palsar For Efficient Land Feature Detection And Biomass Estimation Over Tropical Deciduous Forest

• Published in: GEOGRAPHY, ENVIRONMENT, SUSTAINABILITY
• Main Author: Suman Sinha
• Other Authors: The author sincerely thanks JAXA (Japan Aerospace Exploration Agency, Japan) for providing the ALOS PALSAR data and SAC (Space Application Centre, ISRO, Ahmedabad, India) for providing facilities to process the PALSAR data
• Format: Article in Journal/Newspaper
• Language: English
• Published: Russian Geographical Society 2022
• Subjects: biomass; ALOS PALSAR; polarimetric decomposition; LULC; Sibirica
• Online Access: https://ges.rgo.ru/jour/article/view/2604; https://doi.org/10.24057/2071-9388-2021-095

Information regarding land use and land cover is an important for formulating decision making for land information system. The easiest and most effective way to gather such information is via using Earth observation remote sensing satellites supported by ground data. Synthetic Aperture Radar (SAR), due to its additional unique intrinsic characteristics is favoured over the optical systems for procuring land information. An innovative and effective technique for land feature detection is the use of polarimetric capabilities of SAR. Generally applicable for quad polarized data, this study investigates the polarimetric capabilities of a dual polarized data obtained from ALOS PALSAR, which is not a general notion. The approach applied in the study shows accurate results for detection of land features using polarimetric decomposition of dual polarized ALOS PALSAR data over an area of Munger in the state of Bihar, India. Twelve distinct land cover features are identified in the study area using this approach. The polarimetric products are also investigated for deriving the biomass information for the vegetation cover in the study area. The relation between in-situ biomass generated from floral species-specific volumetric equations and SAR polarimetric products showed a moderate correlation of 0.56 with RMSE=29.13 t/ha and data agreement of 0.62 based on exponential regression model for predicting biomass. The decomposition parameters revealed more evidences for forest structure and feature identification rather than biomass information. The method adopted in the study can be well utilized for land resource information and mapping; hence, natural and man-made resource monitoring and management.