Indicating Saturation Limits of Multi-sensor Satellite Data in Estimating Aboveground Biomass of a Mangrove Forest

Carbon sequestration in aboveground biomass remains understudied because of the difficulties in conducting field observations and saturation of remote sensing datasets. This study aimed to address the challenges associated with estimating forest AGB using remote sensing data from a dense mangrove ecosystem that has reached saturation limits. A mangrove ecosystem can reach saturation limits when the vegetation density and AGB are so high that remote sensing instruments, such as radar and lidar, cannot accurately measure further increases in biomass because of the sensors' limited penetration and resolution capabilities. This study evaluated the potential and limitations of using dual-polarised microwave data from Sentinel-1A and PALSAR-2, as well as spectral reflectance data from Sentinel-2, to estimate the AGB of the Bhitarkanika Wildlife Sanctuary (BWS), which is the second largest mangrove site in India. Using stratified random sampling, 314 elementary sampling units of 20 m × 20 m (0.04 ha) were used to record the diameter at breast height, tree height, and stand density for estimating AGB. Different band combinations of multisensor datasets were utilised to identify the best predictor variables for estimating the AGB and their corresponding saturation limits. Sentinel-1A demonstrated saturation at 123 Mg/ha (R 2 = 0.17) for AGB using VV polarisation, followed by 93 Mg/ha (R 2 = 0.55), 91 Mg/ha (R 2 = 0.26), and 96 Mg/ha (R 2 = 0.17) for the three red-edge bands at wavelengths of 705, 749, and 783 nm, respectively, of Sentinel-2 data. Red-edge bands are sensitive to chlorophyll and vegetation structure, and therefore, S2REP and REIP attained the maximum saturation limit at an AGB of 80 Mg/ha (R 2 = 0.26, 0.3, respectively). The highest correlation (R 2 = 0.9) and the maximum AGB of 326.06 Mg/ha was captured by the variable HH × HV of the L-band of the PALSAR-2 due to its penetration and interaction capacity with biomass components and less susceptible to interference from surface conditions and atmospheric effects. This study confirmed the advantages of longer-wavelength L-band data over C-band and multispectral optical bands for AGB estimation and identified the best predictor variables. This approach highlights the efficacy of different predictor variables for AGB estimation and the complementary strengths of multisensor datasets for navigating saturation limits. This framework could offer a practical guide for variable selection based on forest