Assessment of Forest Above Ground Biomass with SAR Data Using Linear Modeling in Joida Taluk, Uttara Kannada, India

The study was conducted in Joida taluk of Uttara Kannada district to assess the forest aboveground biomass using L band data. In this study, an attempt was made to estimate the aboveground biomass using SAR backscatter. The study area covered dense, moderately dense, and sparse forests. Nearly 0.01 percent of the forest area was sampled through 30 sampling plots. The point center quadrate (PCQ) method was used to select the tree and collected the tree growth parameters viz., tree height, diameter at breast height (DBH), and diameter at the tree base. The sampling plots were randomly selected in all types of forest and the location details such as the latitude, longitude, and altitude were collected from GPS. The tree crown density was measured with a densitometer. Each sample plot of forest aboveground biomass (AGB) was estimated using specific gravity and field-measured forest parameters. The fully polarimetric quad-pol (HH, HV, VV &VH) space borne SAR data of the Phased Array L-band Synthetic Aperture Radar-2 (PALSAR-2) of Advanced Land Observing Satellite-2 (ALOS-2) is used in this study for the natural forest of Joida Taluk of Uttara Kannada, Karnataka, India. The Orthorectification was performed on the radiometrically calibrated SAR data to measure the normalized radar cross-section of all the field-measured forest plot locations. SAR backscatter-based Multiple Linear Regression (MLR) model was implemented to retrieve forest aboveground biomass of the study area. The cross-polarization (HV) had shown a good correlation with forest above ground biomass. Forest Stem Volume varies from 16.11 m3/ha to1235.79m3/ha (12.88 t/ha to 1000.98 t/ha). The higher values are from the plots of dense forests having higher DBH and height. The sampled area included dense forest, moderately dense forest, and sparse forest and hence the variation in biomass has occurred. The Multiple Linear Regression (MLR) analysis was performed to estimate the aboveground biomass of the natural forest areas of the Joida taluk. This study used four combinations (HH &HV, VV &HH, HV & VH, VV&VH) of polarimetric channels in the forest aboveground biomass retrieval. Among them, the combination of HH and HV polarization shows a good correlation with field and predicted biomass. The predicted biomass from HH & HV polarisation varies from 79t/ha to 267t/ha, VV & VH shows the biomass from 62t/ha to 236t/ha, VV& HH showed from 62t/ha to 205 t/ha and HV&VH indicated from 79t/ha to 268t/ha. Among them the HH and HV polarization backscatter can be used to retrieve the aboveground biomass of forests though linear modeling. The RMSE and R2values that were obtained from the MLR for the polarimetric combinations HH & HV and HH & VV were 78 t/ha and 0.86, and 81 t/ha and 0.85 respectively. Forest AGB retrieval from HH & HV polarization-based MLR model has shown the best results. The results indicated that the backscatter from different polarisation could be used for developing the MLR model for estimating above ground biomass.