Genetic Algorithm Approach for Optimization of Biomass Estimation at LiDAR

Abstract

Estimation of Biomass at ICESat/GLAS footprint level was finished by incorporating information from various sensors viz., spaceborne LiDAR (ICESat/GLAS). The biomass estimation accuracies of Genetic Algorithm were studied by optimizing the waveform parameters. Multiple linear regression equation was generated using the most important variables found by Genetic Algorithm. The results of the study were very encouraging. Optimum results were obtained using the top 18 parameters derived from GLAS waveform. The biomass was predicted at small area by Genetic Algorithm with an R2 63% of and RMSE of 18.94 t/ha using the best six variables, viz. wdistance, wpdistance, R50, Ecanopy, Home., wcentroid over to 18 independent variables. The same methodology can be used for biomass estimation over a large area. The estimation is done at Tripura area. The study finally established that Genetic approach is produced the better result to predicting AGB. The best outcome of the study was the formulation of an approach that could result in higher biomass estimation accuracies.