Investigating the performance of genetic algorithms selection method in estimating stand-level structural and biophysical variables of lowland dipterocarp forest from LiDAR data

9th IGRSM International Conference and Exhibition on Geospatial & Remote Sensing

地球科学;计算机科学

Misman, M.A.^1 ; Yaakub, S.Y.^1 ; Omar, H.^1

Geoinformation Programme, Forestry and Environment Division, Forest Research Institute Malaysia, Selangor, Kepong

52109, Malaysia^1

关键词: Airborne LiDAR;    Bio-physical variables;    Genetic algorithm (GAs);    Mean diameter;    Predictor variables;    Selection methods;    Stand densities;    Stepwise regression method;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/169/1/012110/pdf
DOI  :  10.1088/1755-1315/169/1/012110

学科分类：计算机科学（综合）

来源: IOP

【 摘 要 】

Genetic Algorithms (GAs) methods are rarely been used as a selection method in developing the model for estimating forest's variables as compared to stepwise regression method. This study aims to investigate the performance of GAs as a selection method in finding the best predictor variables. Five models were developed to test the performance of GAs methods in estimating stand-level structural and biophysical variables of the forest, namely as mean height (Hm), stand density (S), basal area (G) square mean diameter (Dg), and biomass (B). The results have shown GAs methods produce better model for Hm and Dg as compared to stepwise regression method in term of adjusted R2 and RMSE values. However, models for S, G and B based on stepwise regression method outperformed the GAs methods. This study has shown the capability of GAs in finding the best airborne lidar scanning (ALS) metrics for the development of best model in estimating stand-level structural and biophysical variables of lowland dipterocarp forest.