Incorporating spectral data into logistic regression model to classify land cover: a case study in Mt. Qomolangma (Everest) National Nature Preserve
Abstract:Only few models for land-cover classification incorporated spectral data into ordinary logistic regression (OL model) in the Mt. Qomolangma (Everest) National Nature Preserve (QNNP) in China. In this study, spectral variables were incorporated into OL model and autologistic regression (AL) model to classify six main land covers. Twelve environmental variables and seven spectral variables of 10,000 stratified random sites in the QNNP were quantified and analyzed; OL model, AL model, OL model with spectral data (OLM model), and AL model with spectral data (ALM model) were estimated. The OLM and ALM models produced better estimates of regression coefficients and significantly improved model performance and overall accuracy for the grassland, sparsely vegetated land, and bare land compared with OL and AL models.
Document Type: Research Article
Affiliations: Department of Land-Use/Land-Cover Change and Land Resources Research,Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing, PR China
Publication date: October 1, 2012