Use of logistic regression for validation of maps of the spatial distribution of vegetation species derived from high spatial resolution hyperspectral remotely sensed data

Abstract Logistic regression was used for validation of image-derived maps depicting large woody debris and three Populus spp. in a riparian area in Yellowstone National Park, USA. High spatial resolution hyperspectral imagery was used for data input. The logistic regression model related presence/absence field data for large woody debris and the Populus spp. to the continuous measurement scale output from matched filter image analysis. The agreement between the image analysis and field data was excellent, as measured by Model χ2, residual deviance, and Hosmer and Lemeshow's goodness of fit for the logistic regression. Kappa analysis for all possible thresholds imposed on the image output, and receiver operating characteristic curves, also indicated excellent goodness of fit for the image analysis output. The potential for use of logistic regression for both validation of hyperspectral image classification and calibration of results is discussed.