Digital Numbers, Reflectance, and Atmospheric Correction in Remote Sensing


Abstract

This study focuses on the radiometric transformation of two Landsat images (the first is an ETM+ image acquired on September 23, 2001 and the second is a TM image from June 17, 2003) covering portions of the foothills and mountains west of Calgary, Alberta. A total of four radiometric transformation techniques; viz.converting raw DNs to radiance, converting radiance to TOA reflectance, absolute atmospheric correction using PCI‟s atmospheric correction package and atmospheric normalization via relative atmospheric correction using the liner transformation method, were used on the images.The accuracy assessment suggests that for our study site, atmospheric normalization via relative atmospheric correction using the linear transformation method is the most effective with the RMSE value of 9.5955 and absolute atmospheric correction using PCI‟s atmospheric correction package is the least effective with the RMSE value of 15.0371.