Refined empirical line approach for retrieving surface reflectance from EO-1 ALI images

The refined empirical line (REL) approach was used to convert the Earth Observing 1 (EO-1) Advanced Land Imager (ALI) sensor digital number (dn) to surface spectral reflectance ((rho)/sub (lambda)/). The dn-to-(rho)/sub (lambda)/ relation was derived from a bright target of known reflectance in the image, and the modeled estimates of the image dn at (rho)/sub (lambda)/=0. The mean absolute percent difference ((delta)/sub %/) between (rho)/sub (lambda)/ retrieved from ALI using the REL approach and ground-measured (rho)/sub (lambda)/ for 15 targets on six dates were 42%, 6%, and 13% in the ALI visible, near-infrared (NIR), and shortwave infrared (SWIR) spectral bands, respectively. The (delta)/sub %/ for (rho)/sub (lambda)/ retrieved from ALI without any atmospheric correction were 155%, 9%, and 10% for visible, NIR, and SWIR bands, respectively. For the clear, dry atmospheric conditions in Arizona, REL correction was most crucial for the dark targets in the visible bands. Given the published values of an ALI dn for (rho)/sub (lambda)/=0, the REL offers a simple approach for retrieving reflectance from multiple ALI images for temporal surface analysis.