We have developed an empirical algorithm for removing thin cirrus effects from AVIRIS (airborne visible infrared imaging spectrometer) images at any wavelength in the region 0.4--1 μm using narrow channels near 1.38, 1.24, and 0.66 μm. In this algorithm the imaging pair of 1.38 and 1.24 μm channels is used for cirrus removal over water surfaces, while the imaging pair of 1.38 and 0.66 μm channels is used over land surfaces. The AVIRIS data on which the algorithm is based were measured from an ER-2 aircraft at 20 km altitude during several field programs over both land and ocean. From these data we earlier discovered that narrow channels near the center of the strong 1.38-μm water vapor bands are very sensitive in detecting thin cirrus clouds. The present paper is a logical continuation of that work. Our new algorithm has considerable generality. As a first application, it can be incorporated as a front end to the first-generation moderate resolution imaging spectrometer (MODIS) (on EOS AM and PM platforms) atmospheric correction algorithms for ocean color and land applications in order to yield improved atmospheric aerosol, land surface, and ocean color products. ¿ 1998 American Geophysical Union