Monitoring river ice with landsat images

Abstract

In the northern United States, ice can delay or stop river navigation in the winter and cause unexpected problems and emergencies. As part of a program to develop a river ice forecasting model, photointerpretation techniques were used to map the areal distributions of four classes of river ice along the navigable reaches of the Allegheny, Monongahela, and Ohio Rivers and the Illinois Waterway each winter from 1972 to 1985 from Landsat Multispectral Scanner (MSS Band 2, 0.6–0.7 μm), Thematic Mapper (TM Band 3, 0.63–0.69 μm), and Return Beam Vidicon (RBV, 0.580–0.680 μm and 0.505–0.750 μm) images. The four classes, 1) ice-free, 2) partial gray ice, 3) complete gray ice, and 4) white ice, were usually readily apparent on the images due to differences in gray tones produced by the various ice types and conditions that make up the different classes. Landsat-derived ice observations compared favorably with available ground and aerial observations 64–80% of the time. During mild winters without extensive and long-lasting ice, or along rivers which normally do not have such ice, Landsat images are less-useful because of the time gaps between cloud-or haze-free images. Yet for many rivers in cold regions, Landsat images may be the only source of data on river ice.