Comparison of the Information Content of Data from the LANDSAT-4 Thematic Mapper And the Multispectral Scanner

Abstract

Simultaneous data acquisition by the Landsat-4 Thematic Mapper (TM) and the Multispectral Scanner (MSS) permits the comparison of the two types of image data with respect to engineering performance and data applications. In this paper the "information" contained in data from the visible and near-IR channels is evaluated for five agricultural scenes, leading to the conclusion that the TM provides a significant advance in information gathering capability as expressed in terms of bits per pixel or bits per unit area. The six reflective channels of the TM acquire 18 bits of information per pixel out of a possible 6 × (8 bits) = 48 bits, while the four MSS channels acquire 10 bits of information per pixel out of a possible 4 × (7 bits) = 28 bits. Thus the TM and MSS are equally efficient in gathering information (18/48 = 10/28), contrary to the expected tendency toward lower efficiency as spatial resolution is improved and spectral channels are added to an observing system. The result is attributed to: 1) Superior selection of spectral channels in the TM; 2) Higher precision of the TM data, i.e., lower system noise, and 3) the advantage of higher spatial resolution, even in agricultural areas where fields are larger than the MSS pixel size. Because the MSS lacks a thermal IR channel, the 10-12-micrometer data of the TM at 120-m resolution are analyzed theoretically using an energy balance approach.