IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 42, NO. 12, DECEMBER 2004 2691 Landsat Sensor Performance: History and Current Status Brian L. Markham, Member, IEEE, James C. Storey, Member, IEEE, Darrel L. Williams, and James R. Irons Abstract—The current Thematic Mapper (TM) class of Landsat sensors began with Landsat-4, which was launched in 1982. This series continued with the nearly identical sensor on Landsat-5, launched in 1984. The final sensor in the series was the Landsat-7 Enhanced Thematic Mapper Plus (ETM+), which was carried into orbit in 1999. Varying degrees of effort have been devoted to the characterization of these instruments and data over the past 22 years. Extensive short-lived efforts early in the history, very limited efforts in the middle years, and now a systematic program for continuing characterization of all three systems are apparent. Currently, both the Landsat-5 TM and the Landsat-7 ETM+ are operational and providing data. Despite 20+ years of operation, the TM on Landsat-5 is fully functional, although downlinks for the data are limited. Landsat-7 ETM+ experienced a failure of its Scan Line Corrector mechanism in May 2003. Although there are gaps in the data coverage, the data remain of equivalent quality to prefailure data. Data products have been developed to fill these gaps using other ETM+ scenes. Index Terms—Calibration, characterization, history, Landsat, Thematic Mapper (TM). I. HISTORY T HE Thematic Mapper (TM) was designed in the late 1970s as the successor to the Multispectral Scanner (MSS) systems on the earlier Landsats. Its design offered spa- tial, radiometric, and geometric improvements over the MSS systems (Table I). The design of the TM was also significantly more complicated than the MSS, and there were uncertainties as to how well it would perform. This led to an intensive Na- tional Aeronautics and Space Administration (NASA)-funded Landsat Image Data Quality Analysis (LIDQA) study in the period following the launch of Landsat-4 in July 1982. This study characterized the radiometric, spatial, and geometric properties of the TM data. These studies [1], [2] generally validated the design and early mission performance of the Landsat-4 TM. Landsat-5, launched in 1984, was nearly a carbon copy of Landsat-4, with a few problems corrected that had been observed on Landsat-4. With the end of LIDQA in 1985, the majority of the published Landsat characterization work ceased. Some internal NASA Landsat science office efforts and NASA field campaigns, e.g., the First ISLSCP Field Experiment (FIFE), kept a limited amount of published Manuscript received October 7, 2004; revised November 5, 2004. B. L. Markham, D. L. Williams, and J. R. Irons are with the Land Cover Satellite Project Science Office, NASA Goddard Space Flight Center, Greenbelt, MD 20771 USA (e-mail: Brian.L.Markham@nasa.gov). J. C. Storey is with Science Applications International Corporation, U.S. Ge- ological Survey, Earth Resources Observation System (EROS) Data Center, Sioux Falls, SD 57198 USA (e-mail: James.C.Storey.1@gsfc.nasa.gov). Digital Object Identifier 10.1109/TGRS.2004.840720 characterization and calibration efforts for the TM instruments alive through about 1988 [3]–[5]. The Landsat-4 and -5 satel- lites and sensors were transferred from NASA to the National Oceanic and Atmospheric Administration (NOAA) (in 1983 for Landsat-4) to EOSAT (now Space Imaging, Inc., in 1985). Beginning in about 1990, the United States Geological Survey (USGS), which had assumed responsibility for gener- ating products from archived TM data for U.S. Government users, funded some studies aimed at improving the radiometric quality of TM data. Studies were initiated to understand some of the banding and striping artifacts in TM data [6]. In 1992, Public Law 102-555 required that the U.S. Government de- velop the Landsat-7 system to follow the Landsat-6 system under development by EOSAT at that time. (Landsat-6 was subsequently lost in a launch failure in 1993.) This return of Landsat to the government realm renewed interest in TM data. Landsat-5 TM data were used in the development of the Landsat-7 ground system by NASA and the USGS. The Landsat science team used Landsat-5 TM data in preparation for the launch of Landsat-7. Several papers on calibration and characterization of Landsat-5 TM data appeared in the mid-1990s as a result of these efforts [7], [8]. From the initiation of the Landsat-7 system, greater attention was paid toward the long-term characterization and calibration of the data than for earlier Landsats. In particular, an Image Assessment System (IAS) was incorporated into the ground processing system [9]. The IAS’s role was to characterize and calibrate the instrument and data over the life of the mission. Also, for the first two years of Landsat-7 operations, there was an active Landsat science team, a portion of which was concerned with calibration. This portion of the science team proposed and was funded to continue to provide vicarious calibration of Landsat-5 and -7 instruments and to reconstruct, as well as possible, the radiometric calibration record for the Landsat-4 and -5 TM instruments. Early results of the cali- bration and other Landsat-7 science team efforts appeared in the Special Landsat-7 Issue of Remote Sensing of Environment [10]. Key papers by Thome [11], Vogelmann et al. [12], and Schott et al. [13] address the absolute calibration of the En- hanced Thematic Mapper Plus (ETM+) instrument, and a paper by Teillet et al. [14] discusses the cross calibration between Landsat-5 TM and Landsat-7 ETM+. In 2001, control of the Landsat-4 and Landsat-5 instruments, spacecraft and archived data, reverted back to the U.S. Government, specifically the USGS. With this event, the USGS assumed responsibility for the calibration of the Landsat-4 and Landsat-5 TM data archives and newly acquired Landsat-5 TM data. Work was begun to extend the Image Assessment System to include Landsat-4 and-5 TM data in addition to Landsat-7 ETM+ data. 0196-2892/04$20.00 © 2004 IEEE