The Landsat 9 satellite was launched on September 27, 2021 to continue systematic imaging of the Earth’s land surfaces. Together with Landsat 8, it provides coverage of the entire Earth every 8 days. Landsat 9 carries the Operational Land Imager (OLI), which is practically a copy of the Landsat 8 OLI, and the Thermal Infrared Sensor (TIRS). In this paper we demonstrate the excellent radiometric performance of Landsat 9 OLI over its first several months of operations on orbit. On-board calibrator data were used to assess the sensor’s radiometric performance characteristics. All spectral bands are radiometrically stable to within 0.1%. The signal-to-noise performance is stable and is 3 to 8% better than Landsat 8. The bias stability is better than 1 DN (Digital Number). The validation of the absolute calibration performed with surface measurements indicated the OLI was calibrated to within 5% in spectral radiance and 3% in reflectance. Still, a comparison between Landsat 9 and Landsat 8 OLI derived top-of-atmosphere reflectance indicated small disagreements between the instruments in all spectral bands, The absolute radiometric calibration of Landsat 9 OLI was adjusted to be in closer agreement with Landsat 8 OLI before products were released to the public.
KEYWORDS: Calibration, Landsat, Earth observing sensors, Sensors, Data acquisition, Signal to noise ratio, Data modeling, Lamps, Image processing, Data archive systems
The U.S. Geological Survey (USGS) archive of Earth images acquired by Landsat 1-8 sensors is organized in collections of consistently calibrated, geolocated, and processed data products. Such an organization ensures consistent quality of the archived data within a collection over time and across all instruments within the Landsat mission. In December 2020, the USGS completed reprocessing of the archived Landsat data and released a new collection, Collection 2, which introduced surface reflectance and surface temperature Level-2 products, implemented improved ground control and elevation datasets, and brought several geometric and radiometric calibration enhancements. Radiometric enhancements include absolute and relative gain updates for both imaging sensors, Operational Land Imager (OLI) and Thermal Infrared Sensor (TIRS), aboard Landsat 8 and a change in calculation of bias for OLI. In this paper we present the analysis of Collection 2 Landsat 8 products demonstrating stability over the mission lifetime, the improvement in OLI signal to noise ratio and along-track striping performance resulting from enhanced bias correction, as well as reduction of cross-track striping in TIRS data as a result of relative gain updates. In addition, we discuss effects of two safehold events that Landsat 8 experienced in November 2020 on the radiometric calibration of both sensors and product performance.
KEYWORDS: Earth observing sensors, Landsat, Calibration, Image processing, Data archive systems, Reflectivity, Infrared sensors, Data processing, Data acquisition, Time series analysis
Since the release of Collection 1 in 2016 Landsat imagery in the U.S. Geological Survey (USGS) archive have been organized and managed as a collection of consistently calibrated and processed Earth image data that have been acquired globally over the last nearly 50 years. In addition to Level-2 products and improved geometric accuracy, Collection 2 brings several radiometric updates. In this paper, we will address the Landsat 8 Operational Land Imager (OLI) and Thermal Infrared Sensor (TIRS) absolute and relative gain updates, a change in the OLI bias calculation method and the Landsat 5 Thematic Mapper (TM) thermal band calibration update.
Landsat 9 is planned for launch in December 2020 to continue the mission of observing changes on the Earth’s surface that began in 1972 with the launch of Landsat 1. Like Landsat 8, Landsat 9 will carry two imaging instruments: Operational Land Imager 2 (OLI-2), designed and built by Ball Aerospace**, and Thermal Infrared Sensor 2 (TIRS-2), manufactured by NASA Goddard Space Flight Center (GSFC). As of this writing, both sensors have completed the instrument-level ground testing and are ready for integration into the spacecraft. Data collected during the pre-launch performance testing are analyzed to assess the usability of responses of the video reference pixels (VRPs) located on the focal planes of OLI-2 and Landsat 8 OLI for more accurate detector bias estimates, develop a methodology to correct for nonlinearities in the OLI-2 response and compare it to the OLI correction approach, and determine the spatial performance of TIRS-2.
Landsat-8 has been operating on-orbit for 5+ years. Its two sensors, the Operational Land Imager (OLI) and Thermal Infrared Sensor (TIRS), are continuing to produce high quality data. The OLI has been radiometrically stable at the better than 0.3% level on a band average basis for all but the shortest wavelength (443 nm) band, which has degraded about 1.3% since launch. All on-board calibration devices continue to perform well and consistently. No gaps in across track coverage exist as 100% operability of the detectors is maintained. The variability over time of detector responsivity within a band relative to the average is better than 0.05% (1 sigma), though there are occasional detectors that jump up to 1.5% in response in the Short-Wave InfraRed (SWIR) bands. Signal-to-Noise performance continues at 2-3x better than requirements, with a small degradation in the 443 nm band commensurate with the loss in sensitivity. Pre-launch error analysis, combined with the stability of the OLI indicates that the absolute reflectance calibration uncertainty is better than 3%; comparisons to ground measurements and comparisons to other sensors are consistent with this. The Landsat-8 TIRS is similarly radiometrically stable, showing changes of at most 0.3% over the mission. The uncertainty in the absolute calibration as well as the detector to detector variability are largely driven by the stray light response of TIRS. The current processing corrects most of the stray light effects, resulting in absolute uncertainties of ~1% and reduced striping. Efforts continue to further reduce the striping. Noise equivalent delta temperature is about 50 mK at typical temperatures and 100% detector operability is maintained. Landsat-9 is currently under development with a launch no earlier than December 2020. The nearly identical OLI-2 and upgraded TIRS-2 sensors have completed integration and are in the process of instrument level performance characterization including spectral, spatial, radiometric and geometric testing. Component and assembly level measurements of the OLI-2, which include spectral response, radiometric response and stray light indicate comparable performance to OLI. The first functional tests occurred in July 2018 and spatial performance testing in vacuum is scheduled for August 2018. Similarly, for TIRS-2, partially integrated instrument level testing indicated spectral and spatial responses comparable to TIRS, with stray light reduced by approximately an order of magnitude from TIRS.
Landsat data in the U.S. Geological Survey (USGS) archive are being reprocessed to generate a tiered collection of consistently geolocated and radiometrically calibrated products that are suitable for time series analyses. With the implementation of the collection management, no major updates will be made to calibration of the Landsat sensors within a collection. Only calibration parameters needed to maintain the established calibration trends without an effect on derived environmental records will be regularly updated, while all other changes will be deferred to a new collection. This first collection, Collection 1, incorporates various radiometric calibration updates to all Landsat sensors including absolute and relative gains for Landsat 8 Operational Land Imager (OLI), stray light correction for Landsat 8 Thermal Infrared Sensor (TIRS), absolute gains for Landsat 4 and 5 Thematic Mappers (TM), recalibration of Landsat 1-5 Multispectral Scanners (MSS) to ensure radiometric consistency among different formats of archived MSS data, and a transfer of Landsat 8 OLI reflectance based calibration to all previous Landsat sensors. While all OLI/TIRS, ETM+ and majority of TM data have already been reprocessed to Collection 1, a completion of MSS and remaining TM data reprocessing is expected by the end of this year. It is important to note that, although still available for download from the USGS web pages, the products generated using the Pre-Collection processing do not benefit from the latest radiometric calibration updates. In this paper, we are assessing radiometry of solar reflective bands in Landsat Collection 1 products through analysis of trends in on-board calibrator and pseudo invariant site (PICS) responses.
The Landsat Project is planning to implement a new collection management strategy for Landsat products generated at the U.S. Geological Survey (USGS) Earth Resources Observation and Science (EROS) Center. The goal of the initiative is to identify a collection of consistently geolocated and radiometrically calibrated images across the entire Landsat archive that is readily suitable for time-series analyses. In order to perform an accurate land change analysis, the data from all Landsat sensors must be on the same radiometric scale. Landsat 7 Enhanced Thematic Mapper Plus (ETM+) is calibrated to a radiance standard and all previous sensors are cross-calibrated to its radiometric scale. Landsat 8 Operational Land Imager (OLI) is calibrated to both radiance and reflectance standards independently. The Landsat 8 OLI reflectance calibration is considered to be most accurate. To improve radiometric calibration accuracy of historical data, Landsat 1-7 sensors also need to be cross-calibrated to the OLI reflectance scale. Results of that effort, as well as other calibration updates including the absolute and relative radiometric calibration and saturated pixel replacement for Landsat 8 OLI and absolute calibration for Landsat 4 and 5 Thematic Mappers (TM), will be implemented into Landsat products during the archive reprocessing campaign planned within the new collection management strategy. This paper reports on the planned radiometric calibration updates to the solar reflective bands of the new Landsat collection.
Now in its 17th year of operation, the Enhanced Thematic Mapper + (ETM+), on board the Landsat-7 satellite, continues to systematically acquire imagery of the Earth to add to the 40+ year archive of Landsat data. Characterization of the ETM+ on-orbit radiometric performance has been on-going since its launch in 1999. The radiometric calibration of the reflective bands is still monitored using on-board calibration devices, though the Pseudo-Invariant Calibration Sites (PICS) method has proven to be an effective tool as well. The calibration gains were updated in April 2013 based primarily on PICS results, which corrected for a change of as much as -0.2%/year degradation in the worst case bands. A new comparison with the SADE database of PICS results indicates no additional degradation in the updated calibration. PICS data are still being tracked though the recent trends are not well understood. The thermal band calibration was updated last in October 2013 based on a continued calibration effort by NASA/Jet Propulsion Lab and Rochester Institute of Technology. The update accounted for a 0.036 W/m2 sr μm or 0.26K at 300K bias error. The updated lifetime trend is now stable to within +/- 0.4K.
Landsat-8 and its two Earth imaging sensors, the Operational Land Imager (OLI) and Thermal Infrared Sensor (TIRS) have been operating on-orbit for 2 1/2 years. The OLI radiometric calibration, which is monitored using on-board lamps, on-board solar diffusers, the moon and vicarious calibration techniques has been stable to within 1% over this period of time. The Coastal Aerosol band, band 1, shows the largest change at about 1% over the period; all other bands have shown no significant trend. OLI bands 1- 4 show small discontinuities in response (+0.1% to 0.2%) beginning about 7 months after launch and continuing for about 1 month associated with a power cycling of the instrument, though the origin of the recovery is unclear. To date these small changes have not been compensated for, but this will change with a reprocessing campaign that is currently scheduled for Fall 2015. The calibration parameter files (each typically covering a 3 month period) will be updated for these observed gain changes. A fitted response to an adjusted average of the lamps, solar and lunar results will represent the trend, sampled at the rate of one value per CPF.
The Operational Land Imager (OLI) on Landsat-8 has been collecting imagery on orbit for 17 months. The radiometric performance of the OLI is monitored using on-board systems (lamps and solar diffusers) as well as by reference to lunar and ground measurements and other satellite systems. Over this nearly 1½ years of operation the OLI has been extremely radiometrically stable in all of its 9 spectral bands. Only the shortest wavelength band, centered at 443 nm, which has degraded about 0.8%, has changed by more than the variability among the measurements (~0.2%). This consistency between the lamps, diffusers, moon, and ground measurements lends high confidence to these statements, which is unusual for a system so early in its lifetime. Comparisons to other satellite systems and ground measurements show that the OLI is calibrated to within requirements and generally better than 3% in both radiance and reflectance.
From the Landsat program's inception in 1972 to the present, the earth science user community has benefited from a
historical record of remotely sensed data. The multispectral data from the Landsat 5 (L5) Thematic Mapper (TM) sensor
provide the backbone for this extensive archive. Historically, the radiometric calibration procedure for this imagery used
the instrument's response to the Internal Calibrator (IC) on a scene-by-scene basis to determine the gain and offset for
each detector. The IC system degraded with time causing radiometric calibration errors up to 20 percent. In May 2003
the National Landsat Archive Production System (NLAPS) was updated to use a gain model rather than the scene
acquisition specific IC gains to calibrate TM data processed in the United States. Further modification of the gain model
was performed in 2007. L5 TM data that were processed using IC prior to the calibration update do not benefit from the
recent calibration revisions. A procedure has been developed to give users the ability to recalibrate their existing Level-1
products. The best recalibration results are obtained if the work order report that was originally included in the standard
data product delivery is available. However, many users may not have the original work order report. In such cases, the
IC gain look-up table that was generated using the radiometric gain trends recorded in the NLAPS database can be used
for recalibration. This paper discusses the procedure to recalibrate L5 TM data when the work order report originally
used in processing is not available. A companion paper discusses the generation of the NLAPS IC gain and bias look-up
tables required to perform the recalibration.
The National Landsat Archive Production System (NLAPS) has been the primary processing system for Landsat data
since U.S. Geological Survey (USGS) Earth Resources Observation and Science Center (EROS) started archiving
Landsat data. NLAPS converts raw satellite data into radiometrically and geometrically calibrated products. NLAPS
has historically used the Internal Calibrator (IC) to calibrate the reflective bands of the Landsat-5 Thematic Mapper
(TM), even though the lamps in the IC were less stable than the TM detectors, as evidenced by vicarious calibration
results. In 2003, a major effort was made to model the actual TM gain change and to update NLAPS to use this model
rather than the unstable IC data for radiometric calibration. The model coefficients were revised in 2007 to reflect
greater understanding of the changes in the TM responsivity.
While the calibration updates are important to users with recently processed data, the processing system no longer
calculates the original IC gain or offset. For specific applications, it is useful to have a record of the gain and offset
actually applied to the older data. Thus, the NLAPS calibration database was used to generate estimated daily values for
the radiometric gain and offset that might have been applied to TM data.
This paper discusses the need for and generation of the NLAPS IC gain and offset tables. A companion paper covers the
application of and errors associated with using these tables.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.