The Multispectral Scanner (MSS) sensor acquired imagery of the earth from July 1972 to January 1999 on board Landsats 1 through 5. Imagery collected during the first decade of these missions predates most other operational imaging efforts, providing a unique view of the earth not available elsewhere. These early missions also offered the first digital remote sensing products to the earth science community.
After the Thematic Mapper (TM) sensor on Landsat 5 failed in November 2011, the MultiSpectral Scanner (MSS) instrument was brought back online. From June 2012 until January 2013, over 15,000 MSS scenes were collected. Due to recent modifications to the Level-1 Product Generation System (LPGS), these data are now available to all users.*
*While bands 1, 2 and 3 contain usable data, band 4 has not produced any data since 1995. Also, a strip of 8 pixels are missing in the center of the image, equating to approximately 18,000 pixels. This affects all bands.
Over the years, technological advances in satellite and sensor designs and enhanced ground processing techniques have allowed for greatly improved geometrically and radiometrically processed data products. Precise location of the early MSS Landsat satellites was not well known, as is now the standard on modern systems. Therefore, the image processing from these earlier missions often required manual intervention to create products that were comparable to modern approaches.
With the advent of Web-enabled Landsat imagery, the demand for MSS products significantly outpaced the product generation system in place. The manual intervention required to process images to geometrically accurate, terrain-corrected products could not keep up with demand. Using the automated systems, only a small number of MSS scenes could be processed to a terrain-corrected product. The majority could only be processed with systematic geometry, resulting in a less accurate data product. It was also necessary to produce MSS products that are compatible in format and metadata with the Thematic Mapper (TM) and Enhanced Thematic Mapper Plus (ETM+) products. These data types are produced with the Level-1 Product Generation System (LPGS).
To resolve these issues, MSS data products with improved geometric correction were released in fall 2010. The products are processed on LPGS, similar to all other Landsat data types. Also, MSS imagery has been cross-correlated to radiance using the Normalized Radiance method developed at South Dakota State University (SDSU) to aid in the analysis of time-based studies involving multiple date scenes of MSS imagery.
All MSS products are processed to generate a geolocated, terrain-corrected product (L1T) in the Universal Transverse Mercator (UTM) projection cast on the WGS-84 datum at a 60-meter pixel size. Ground control is obtained in an automated fashion from the Global Land Survey (GLS1975 - Landsat 1, Landsat 2, and Landsat 3) using a hierarchical image-matching technique for initial fits, and then with GLS2000-derived control for final refinements.
If adequate control is not available due to cloud cover or significant ground changes, a systematically corrected (L1G) image is produced instead. Landsat MSS L1G products can vary significantly in geometric quality and are generally not useful for time-based studies without manual processing by the end user.
There are several known data anomalies associated with MSS data, described on the Landsat Known Issues web page.
A final set (grid) of verification points is processed using cross-correlation techniques based on the GLS 2000 dataset as a geometric reference. An accuracy report and a graphic overview of product accuracy are provided with each L1T MSS image produced by the LPGS system.
The product metadata also includes an RMSE for the overall scene and for each quadrant of each scene. These visual products provide a quick-look of product accuracy but may provide a pessimistic view of overall accuracy due to blunders in the correlation surface caused by clouds, cloud shadow, or change on the ground.
The Image Verify product is experimental; we encourage your input on how we might improve it.
|Path/Row : 36/37 (WRS-1)
RMSE by quadrant and scene
Verification report contains
LPGS processing will use cross-calibration gains, biases, and time-dependent variables to map the MSS archive data to radiance using the Normalized Radiance method developed at SDSU. Radiance values are scaled to a 0-255 range with 0 reserved for fill, 1 (QCalMin) for low saturation value (LMin), and 255 (QCalMax) for high saturation value (LMax). LMin, Lmax, QCalMin, and QCalMax are reported in the image metadata for use in conversion to top-of-atmosphere (TOA) radiance. All MSS sensors are cross-calibrated to Landsat 5 MSS, so reflectance calculations need to use Landsat 5 MSS parameters for ESUN.
Eventually, a consistent, cross-calibrated archive of Landsat data will be available, as increasing knowledge of this oldest dataset expands and trending continues. This large-scale effort may require regular reprocessing of MSS data. Users are encouraged to report any problems so that new solutions for future processing enhancements can be investigated.