Landsat Geometry

Landsat Levels of Processing

All Landsat Level-1 data products are created using the best available processing level for each particular scene. The processing level used is determined by the existence of ground control points, elevation data provided by various digital elevation models, and/or payload correction data collected by the spacecraft and sensor. 

Return to Landsat Geometry Overview

 

Landsat satellite data are processed into Level-1 scenes using the Landsat Product Generation System (LPGS). Each Landsat scene is processing to one of the three following processing levels:

  • Precision and Terrain Correction (L1TP)
    All Landsat scenes are attempted to process to L1TP, using Ground Control Points (GCPs) and a digital elevation model (DEM).  In some cases (and more likely in older Landsat data), scene and/or sensor issues, or insufficient reference data can cause L1TP processing to fail. Scene issues include snow, ice, and clouds, which prevent an accurate registration of GCPs within a scene. Sensor issues include measurement/outliers in the spacecraft or instrument telemetry of an interval which also affect the usability of the GCPs. At these times, a L1GT or a L1GS product will be created instead.  Click here to see more details about L1TP.
     
  • Systematic Terrain Correction (L1GT)
    L1GT products are created when the systematic product has consistent and sufficient locational accuracy to permit the application of a terrain model.
    Click here to see more details about L1GT.
     
  • Systematic Correction (L1GS)
    L1GS products are created when the locational accuracy is not sufficient to apply terrain correction, such as: Insufficient number of ground control points, such as small islands or Antarctic; opaque clouds that obscure the ground; or locational errors greater than the search distance for ground control.
    Click here to see more details about L1GS. 

Landsat data are also characterized by Collection Tiers, which make up the inventory structure for Landsat Collections Level-1 data. Each scene is placed into the appropriate tier, based on data quality and the level of processing. Visit the Collection 2 Tiers section on the Landsat Collection 2 page to learn more about Landsat Collection tiers.

 

How Geometric Errors affect Landsat Data Quality

When scenes fail to process to L1TP, the loss of the ability to use GCPs affect data from each sensor differently. Details for each sensor are described below. The USGS Cal/Val team continually works to improve the quality of Landsat data. Data improvements will be implemented into future Landsat Collections. Users are encouraged to contact the USGS with scenes that are found to be problematic, as these will be used as testing data to prove data improvements. Every best effort is always made to improve data quality.

Landsat 8

If a Landsat 8 scene fails to generate a L1TP product, the scene is processed to L1GT, without the use of GCPs. Since the pointing accuracy of Landsat 8 spacecraft is excellent (about 17-m, circular error (CE)-90, terrain correction helps to account for the higher order/relief induced distortions even when GCPs are not used. So, while Landsat 8 L1GT products may not be as accurate as L1TP products, they are typically better than 30-meters (m) (one pixel), even without the use of GCPs.

Landsat 7

While the Landsat 7 spacecraft pointing-accuracy requirement is 250-m, in most cases, it is 10- m. Although Landsat 7 data are not as accurate as Landsat 8, the advantage of correcting the higher-order terrain-induced effects outweighs the inaccuracy introduced due to the small mis-registration error between the DEM and the scene due to its pointing errors. Therefore, Landsat 7 scenes that fail to generate L1TP products are corrected for terrain without the use of GCPs (L1GT).

It should be understood that In some rare cases, due to errors / inconsistencies in the spacecraft and instrument telemetry, scenes can exhibit errors that may introduce a simple shift (on the order of a few hundred meters) to higher order distortions that can affect the entire scene producing inconsistent internal geometry. In this case, the L1GT products could be geometrically inaccurate by several hundred meters. The EROS Cal/Val team is continuing to work in improving the geometric accuracies of these scenes.

Landsat 4-5 Thematic Mapper (TM)

There are no specific pointing accuracy requirements for Landsat 5 (or for Landsat 4). Based on EROS Cal/Val analysis, the pointing accuracy of a Landsat 5 scene can vary anywhere from 0 m to 10 kilometers (km), depending on the time-period within the life of the mission, telemetry-related issues, and the type of telemetry available to estimate the spacecraft position (no payload correction data (PCD) or use of two-line elements (TLE) than spacecraft position information. Most Landsat 5 TM scenes in the archive are expected to be within 1 km, but scenes with offsets between 1 and 3 km are not uncommon. LPGS can estimate the scene’s geometric offsets when their offsets are within 4 km, but scenes with offsets greater than 4 km cannot be registered with the GCPs. These scenes will be processed as L1GS without the use of GCPs or DEM.

Unlike Landsat 8 and Landsat 7, TM scenes are likely to have been acquired with large pointing error (> 500 m). Therefore, applying terrain correction without the use of GCPs will introduce large incorrect relief-adjustments to the scene. Therefore, all TM scenes are processed as L1GS products if GCPs are not used in data correction.

Landsat 1-5 Multispectral Scanner (MSS)

MSS geodetic accuracy is known to have a much greater offset than TM data.  The EROS Cal/Val Team is looking to address MSS geometric/geodetic data accuracy in the future.

 

Additional Information on Processing Corrections Accuracy

Users need to be aware that since the accuracy of L1TP, L1GT, and L1GS products vary, stacking scenes with different corrections will introduce misalignment between the images. The magnitude of misalignment between the products is dependent on the sensor, product types, and can vary anywhere from less than one pixel to hundreds of pixels. All L1TP Tier-1 products are expected to be consistent to within 12-m. Although Tier-1 and Tier-2 products are expected to be aligned to within 1 pixel in general, it is not uncommon to observe misalignments more than 1 pixel between Tier-1 and Tier-2 products (for any sensor combinations) or between any Tier-2 products.

Unfortunately, since Landsat data are processed using automated approaches when attempting to produce a L1TP product, it is not possible to provide estimated offsets for L1GT and L1GS products. A designation of L1GT or L1GS for a given scene is due to the fact that an accurate geometric offset could not be estimated using the GCPs through these automated approaches. These product designations also indicate no information is available on the magnitude of the offsets between the imagery and ground control. This can also be an indication that any cloud-free (snow/ice free) scene that failed to generate a L1TP product is likely to have poor telemetry data. This is especially true for the data from earlier Landsat missions. Scenes with poor telemetry data are highly susceptible to higher-order distortions and cannot be corrected to match the requirements of an orthorectified product even when approximate offsets are known.

 

Graphic displaying pixel levels of geometric distortion in a sample Landsat scene

In Pixel units (geometric error in the scene with respect to an orthorectified image)
0 <= Green < 0.5
0.5 <= Cyan < 1
1 <= Green < 2
2 <= Green < 3
Red >= 3

How Users can Best Utilize Scenes of Different Corrections

Geometric registration of the data with further processing is dependent on the whether the scene can be corrected using a simple set of biases as offsets. In some cases, as mentioned earlier, incorrect measurements / outliers in the instrument or spacecraft telemetry can potentially introduce higher-order distortions in the L1GS products that cannot be corrected by offsets or using a linear model. Secondly, terrain-induced effects cannot be corrected without the use of DEM.

The image to the right is an example of the higher-order distortion that can be observed in a L1TP product that was originally processed as a L1GS product. This is an example of a scene that has bad telemetry data, resulting in not only large geodetic offset, but also an inconsistency in the internal geometry of the scene that cannot be corrected using a simple polynomial or affine transformation.The L1GS product’s geometric offset was measured by comparing against a L1TP product. The measured offsets (>10 km), determined manually, were then used within the Image Assessment System (IAS) to produce a L1TP product. Several software packages and most GIS systems will allow a user to pick control manually and apply a given spatial transformation to an image file.

The color-coded graphics observed in the figure shows the higher order distortions present in the image even when it is corrected for the large geodetic offsets within the scene.

 

 

Precision and Terrain Correction (Level-1TP, L1TP)

Precision and Terrain Correction provides radiometric and geodetic accuracy by incorporating ground control points while employing a Digital Elevation Model (DEM) for topographic displacement. Geodetic accuracy of the product depends on the image quality and the accuracy, number, and distribution of the ground control points (GCP):

  • Precision fit and verification RMSE estimates are only available for L1TP products. The precision fit estimate (RMSE_Model) quantifies how well the control points used in the precision registration match the reference GCP database. The verification estimate (RMSE_Verify) of MSS and TM data quantifies how well the image matches an independent set of GCPs in the reference GCP database.
  • The specification for L1TP product acceptance varies by sensor. The specification is rigid for Landsat 8, Landsat 7, and Landsat 4-5 Thematic Mapper (TM) data, many of which have excellent internal geometry. Given the poor internal geometry of the Multispectral Sensor (MSS) aboard Landsat 1-5, the use of ground control even for data with large RMSE was considered preferable, to creating a large proportion of data as L1GS with the internal geometry uncorrected. 
  • The information provided in the metadata file can be used to evaluate the geodetic accuracy of the L1TP data product.

Systematic Terrain Correction (Level-1GT, L1GT)

Systematic Terrain Correction provides systematic, radiometric, and geometric accuracy, while employing a Digital Elevation Model (DEM) to correct for relief displacement:

  • Landsat 7 scenes without sufficient control to produce L1TP images are processed to an L1GT.
  • Landsat 8 scenes without sufficient ground point control to produce L1TP products are processed as L1GT. The accuracy of the L1GT systematic product approaches that of an L1TP product. Registration to the shared Ground Control Point reference data set improves the co-registration to the other Landsat sensors. For scenes where the reference database error exceeds 30 m the L1GT images will have better absolute accuracy than Landsat 8 L1TP data, but may not be co-registered to within 30 m.
  • Landsat 7 and Landsat 8 data over Antarctica are processed to an L1GT since it has not been possible to generate ground control in Antarctica suitable for the generation of an L1TP product.

Systematic Correction provides systematic radiometric and geometric corrections, which are derived from data collected by the sensor and spacecraft.

  • Landsat scenes processed as L1GS do not have sufficient geodetic accuracy to include in image-to-image analysis without further image-specific evaluation and registration.
  • Landsat ETM+ geometric accuracy of the systematically corrected product should be within 125 m, 90 percent of the time for low-relief areas at sea level based on pre-fit estimates.  Error increases as distance and elevation increase from low relief areas.
  • Landsat TM geometric accuracy for L1GS products should be within 700 m, 90 percent of the time for low-relief areas at sea level based on pre-fit estimates.
  • Landsat MSS geometric accuracy for L1GS products is substantially worse than later sensors. Both the internal geometry and locational accuracy will require manual registration of the images
  • Landsat TM and MSS images may be offset from its correct spatial location by thousands of meters, preventing the use of terrain correction for systematic products.

The success rate for creating L1TP products varies by sensor, but also by cloud cover. The table below displays, for Landsat 8, each cloud-cover class and lists the proportion of images that process to an L1TP, those that fallback to an L1GT or L1GS after a failed attempt to produce an L1TP (clouds or poor ephemeris data), and the proportion that are planned to produce an L1GT (night, Antarctica or insufficient land features). These values were generated from the most recent Landsat Product Generation System (LPGS) software version.

Note: The information shown below are based on Landsat Collection 1; updated graphics are being created for Collection 2 and will be added to this page when they are complete.

Landsat Level-1 Products Registration Success by Cloud Cover

Landsat Level-1 Products Registration Success by Cloud Cover, based on values generated February 28, 2019. 

Table Explanation
*Landsat 1-7: All path/row combinations are separated internally into two groups; path/rows that will undergo precision and terrain correction, and path/rows that will not be precision and terrain corrected. If precision and terrain correction is attempted and successful, the scene becomes an L1TP. If precision correction is unsuccessful, the scene becomes an L1GT FB (fallback). If precision and terrain correction is not applied, the scene will become an L1GS (or L1GT for Landsat 7).
*Landsat 8: All path/rows are separated into two groups; path/rows that have produced at least one L1TP scene, and path/rows that have never produced an L1TP scene. Precision and terrain correction is attempted on all Landsat 8 scenes.  If precision correction fails from the path/row group that has produced at least one L1TP, the scene becomes an L1GT FB (fallback). If precision correction fails from a path/row that has never produced an L1TP (which is likely), then the scene will become an L1GT (no fallback).

RMSE Distribution Plots

The distribution of the RMSE for each sensor shows significant improvement as sensor and spacecraft technologies evolve, as of February 28, 2019.

Landsat 8 OLI Collection 1 L1TP RMSE

Landsat 8 OLI Collection 1 L1TP RMSE: Equal to or less than 12 meters in 92.2 percent of the data. 

Landsat 7 ETM+ Collection 1 L1TP RMSE

Landsat 7 ETM+ Collection 1 L1TP RMSE: Equal to or less than 12 meters in 96.3 percent of the data. 

Landsat 4-5 TM Collection 1 L1TP RMSE

Landsat 4-5 TM Collection 1 L1TP RMSE: Equal to or less than 12 meters in 95.4 percent of the data. 

Landsat 1-5 MSS Pre-Collection L1T RMSE

Landsat 1-5 MSS Collection 1 L1T RMSE: Less than 12 meters in 1.23 percent of the data.