The USGS Earth Resources Observation and Science (EROS) Cal/Val Center of Excellence (ECCOE) aims to help the world produce and access the best quality Earth observation remote sensing data possible.
ECCOE does this in many ways, from calibrating Landsat to be the gold standard for other satellites to helping compile a database of details about satellites past, present and future. Now, ECCOE has developed a new tool that helps users visually compare data from satellite sensors and analyze their strengths and weaknesses over different terrain.
In its initial release, the Land Product Characterization System (LPCS) enables the comparison of data from the Landsat 9 and 8 OLI, Landsat 7 ETM+, Sentinel-2A and Sentinel-2B MSI, MODIS and VIIRS sensors.
Serving as a backdrop for the tool are six locations in Algeria, Libya and Mauritania endorsed by the Committee on Earth Observation Satellites (CEOS) as reference sites for the calibration of orbiting space-based optical imaging sensors, chosen for their desert environments that are uniform, stable and unchanging. Calibration involves continually comparing measurements from the sensors to the ground references, called Pseudo-Invariant Calibration Sites (PICS), and adjusting to ensure data quality.
These PICS allow for trending, which USGS EROS contractor and LPCS Systems Engineer Lead LaDonn Powell described as graphing out the reference site data over a period of time to see how radiometric performances compare across different sensors, or whether a trend of changes can be identified for a particular sensor. Powell gave some examples of what scientists might be looking for: “If you wanted to see degradation in sensor performance or understand seasonality in sensor trends, maybe some physical on-orbit changes in sensors, even month-to-month or year-to-year.”
LPCS can be filtered for satellites, data product levels, attributes of radiance or reflectance, spectral bands, and calibration sites. This allows, for example, for comparisons of bands within one sensor or a single band among multiple sensors, Powell said. Tips for using LPCS can be found here.
“We’re able to plot things out in a chart and visually see where there are differences—how one sensor compares to other sensors, if there are outliers, or if there’s a trend going in a different direction than expected, that type of thing,” Powell said.
With Analysis Ready Data (ARD) and interoperability, there is a lot of focus on free and open data access and using that data for scientific analysis. LPCS fits well in the realm of those large global data sets to inform users of the quality of the data and the differences between the sensors.
Future planned LPCS releases will add more characterization methods beyond PICS, including band-to-band (B2B), image-to-image (I2I), simultaneous nadir overpass (SNO) and signal-to-noise ratio (SNR). Additional sensors will include GOES-16 ABI and Landsat 4-5 TM. Additional locations will include the Mojave Desert and about a dozen of each of USGS geometric supersites and in-situ sites.
Powell is pleased that this initial LPCS release could happen just in time for the Joint Agency Commercial Imagery Evaluation (JACIE) 2023 Workshop, which ECCOE supports, on March 27-30, 2023, at the USGS Headquarters in Reston, Virginia. A JACIE poster will spotlight tools like the LPCS; the Land Remote Sensing Satellites Online Compendium, a database of satellites and sensors; and the Spectral Characteristics Viewer, which helps users choose the best satellite bands for their particular research.