With the launch of Landsat 9 in September 2021, an optimal opportunity for in-flight cross-calibration occurred when Landsat 9 flew underneath Landsat 8 while being moved into its final orbit. Since the two instruments host nearly identical imaging systems, the underfly event offered ideal cross-calibration conditions. The purpose of this work was to use the underfly imagery collected by the instruments to estimate cross-calibration parameters for Landsat 9 for a calibration update scheduled at the end of the on-orbit initial verification (OIV) period. Three types of uncertainty were considered: geometric, spectral, and angular (bidirectional reflectance distribution function—BRDF). Differences caused by geometric uncertainty were found to be negligible for this application. Spectral uncertainty was found to be minimal except for the green band when viewing vegetative targets. BRDF models derived from the MODIS BRDF product indicated substantial error could occur and required development of a mitigating methodology. With these three contributions of uncertainty properly addressed, it was estimated that the total cross-calibration uncertainty for underfly data could be kept under 1%. The data collected during the underfly were filtered to remove outliers based on uncertainty analysis. These data were used to calculate the TOA reflectance and radiance cross-calibration values for each spectral band by taking the ratio of Landsat 8 average pixel values to Landsat 9. Initial results of this approach indicated the cross-calibration may be as accurate as 0.5% in reflectance space and 1.0% in radiance space. The initial results developed in this study were used to refine the cross-calibration of Landsat 9 to Landsat 8 at the end of the OIV period.
The Landsat 8 and 9 Underfly Event occurred in November 2021, during which Landsat 9 flew beneath Landsat 8 in the final stages before settling in its final orbiting path. An analysis was performed on the images taken during this event, which resulted in a cross-calibration with uncertainties estimated to be less than 0.5%. This level of precision was due, in part, to the near-identical sensors aboard each instrument, as well as the underfly event itself, which allowed the sensors to take nearly the exact same image at nearly the exact same time. This initial calibration was applied before the end of the on-orbit initial verification (OIV) period; this meant the analysis was performed in less than a month. While it was an effective and efficient first look at the data, a longer-term analysis was deemed prudent to obtain the most accurate cross-calibration with the smallest uncertainties. The three forms of uncertainty established in the initial analysis, dubbed “Phase 1”, were geometric, spectral, and angular. This paper covers Phase 2 of the underfly analysis; several modifications were made to the Phase 1 process to improve the cross-calibration results, including a spectral correction in the form of a spectral band adjustment factor (SBAF) and a more robust filtering system that used the statistics of the reflectance data to better include important data compared to the more aggressive filters used in Phase 1. A proper uncertainty analysis was performed to more accurately quantify the uncertainty associated with the underfly cross-calibration. The results of Phase 2 showed that the Phase 1 analysis was within its 0.5% uncertainty estimation, and the cross-calibration gain values in this paper were used by USGS EROS to update the Landsat 9 calibration at the end of 2022.
Three advanced methodologies were performed during Landsat-9 on orbit and initialization and verification (OIV): Extended Pseudo Invariant Calibration Sites Absolute Calibration Model Double Ratio (ExPAC Double Ratio) and Extended Pseudo Invariant Calibration Sites (EPICS)-based cross-calibration utilizing stable regions in Northern African desert sites (EPICS-NA) and a global scale (EPICS-Global). The development of these three techniques was described using uncertainties analysis. The ExPAC Double Ratio was derived based on the ratio between ExPAC model prediction and satellite measurements for Landsat-8 and Landsat-9. The ExPAC Double Ratio can be performed to determine differences between sensors ranging from visible, red edge, near-infrared, to short-wave infrared wavelengths. The ExPAC Double Ratio and EPICS-based inter-comparison ratio uncertainties were determined using the Monte Carlo Simulation. It was found that the uncertainty levels of 1–2% can be achieved. The EPICS-based cross-calibration results were derived using two targets: EPICS-NA and EPICS-Global, with uncertainties of 1–2.2% for all spectral bands. The inter-comparison results between Landsat-9 and Landsat-8 during the OIV period using the three advanced methods were well within 0.5% for all spectral bands except for the green band, which showed sub 1% agreement.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.