Sujit Roy of the Harmonized Landsat Sentinel-2 (HLS) team within the Interagency Implementation and Advanced Science Team (IMPACT) submitted an abstract to the IEEE International Geoscience and Remote Sensing Symposium (IGARSS) 2023 conference in Pasadena, CA. The abstract details a novel machine-learning based approach to adjust Sentinel-2 surface reflectance to match Landsat surface reflectance values, accounting for the differences in each sensor’s bandpass. With the expansion of the HLS data products to global coverage, the legacy bandpass adjustment algorithm (optimized over the United States) provides an opportunity for modernization. The machine-learning based approach under development by the HLS team in IMPACT accounts for more diverse land cover classes, improving the capability of the two HLS products to be used interchangeably. An additional benefit of the modernized algorithm is the ability for the bandpass adjustment to be applied in either direction (i.e., mapping Sentinel-2 reflectances to Landsat or mapping Landsat reflectances to Sentinel-2). While in a preliminary stage, if the new approach proves to be successful it provides a great opportunity for harmonization of Landsat and Sentinel-2 products in the future.
