A general spectral bandpass adjustment function (SBAF) for harmonizing landsat-sentinel over inland and coastal waters

Abstract

Landsat 8/9 Operational Land Imager (OLI) and Sentinel-2 Multispectral Imager (MSI) are the most relevant medium spatial resolution data sources for aquatic applications, and integrating these spectral images into a single product constellation offers significant potential for monitoring dynamic processes over coastal and inland waters. Due to water's inherently low reflectance values, small differences in the relative spectral responses (RSR) between the two sensors can result in significant discrepancies in water reflectance retrievals. To ensure compatibility and consistency in harmonized products for aquatic studies, spectral bandpass adjustment function (SBAF) for cross-calibration between OLI and MSI sensors must be carefully derived and applied. This study provides a global analysis of 4047 match-ups of atmospherically corrected Landsat-8/9 OLI and Sentinel-2 A/B MSI L1 products over inland and coastal waters to generate a new general SBAF for aquatic studies. Atmospheric correction was performed using the 6 S V radiative transfer model, a widely validated approach for aquatic remote sensing applications. The selected images were sensed ≤30 min apart on the same day, under <5 % cloud cover, ≤5° solar zenith difference across different atmospheric conditions and aquatic systems (928 coastal and inland water bodies) over the world. A robust quality-controlled protocol was developed to remove low-quality image pairs under sun/sky glint, ice/snow surface, and cirrus clouds. Following this procedure, 2,2 million quality filtered water reflectance pixels were extracted. The SBAF coefficients (i.e., slope and offset) were derived through statistical regression between Landsat-8/9 and Sentinel-2 reflectance values. Additionally, we simulated sensor band responses using a global in-situ hyperspectral water dataset and calculated the SBAF coefficients for comparison with the pixel-based results. The application of SBAF was demonstrated through comparative analyses of spectral reflectance from Landsat-8/9 and Sentinel-2 before and after the cross-calibration. Our findings underscore the effectiveness of these coefficients in reducing spectral discrepancies between Landsat-8/9 and Sentinel-2 water reflectance measurements.