ACIX-III AQUA: Evaluation of atmospheric correction processors for hyperspectral satellite over inland and coastal waters - A comparative analysis of atmospheric correction methods applied to PRISMA and EnMAP using in situ reflectance data

Résumé fourni par la source

This report shows the main outcomes of the Atmospheric Correction Intercomparison Exercise (ACIX-III Aqua), aiming to test atmospheric correction (AC) processors on spaceborne imaging spectroscopy. Developed under the same concepts of ACIX-Aqua that enabled an evaluation of eight state-of-the-art AC processors available for multi-spectral satellite data, ACIX-III Aqua focuses on hyperspectral data, particularly as those acquired by the Italian Space Agency (ASI) satellite hyperspectral mission PRecursore IperSpettrale della Missione Operativa (PRISMA). Launched in March 2019, the PRISMA mission has acquired thousands of images over various marine and freshwaters, where in situ measurements were collected to enable match-up generation with satellite products. The in situ dataset originates from two sources: hyperspectral data contributed by the international community (called Community Validation Database, CVD) and multispectral measurements provided by the AERONET-OC network. This extensive data set allows the performance of each AC processor to be evaluated in different spectral regions and on eight different Optical Water Types (OWTs). Five AC models are participating in ACIX-III Aqua: ACOLITE, hGRS, POLYMER, iCOR, and MIP. Additionally, the standard AC product of PRISMA L2C was included in the comparison. Moreover, T-Mart, which focuses solely on adjacency effect correction, was also evaluated in combination with the ACOLITE model. PRISMA data were selected based on the synchronicity with the in situ measurements and the percentage of cloud cover, which was required to be less than 2%. The dataset finally consists of 174 images with in situ data derived from the AERONET-OC network and 65 images with in situ data obtained from the CVD network, for a total of 239 images. The match-up analysis is provided both per site and across the OWTs; the Quality Water Index Polynomial (QWIP) is also included in the analysis to identify Rrs spectra that fall outside the general trends observed in aquatic optics for optically deep inland and coastal waters. So far ACIX-III Aqua results seem consistent with previous findings of ACIX-Aqua for Landsat-8 and Sentinel-2. In particular, we observe largest differences between satellite and in situ measurements associated with the blue bands and NIR bands, and that each AC model has different degrees of accuracy depending on OWTs. Further results will be presented based on the latest progress in data analysis to understand how the variations in AC performance are related to the increased hyperspectral information content of PRISMA and how the results of this exercise compare to those obtained with other satellites to support the generation of improved products for aquatic applications from hyperspectral data collected by other current and future missions (e.g., CHIME). To the aim, the last section shows the first results obtained with EnMAP for 2022 to 2023 by applying three of the AC processors used with PRISMA (i.e. ACOLITE, MIP and POLYMER), and also including the PACO land AC processor in combination with WASI (PACO-WASI). In this case, the dataset consists of 25 images with in situ data derived from the AERONET-OC network and 21 images with in situ data obtained from the CVD network, for a total of 46 images.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.