Analysis of hyperspectral and multispectral reflectance spectra in the Black Sea coastal area near the Danube delta: comparison of PRISMA and Sentinel-2 observations

Abstract

Abstract In this study we investigate the possibility to distinguish between different water types in the Danube delta and the Black Sea coastal zone based on the reflectance spectra. For this we use hyperspectral satellite images from PRISMA (PRecursore IperSpettrale della Missione Applicativa) and multispectral images from Sentinel-2 MSI (MultiSpectral Instrument) in July 2020. The visual inspection of the available acquisitions differentiates between four aquatic types: lagoons; river and stream plumes mixing with marine waters; turbid and optically complex coastal waters; and optically clean waters away from the shore. For each of the four types we determine the characteristic averaged reflectance spectra from PRISMA and Sentinel-2 for the spectral range from 400 nm to 2500 nm. It is found that Sentinel-2 spectra are very similar for the 4 water types, in contrast to the PRISMA spectra which give substantial additional information. Further we analyse the gradual modification of the characteristic hyper- and multispectral spectra from the shoreline to the open sea (∼25 km distance) in 15 locations passing through areas of varying turbidity. We demonstrate that the intensity of surface reflectance from the hyperspectral instrument decreases gradually with distance from shoreline, clearly showing the transition zone between riverine and marine water. The multispectral reflectance spectra for the same study areas did not present such consistent behaviour. The presented results demonstrate the benefits of hyperspectral over multispectral images for turbid aquatic regions in the Black Sea coastal zone. They show that with little requirements regarding pre-processing and computational resources hyperspectral data can contribute greatly to classification of water types, in respect of their turbidity.