Spaceborne hyperspectral remote sensing — imaging spectroscopy that acquires reflectance data across hundreds of narrow, contiguous spectral bands — offers transformative capability for inland water quality monitoring compared to conventional multispectral sensors. Where Sentinel-2's 13 spectral bands must approximate water quality parameters through band ratio indices, hyperspectral sensors resolve the full visible-to-SWIR spectral curve of the water-leaving radiance signal, enabling application of physically-based bio-optical models that retrieve water quality parameters (chlorophyll-a, phycocyanin, CDOM absorption, turbidity) with accuracy approaching in-situ measurement.
Two operational spaceborne hyperspectral sensors are currently available for inland water applications: ASI's PRISMA (PRecursore IperSpettrale della Missione Applicativa), launched in 2019, provides 239 spectral bands from 400–2500nm at 30m spatial resolution and 30km swath width; DLR's DESIS (DLR Earth Sensing Imaging Spectrometer), operational on the ISS since 2018, provides 235 spectral bands from 400–1000nm at 30m spatial resolution and 30km swath width. Both sensors represent the transition from research to operational hyperspectral capability in advance of the ESA CHIME and NASA SBG missions planned for the late 2020s.
This evaluation processed 847 PRISMA and 312 DESIS acquisitions over 24 inland water bodies across four continents, spanning lakes (oligotrophic to hypereutrophic), reservoirs, and large river systems. In-situ validation data comprised 1,840 coincident water quality measurements collected within ±3 hours of satellite overpass. Three retrieval algorithms were evaluated: Mixture Density Network (MDN) bio-optical inversion; QAA (Quasi-Analytical Algorithm); and a Random Forest approach using full-spectrum reflectance as input features.
For chlorophyll-a retrieval in eutrophic to hypereutrophic systems (the operationally critical range for cyanobacterial bloom detection), PRISMA with MDN retrieval achieved RMSE of 8.2 μg/L and R² of 0.93 across the full evaluation dataset. DESIS performance was comparable (RMSE 9.4 μg/L, R² 0.91) in oligotrophic to mesotrophic systems but degraded more severely in highly turbid conditions due to its limited spectral range (400–1000nm) excluding the SWIR bands used by MDN for turbidity correction. For phycocyanin retrieval — the pigment most specific to cyanobacteria and therefore most valuable for bloom species discrimination — PRISMA's SWIR coverage provided a 23% RMSE improvement over DESIS.
Operational deployment recommendation: PRISMA is preferred for cyanobacterial bloom monitoring and full-parameter retrieval in complex optically active water bodies. DESIS provides an adequate, freely accessible alternative for chlorophyll-a and CDOM retrieval in clear to moderately turbid systems where SWIR-based turbidity correction is not required. The Prime Logic Monitoring Command Center's Water Quality Intelligence module supports both sensor data streams with automated Level-2 processing, MDN retrieval, and lake health scoring against WHO recreational water guideline thresholds.