POSTER SESSION 3
15:30–17:30, Wednesday, September 16
Poster Session | 1 | 2 | 3 | 4 | Schedule at a Glance
ABSTRACT 1123 | POSTER 3
USING HYPERSPECTRAL UNMIXING TO DERIVE OPTICAL WATER TYPES FROM SATELLITE DATA
Remote sensing is a powerful tool for monitoring marine ecosystems, where satellites have enabled observations at vast spatial and temporal scales. Historically, the main application of ocean-color satellites has been in open-ocean Case-1 waters, which are less optically complex than coastal and inland regions that more often exhibit Case-2 conditions. In contrast, the higher optical variability of Case-2 waters makes it challenging to develop general ocean-color algorithms that perform well across all environments.
Optical Water Types (OWTs) have therefore been proposed to add nuance beyond the binary Case-1/Case-2 classification, and have been shown to improve water-quality parameter (WQP) retrievals by selecting and blending algorithms according to the prevailing water type. To derive such OWTs from observed reflectance spectra, fuzzy c-means clustering has become a common approach, yielding a set of cluster centroids that are interpreted as different water types.
In parallel, spectral unmixing has emerged as a powerful technique in hyperspectral remote sensing, estimating constituent spectra (endmembers) and the fractional abundance of each endmember within every pixel. This suggests a clear conceptual parallel with OWT classification, raising the question of how effectively spectral unmixing can be used to derive OWTs, and what the associated trade-offs are.
In this work, we investigate the use of spectral unmixing to derive OWTs from hyperspectral data from the PACE and HYPSO satellite missions.
Aria Alinejad*, Norwegian University of Science and Technology, [email protected], https://orcid.org/0009-0006-1673-8394
Vishnu Perumthuruthil Suseelan, Norwegian University of Science and Technology (NTNU), [email protected]
Poster Session | 1 | 2 | 3 | 4 | Schedule at a Glance
