A VAE trained on coastal bio-optical data retrieves phytoplankton absorption spectra and chlorophyll a from EMIT/PACE hyperspectral reflectance with more stable performance than an MDN baseline.
Mixture density networks for re-constructing historical ocean-color products over inland and coastal waters: demonstration and validation
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Variational Autoencoder Framework for Hyperspectral Retrievals (Hyper-VAE) of Phytoplankton Absorption and Chlorophyll a in Coastal Waters for NASA's EMIT and PACE Missions
A VAE trained on coastal bio-optical data retrieves phytoplankton absorption spectra and chlorophyll a from EMIT/PACE hyperspectral reflectance with more stable performance than an MDN baseline.