Deep Neural Networks are trained on ARC supercomputer facilities to replicate the precise output of MODTRAN 6.0 calculation. The required training data is collected by running the inefficient MODTRAN 6.0. An appropriate deep neural network architecture in order to obtain the required accuracy in emulating MODTRAN 6.0 needs to be researched based on the science of the problem.
More »Visible Shortwave Infrared (VSWIR) atmospheric correction is key for global terrestrial & aquatic ecosystem studies. Capability need: 2 order of magnitude speedup. 0.1 seconds per iteration enables accurate iterative numerical inversion. Potential applications include: • Atmospheric correction for airborne VSWIR spectrometers (AVIRIS-NG, AVIRIS-C, ACIS, and future instruments). • Orbital VSWIR spectrometers (EMIT, SBG) • Extensible to water column modeling for ocean color, inland water quality, and coastal ecosystem studies (PRISM, CORAL, PACE, EV-S proposals) • Extensible to thermal IR for HyTES and longwave IR imaging spectrometers for Earth and planetary
More »Organizations Performing Work | Role | Type | Location |
---|---|---|---|
Jet Propulsion Laboratory (JPL) | Lead Organization | FFRDC/UARC | Pasadena, California |
University of Massachusetts-Amherst (UMASS) | Supporting Organization | Academia | Amherst, Massachusetts |