613 Seminar/Meet with Speaker: Adam Milstein/MIT Lincoln Laboratory

See TEAMS meeting information following abstract
Adam B. Milstein
MIT Lincoln Laboratory

Computational Imaging in Infrared Remote Sensing of the Atmosphere

Abstract: NASA’s recent Earth science and technology incubation efforts have emphasized smaller, more flexible hardware platforms and improved science algorithms that can fuse information from heterogeneous sources.  We describe two recent efforts at MIT Lincoln Laboratory (MIT-LL) addressing these goals with infrared remote sensing.  First, we describe our recent hardware efforts.  The Computational Reconfigurable Imaging Spectrometer (NASA) is new IR instrument concept that uses a coded mask and platform motion to make spectral measurements.  We show that this approach offers more flexible scan modes, as well as sensitivity advantages over traditional designs, enabling smaller infrared instruments and reducing cooling requirements.  Second, we describe our recent science algorithm efforts. MIT-LL has developed neural network algorithms for NASA’s hyperspectral IR sounder instruments, including temperature and moisture first guess for the AIRS Level 2 retrieval, resulting in improved global accuracy in cloudy atmospheres versus previous versions. We describe our recent efforts aiming to fuse modern AI and physics-based approaches in a more explainable way, with application to improving vertical resolution in the planetary boundary layer.

Bio: Dr. Adam B. Milstein is a Member of the Technical Staff at MIT Lincoln Laboratory. His research interests include neural network retrieval algorithms for hyperspectral infrared sounding, computational imaging, and small satellite instrument development.  Dr. Milstein is currently PI on NASA funded efforts including the Science of Terra, Aqua, and Suomi NPP and Advanced Component Technology under the Earth Science Technology Office.

613 Seminar Series Coordinators 

Please do not forward the meeting invitation to outside recipients as the presentation, discussions, and questions may contain proprietary or confidential information.