A008-0018
Meso-anywhere sectors from GOES-R - optical flow techniques for global 1-minute imagery
Meso-anywhere sectors from GOES-R - optical flow techniques for global 1-minute imagery
Monday, 7 December 2020
Poster
Abstract:
Novel technologies in video processing and frame interpolation allow for unique opportunities in expanding the temporal resolution of ABI observations from the GOES-R series of satellites. Open-source products using OpenCV and other solutions developed for the film and media industries can be used to generate scientific datasets of interest, but require detailed analysis to determine the suitability of these products for operational and research use.
In this presentation, 1-minute observations, created by pyramid-based optical flow schemes, will be presented, using CONUS (native resolution: 5-minute) and full-disk (10-minute) scans from the ABI instrument aboard GOES-16 and GOES-17 as source material. Qualitative and quantitative analysis of simulated 1-minute imagery made against 'native' 1-minute imagery will be demonstrated, as well as strategies to optimize accuracy and reduce error in the simulated 1-minute imagery.
Applications include media and public affairs use, as well as generation of custom meso-sector imagery to expand the capability of the current GOES-R platform from having two 1-minute meso-sector scans, to potentially limitless scan capability. Opportunities for applications to RGB and other retrieved properties will also be presented.
In this presentation, 1-minute observations, created by pyramid-based optical flow schemes, will be presented, using CONUS (native resolution: 5-minute) and full-disk (10-minute) scans from the ABI instrument aboard GOES-16 and GOES-17 as source material. Qualitative and quantitative analysis of simulated 1-minute imagery made against 'native' 1-minute imagery will be demonstrated, as well as strategies to optimize accuracy and reduce error in the simulated 1-minute imagery.
Applications include media and public affairs use, as well as generation of custom meso-sector imagery to expand the capability of the current GOES-R platform from having two 1-minute meso-sector scans, to potentially limitless scan capability. Opportunities for applications to RGB and other retrieved properties will also be presented.