450 Artificial Intelligence to Improve Detection and Classification of GOES-R Ground System Product Artifacts

Tuesday, 30 January 2024
Hall E (The Baltimore Convention Center)
Brad Brown-Bergtold, L3Harris Technologies, Inc., Melbourne, FL; L3Harris Technologies, Melbourne, FL; and J. Masson and B. Brown-Bergtold

Quality Watch is an automated AI/ML ground processing artifact detector and classifier trained to operate on ABI imagery. Its primary purpose is detecting ground system issues that affect product generation. Training data was generated for missing blocks, Caterpillar Tracks, Solar Avoidance, and Shark Fin artifacts using ABI L1b Radiance imagery available from NOAA’s Big Data project. A classifier was trained via the ENVI Deep Learning module hosted in an AWS cloud instance. The resulting classifier was hosted in an AWS cloud instance and processed GOES-16 ABI data in near real time, easily keeping up with the high volume of product production. Detection results were indexed and displayed on a live dashboard. Quality Watch provides operators a unique, automated method to view GOES-R GS Data Operations system behavior and product artifacts.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner