E26 Recovering Historic Geostationary Satellite Data: Preserving ATS and SMS/GOES Observations (1966-1996)

Monday, 29 January 2024
Hall E (The Baltimore Convention Center)
Jerrold Robaidek, CIMSS, Madison, WI; and D. A. Santek, T. J. Schmit, J. Phillips, D. Forrest, A. De Smet, M. Drexler, and S. Reiner

Geostationary weather satellite data offers invaluable insights into Earth's climate and weather patterns. However, a digital gap exists for 1966-1974. Helping address this, over 66,000 ATS-1 and ATS-3 paper records from this period were scanned and digitized. Additionally, while a digital dataset for SMS and GOES (1978-1996) is available, over 900,000 images from this dataset were also reprocessed resulting in thousands of previously unaccounted for images.

For ATS data, paper images were scanned into a digital format, then Optical Character Recognition (OCR) was used to extract metadata. The metadata were then used to label images with host satellite, image dates, and times. In addition, state-of-the-art Language Learning Models (LLMs) were used to aid metadata validation. Software was also written to detect the Earth limb and center, which aided in automatic image alignment and navigation. In the SMS/GOES effort, we reprocessed data from the original tapes, revealing images previously missed due to errors in the original data, and incorporated external navigation for images lacking it.

Progress, reprocessing challenges and datasets release schedule will be discussed.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner