Monday, 13 January 2020: 2:15 PM
156A (Boston Convention and Exhibition Center)
As new observation platforms are launched into orbit, the amount of satellite data generated globally is increasing rapidly. Transmission of large amounts of data to ground stations is limited by available bandwidth. Also, the amount of data that can be ingested into numerical models is limited, especially if the models are run within operational time constraints. In this study, a deep learning approach is applied as a way to compress satellite data intelligently. By learning the latent representations of the complex patterns in high dimensional satellite data, deep learning provides a method for creating statistically-optimized lossy representations of the of the observational data as well as a method to decompress the resulting latent representations. This methodology is compared to other approaches to compress observational satellite data for transmission and consumption by numerical models.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner