Lessons Learned from Building an Earth Observation Digital Twin in One Year

Every day, the National Environmental Satellite, Data, and Information Service (NESDIS) produces hundreds of Earth data products. NESDIS employs complex strategies to govern the ground system, which is the sensors-to-user pipeline for downlinking, ingesting, processing, and disseminating data. Technologies like machine learning and computer vision provide opportunities to enhance the ground system. Digital Twins can serve a dual role by optimizing the ground system's performance and acting as a sandbox for testing enhancements without disrupting existing operations.

This presentation describes our rapid efforts to build an Earth Observation Digital Twin (EO-DT) prototype to enhance data discovery, reduce dissemination challenges, and automate data quality control within NESDIS. We focus on critical features, such as rapid data cataloging, querying, as well as data anomaly detection using computer vision. Using an air quality case study, we also explore fusing data products from multiple sources, which can create analysis-ready datasets for seamless end-user access. Numerous digital twin projects are underway, and the community envisions this effort will lead to a federation of interconnected digital twins. We will share our software approach, enabling technologies, successes, and lessons learned to work toward building an EO-DT and ensuring its interoperability with sibling efforts.