Transforming Data to Information: The Ongoing Evolution of Satellite Products

Satellites are a critical data source for hazard mitigation and response. In particular, next generation satellites, such as GOES-16, Himawari-8, and NOAA-20, have the potential to revolutionize hazard detection, characterization, nowcasting, and response. In order to maximize the value of satellite measurements, sophisticated computer algorithms are needed to transform satellite data and other relevant data sources to actionable information. Operational meteorological satellites, alone, currently provide over 1 trillion earth observations each day, so human expert analysis of satellite imagery must be supplemented with information derived from automated processing. The traditional satellite product suite consists of imagery and a complementary set of satellite derivable geophysical parameters. The traditional satellite specific product suites, while valuable, generally require human experts to manually blend many relevant products and extract pertinent information for decision-making. Given the enormous data volume and the rapidly evolving nature of many hazards, the traditional product paradigm does not allow for full utilization of satellite measurement capabilities. Satellite product capabilities have not evolved as quickly as the measurement capabilities, so there is a tremendous opportunity for further exploitation of satellite measurements. One way to improve satellite measurement exploitation is to build upon traditional (foundational) products and develop multi-data-source application systems that directly support decision-making. In this presentation, the need for higher-order integrated satellite products is motivated and illustrated using applications aimed at mitigating volcanic, severe weather, and fog related hazards. More specifically, the development and deployment of the VOLcanic Cloud Analysis Toolkit (VOLCAT), the Probability of Severe (ProbSevere), and the fog/low stratus (FLS) application systems illustrate the promise of higher order products and machine learning based data integration. Key research and operational challenges, associated with higher order products, will also be highlighted.