A Contemporary Approach to Streaming Weather Data to Better Support Nowcasting

Nowcasting, which includes comprehensive awareness of current weather and prediction of weather in the very near future, is becoming an increasingly important meteorological discipline because of its potential to inform the public of severe weather with better fidelity and timeliness and to address the needs of emerging new commercial businesses such as commercial space launch and drone operations. Nowcasting is often constrained by the inability to receive satellite, radar, radiosonde, and other observational data in a sufficiently timely manner. A confluence of trends in remote sensing and information technology, and regional and microscale weather prediction are enablers to address and overcome this challenge. Today’s environmental observations are more rapidly sensed at higher resolution and frequency than ever before. This trend will continue. Cloud computing environments and increasingly higher data rate networks provide ubiquitous, low cost access to highly available and scalable computational resources, and the capability to distribute observation data more effectively. Nowcasting applications are leveraging advances in weather model science, including the capability to use a wide variety of observation data types that execute over smaller regions more frequently. What remains is the need for a contemporary, open, standards-based data format and transport technology that minimizes the time between observation and use by nowcasting applications. Our paper introduces Streaming Objects that enable concurrency between data acquisition, product generation, and product distribution. Each Streaming Object is a Unidata Common Data Model (CDM) dataset that is consumable independently by a weather application. A Streaming Object also defines its relationship with a subsuming traditional weather product that typically includes a greater timespan and geographic region. Streaming objects are configured to balance the need to contain a cohesive amount of data and minimize the time between when remote sensing occurs and when the observational data is available to the weather application. Our demonstration system makes use of the GOES-R Rebroadcast (GRB) to generate Radiances, Cloud and Moisture Imagery, and Lightning Detection Streaming Objects using a streaming-enhanced version of the open source Community Satellite Processing Package (CSPP) for GOES-R developed by the Cooperative Institute for Meteorological Satellite Studies. Our demonstration system is a cloud-native application with a web-browser thin client. Our paper describes our demonstration system and documents its performance characteristics.