Summary

Operational forecast models have been deployed using the ADCIRC model via the Floodwater system. These models produce time series data that is useful for evaluating how the operational system is running and the differences in predictions due to meteorology or model geometry. Previously, this has been visualized by either (a) pre-processing select locations within the model for time series visualization or (b) bulk pre-generating time series throughout the domain. Each of these approaches has their disadvantages that this work seeks to resolve: Predefining locations can be limiting, and bulk processing all locations can be time consuming and highly inefficient.

This project will automate the conversion of forecast model output into cloud-native formats and provide facilities for connecting these data to web applications. Cloud-native formats reorganize data so that subsets (such as time series data for a single location) can be efficiently queried without loading the entire model output. The workflows will be automated using a workflow manager running within the cloud. An API will be provided for querying the data and returning JSON format which can be readily used by web applications. The team will incorporate the time series API into the FloodID application and work with Blanton to connect the system to the APSViz interface. The time series API will be deployed ahead of the 2025 hurricane season, however, it is expected that significant progress will be made prior to that and initial testing of the system can begin in the fall/winter of 2024.

Investigator