A digital twin for the Advanced Quantitative Precipitation Information system

Abstract

With a population in the Bay Area exceeding seven million, understanding potential flooding and transportation impacts is essential for emergency and city managers. Weather prediction in the Bay Area is complicated by the geography, which creates gaps in radar coverage, and by the phenomenon known as an Atmospheric River (AR). Midlatitude cyclones and the Pineapple Express derive their strength from ARs. ARs contribute 20% to 50% of California's precipitation and streamflow over 5-15 days throughout the year. The ability to identify ARs through satellite imagery and forecast their timing is crucial for accurate predictions in the Bay Area. With precise and timely forecasts, emergency managers, reservoir operators, and water managers can respond proactively. The Bay Area Advanced Quantitative Precipitation Information (AQPI) System integrates multiple data sources—such as satellites, radars, and sensors—and uses model processing to generate forecasts ranging from a few minutes (Nowcast) to several months (long term). Like any complex system, its successful deployment and operation require many components. There is also both a short-term and long-term perspective when it comes to water resource management. Evaluations of technical and policy changes must consider all aspects. Adopting the Complex, Large-scale, Interconnected, Open, Sociotechnical (CLIOS) process from Joseph Sussman allows engineers, policymakers, and institutions to assess the impacts of technical and policy decisions on water systems. The CLIOS process offers an approach to evaluate a system, while Model-Based Systems Engineering (MBSE) provides a set of standards and protocols for developing a Digital Twin of the system. Based on physical systems, Digital Twins enable engineering teams to assess potential updates and changes to a system before expending resources. This dissertation describes the development of a Digital Twin for the AQPI Public Website. Following the CLIOS process, it outlines the approach for creating an MBSE AQPI model, showing how each step of the CLIOS process aligns with various diagrams and features of the System Modeling Language. It captures the AQPI system data flow dependencies and user interface requirements to build a Digital Twin. An extensive review of the identified Use Cases and supporting activities provides details on the exchanges and is validated by simulation execution using data files directly from the AQPI system.