California tests flood 'digital twins' as a communication device, not a modeling upgrade

A UC Berkeley-led team is building a hydrodynamic digital twin of the Lower Pajaro River watershed, including the levee system now being rebuilt, in partnership with the Pajaro Regional Flood Management Agency and state, municipal and community organizations ^[7]. The interesting part is not the physics. It is the assertion that a three-dimensional simulation of a specific management decision will land with residents and policymakers in a way that a FEMA map does not ^[3].

Roughly 1.4 million Californians live in areas at high annual flood risk, a number climate change is pushing up ^[1]. According to the project's framing, the standard instruments fail as communication: looking at a FEMA map, or learning that a house sits in a 100-year floodplain, does not tell a person how likely a flood is or what happens when one arrives ^[2]. "When you look at a flood risk map, you only see a flat depiction of the hazard area, and this is what informs most of the decision-making around reducing risk," said Anna Serra-Llobet, a senior scientist at the Berkeley Center for Smart Infrastructure, who co-leads the project ^[4][5]. She frames the twin as a way to see flooding in three dimensions and watch processes change over time ^[4].

The test site is well chosen, which is to say it is a place where the old approach visibly did not work. Watsonville and the unincorporated community of Pajaro sit on either side of the Lower Pajaro River and have had five major floods since the levee system was finished in 1949 ^[6], an average of about one every 15 years ^[14]. The March 2023 flood caused an estimated $300 million in damage ^[8]. The levee upgrade arrived after decades of delays ^[9].

The substantive case for visualization is the setback levee, which is pulled back from the channel to partially restore floodplain ^[11]. Done well it can cut flood risk while adding groundwater recharge, wetland habitat and park space ^[11]. It can also displace an established residential community that may already be socially or economically vulnerable ^[12]. That is a distributional trade, not a hydraulic one, and the Berkeley team's argument is that a twin can carry hazard, ecology and social vulnerability in one object and make the pros and cons of a given scenario explicit ^[13][10]. Kofi Boone of North Carolina State University is on the project for community engagement and environmental justice expertise ^[15], which is a reasonable signal that the deliverable is a deliberation aid rather than a report.

The tooling is imported. The model is being developed by Berkeley PhD candidate Gabriela Paredes with VRVis GmbH researchers Jurgen Waser and Silvana Rauer-Zechmeister ^[15], using software called scenarify from the Vienna Research Center for Visual Computing; the project would be the first U.S. digital twin for flood risk management built with it ^[16]. VRVis already runs flood-scenario twins in Austria, Germany and Denmark ^[17]. Adda Athanasopoulos-Zekkos, an associate professor of civil engineering at the Center for Smart Infrastructure, co-leads with Serra-Llobet ^[5].

What to watch: whether the twin is finished and used while the Pajaro levee design still has open decisions, or arrives after the concrete. Watch also for any evidence that viewing a scenario changed a stated preference or a vote, since the project's whole premise is comprehension, and comprehension is measurable. The team says the approach could be extended to other flood-prone regions of California if it succeeds ^[18], but success here has not yet been defined in the material.