USC Viterbi Engineers Develop Models to Test LA River Restoration Concepts

Much of the Los Angeles River runs through a concrete channel built for flood control: efficient at moving stormwater, but cut off from the floodplain and habitat that once surrounded it. At Taylor Yard, a former rail yard in the Cypress Park neighborhood of northeast Los Angeles, one of the few remaining stretches of open land along the river offers a rare opportunity: to restore those natural connections and improve public access while keeping flood protection intact.

Since 2024, a team of USC researchers and students has been exploring how to do exactly that, using physical and computational models to test whether restoration concepts can work before they are built.

As part of the USC River Team - a collaboration between USC School of Architecture, USC Viterbi School of Engineering, USC School of Cinematic Arts, and USC Dornsife Public Exchange - USC Viterbi researchers bring two complementary tools to the design process: a 60-foot physical hydraulic model built at a scale of 1:120, and Celeris, a browser-based computational simulation.

The physical model lets researchers observe selected designs under scaled flood flows, capturing three-dimensional effects that simulation cannot fully reproduce. Celeris runs faster than real time, allowing the team to compare many configurations quickly. Together, the tools give designers and project partners an evidence base for evaluating proposals as they emerge.

The USC River Team

At the USC School of Architecture, Associate Professor Alexander Robinson and his research group develop landscape concepts that account for habitat, public access and site constraints. To evaluate those restoration concepts as they emerge, Mitul Luhar, associate professor at USC Viterbi's Department of Aerospace and Mechanical Engineering and Sonny Astani Department of Civil and Environmental Engineering, guides physical hydraulic modeling with a team of architecture and engineering graduate and undergraduate students. Luhar's colleague Patrick Lynett, a coastal engineer and professor of civil and environmental engineering, adapts computational simulations to test how design changes affect river flow and stormwater channeling.

Associate Professor Andreas Kratky and colleagues at the USC School of Cinematic Arts contribute interactive visualization and augmented-reality tools that enable project partners to easily test flow scenarios and site constraints alongside the physical model. The City's Bureau of Engineering provides the Hydraulic Research Laboratory where the model is based; the Mountains Recreation and Conservation Authority is a project partner; and the team coordinates with the 100 Acre Partnership at Taylor Yard with support from USC Dornsife Public Exchange.

Modeling the river's response

Recently released by Public Exchange, the Taylor Yard River Connection Study outlines three design concepts: North Island routes a side channel around a vegetated island; South Tributary creates a groundwater-fed basin and shaded stream connected to the river; and Parabolic Terracing reshapes the riverbank to make more room for floodwaters, habitat and public use. The designs must account for contamination, power and transportation infrastructure, and historic railway features.

Earlier Taylor Yard proposals faced uncertainty about flood effects and the high cost of soil remediation. The USC River Team took a different approach, introducing hydraulic modeling to the design process from the outset. Engineering analysis enters the design process through two tools: a physical model and a computational simulation.

Luhar and his research team apply fundamentals of fluid mechanics to measure how changes to the channel and vegetation affect water movement. These tests are conducted in a 60-foot model of the Taylor Yard reach, built at a scale of 1:120 in the City's Hydraulic Research Laboratory. The USC River Team and project partners can reshape the channel and add calibrated vegetation of different heights and densities, while cameras track water levels and overtopping. Tests represent flood events ranging from 5-year to 100-year scenarios. "From the start, we wanted to make this model interactive and a shared design resource," said Luhar.

Lynett applies his expertise in coastal engineering and computational modeling through Celeris, which he adapted to simulate river flow from a tool originally developed for coastal waves and harbor engineering problems. "River-flow equations are a simplified version of the equations used for coastal waves, so I could adapt Celeris to simulate the river," he explained.

Users can sketch a side channel or add vegetation and see how the simulated flow changes. They can also test responses to flooding: "If a design flood overtops the riverbank in one area, you can draw in a higher flood wall or add vegetation and see how the flow changes," Lynett said. The USC River Team tried dozens of variations in the simulation, adjusting channel routes and inlet and outlet locations around contamination and infrastructure before transferring selected designs to the physical model.

Measurements from the hydraulic model and ongoing fieldwork also help calibrate Lynett's predictions.

A third tool bridges the technical and the collaborative. Kratky and colleagues at the USC School of Cinematic Arts created an augmented-reality interface that overlays flow scenarios and information, including vegetation, contamination and easements. At workshops, engineering consultants and 100 Acre Partnership representatives used the iPads to explore the overlays, annotate them and sketch possible changes.

Early results and next steps

Overall, USC Viterbi's hydraulic analysis informs the USC Architecture design concepts, and site constraints and partner priorities help determine which alternatives the engineers test. Lynett described the engineering team's role as equipping designers and stakeholders with "the best information that we have available to us on how this system functions," so they can assess how proposed features might behave together.

Early results point to both possibilities and trade-offs. Some North Island configurations reduced potential overtopping near the site, though downstream effects may require mitigation. Vegetated terracing increased flood capacity in some tests; further work will assess performance and constructibility. Modeling found that the groundwater-focused South Tributary would not affect storm-event flow in the main channel.

Contamination, utilities and coordination with proposed public parkland remain part of the design challenge. Luhar hopes the next phase will also bring real-time river data into the models. "The big advance we're hoping for is starting to do monitoring on the river, to see whether what our models predict is consistent with what we're observing," he said. Comparing predictions with river conditions could help answer open questions about sediment and groundwater - and show where approaches tested at Taylor Yard might work elsewhere along the LA River.

Published on November 10th, 2026

Last updated on November 10th, 2026