Q&A with Dr. Feyera Hirpa and Matthew Bachman on Their Presentation at FMA for Strengthening California’s Flood Resilience through Innovation and Collaboration

By Quinn Fenger, Feyera Hirpa, and Matthew Bachman

California’s Central Valley is the state’s agricultural heartland and is home to some of its greatest flood risks. Every five years, the Central Valley Flood Protection Plan (CVFPP) is updated to assess how those risks are changing over a 50year horizon. The next iteration, due in 2027, leans on new climate science, enhanced modeling and closer collaboration among agencies and consultants. To understand why this work matters, we spoke with two of the people leading the technical analysis at GEI Consultants: Dr. Feyera Hirpa, a climate scientist with more than a decade of experience in flood prediction and hydrologic modeling, and Matthew Bachman, a water resources engineer who also has more than a decade of experience developing and applying water resources planning models.

In this conversation, Hirpa and Bachman explain how the 2027 update differs from past plans and what insights attendees at their presentation at the Floodplain Management Association (FMA) conference can expect.

Q: For readers who aren’t familiar, what is the Central Valley Flood Protection Plan and why does it matter?

A: Feyera Hirpa: The CVFPP is a living document that’s updated every five years. It quantifies how flood risk will evolve in the Sacramento and San Joaquin basins over the next half century, and it becomes the reference for flood risk management, planning, and infrastructure investments. Agencies use it to decide, for example, whether to raise levees around cities or make other long term improvements. Even though we didn’t work on earlier editions, we looked closely at them; they’ve been published since 2012 and show how quickly modeling techniques and climate data have advanced.

Q: What’s different about the 2027 update?

A: Hirpa: Two big changes are the data and the models. Climate models are now more detailed and accurate, and we have higher resolution geospatial data. We’re running the variable infiltration capacity (VIC) model on a three-kilometer grid instead of six, which provides four times the spatial resolution. There’s also a new routing model called mizuRoute, with a different modelling philosophy compared to the previous iterations. Beyond the technical upgrades, the climate analysis itself is different: we have used a framework called Decision Scaling. Rather than simply running climate projections through a hydrologic model, we first stress tested the watershed to see what kinds of climate conditions would break the system and then we overlaid future climate projections to understand the likelihood of those stressful events. It’s a bit of an inverted approach compared to past iterations.

A: Matthew Bachman: Another major change is that we expanded the range of climate futures we evaluated. In the previous (2022) iteration, we evaluated three future scenarios: high, median, and low climate change. For this update, we analyzed 24 combinations of future warming and precipitation changes. This gives us a more robust picture of the future and helps planners prepare for extremes. That’s part of the collaboration theme—bringing in other agencies and researchers to evaluate different methods and compare results.

Q: What innovations or tools stand out for you?

A: Bachman: In addition to the modeling and framework updates, we used an improved downscaling technique to translate global climate model outputs more granularly into local hydrologic data. Some of the data sets we’re using didn’t exist twenty years ago. And while most of our models remain physics-based, there’s growing interest in machine learning tools to calibrate and refine simulations.

Q: Were there any results that genuinely surprised you?

A: Bachman: Even though we anticipated earlier, flashier storms, I was struck by how much the volume of runoff shifts to the earlier part of the year and how dramatically the spring snow‑melt peak declines. The magnitude of the change exceeded my expectations. When we talk about a “100-year” flood today, the models suggest that by 2077 the magnitude of that flood could double. That’s stark.

A: Hirpa: Exactly. Warmer air holds more moisture, so extreme precipitation events become more likely, and the effect on flood peaks is huge. And remember, the snowpack acts like a natural reservoir. On average the Sierra snowpack provides about a third of California’s water supply, but as more precipitation falls as rain instead of snow, we lose that storage and get more runoff all at once. We expected changes, but the size of those changes—more frequent, higher peaks—was eye opening.

Q: How will these changes affect infrastructure and communities?

A: Hirpa: In simple terms: if your property is currently exposed to flooding, it will get worse. Peak flows will increase in both the Sacramento and San Joaquin valleys. We don’t model inundation directly—that’s another team’s role—but our results feed into those hydraulic models that map flood extents.

Q: What do you hope attendees take away from your presentation?

A: Bachman: Even if you follow climate science, you might not realize how quickly extremes are shifting. The magnitude of the changes we’re seeing from this modeling are more striking than people might expect. I hope our work underscores the need for infrastructure investment, especially in flood protection, and encourages decision makers to prepare for a wider range of future scenarios.

A: Hirpa: My takeaway is that we can’t ignore climate warming. It’s not just an abstract idea; it has real implications for flood risk and water supply. As snow shifts to rain and peaks intensify, we need to adjust our lifestyles and investments accordingly. This update provides the information to do that.

Q: Are there remaining challenges or uncertainties you’d like to see addressed in future work?

A: Hirpa: Uncertainty is always there. Climate data contain noise, and even with better models we can’t predict with absolute certainty what will happen in 50 years. But modelling techniques are improving, computational power is increasing, and data sets are becoming richer. As we refine our models and incorporate more observations, perhaps with emerging AI tools, the signal will become clearer. My hope is that future updates will provide even greater confidence and resolution.

A: Bachman: I’d add that we need broader engagement. It’s not enough to produce technical reports; we have to communicate the risks clearly and work with communities, state agencies, and researchers to translate these insights into action.