BAY DELTA SCIENCE CONFERENCE: Delta Flood Risk Under Climate Change: Key Findings from the Delta Adapts Flood Risk Analysis

The Delta Stewardship Council, with the assistance of partner agencies and consultant AECOM, developed a climate change vulnerability assessment for the Sacramento-San Joaquin Delta as part of the Delta Stewardship Council’s Delta Adapts: Creating a Climate Resilient Future initiative.  The climate change vulnerability assessment is the precursor to developing an adaptation plan for the Delta that will detail the strategies and tools that state, regional, and local governments can use to help communities, infrastructure, and ecosystems thrive in the face of climate change.

The climate change vulnerability assessment, finalized in spring of 2021, included an analysis of the flood risk for the Delta that was used to develop flood maps that show existing and future projected flood risk throughout the Delta in terms of the likelihood of flooding. These maps will be used in a wide array of public outreach and education settings as well as policy and investment decision making. Ultimately, the results will inform future climate change adaptation decisions, investments, and policies.

At the 2021 Bay-Delta Science Conference, Andrew Schwarz,  State Water Project Climate Action Advisor with the Department of Water Resources and one of the collaborators on the assessment, shared key flood risk hazard assessment findings. 

He began by acknowledging the other collaborators on the project: Cory Copeland (DSC), Kaylee Griffith (DSC), Justin Vandever (Aecom), Phillip Mineart (Aecom), and Ricky Torres-Cooban (Aecom).   He also acknowledged the assistance of DWR staff, CVFPP staff, and the Delta levee district engineers that helped QA-QC the data for levee elevations used throughout the process.

The assessment looked at several asset resource categories.  In this presentation, Mr. Schwarz will discuss the flood risk in the Delta in general, but there is much more detail in the report on the flood risk of each of these different resource and asset categories.

The Flood Hazard Analysis covers the entire watershed flowing into the Delta, and with the Delta being at the bottom of the watershed, what happens upstream of the Delta matters.  They evaluated sea level rise, storm surge, tides, inflow from the river system, the effects of climate change on river flows, and the possibility of high tide and storm surge co-occurring at the same time as a high inflow event.

Mr. Schwarz said that because it is very complicated, they looked at flooding due to levee overtopping only and not any other modes of failure, so in some ways, it’s a very conservative look at what potential flooding in the Delta might look like in the future.  He also noted that the analysis was a ‘no action scenario,’ meaning that only the current levee elevations were used for the analysis. 

Over time, levees will compact and consolidate and fall a little bit, particularly in the Delta, because of the organic soils,” he said.  “There also is a lot of building maintenance and improvement projects that are going on in the Delta, and so those levees go up over time, too. So we decided to stay with what they are today, and we did not account for either of those factors because it’s very difficult to understand and know exactly how fast that compaction will occur and where.  It’s also difficult to know where the investment will occur and what will be done with the level of investment. So the DSC staff will walk that out during the adaptation strategy.”

Traditional approach

Traditionally, the way that a flooding analysis might be done is to take a scenario, a particular set of conditions, and say, ‘if we get this condition, this is what will be flooding.’ 

However, in the Delta, many different conditions could affect flooding:

  • The mean water level and sea level rise on top of that: how much sea level rise by 2050?  Will it be two inches or two feet?
  • There is a daily tidal cycle, as well as King Tides
  • Riverine inflow: Are we modeling a 100-year event? Is that a 100-year event on every single one of the tributaries? Or is it just one of the tributaries? And is it a 200-year event on the San Joaquin, or is it an 80-year event on the San Joaquin?

There’s just an infinite number of possibilities, and the co-occurrence of any individual possibility is infinitesimally small,” said Mr. Schwarz.  “So if you do it that way, you’re essentially going to be wrong about what you’re saying the future condition is.”

The decision-scaling approach

So rather than the traditional approach, the collaborators took a decision scaling approach, which is a strategy for decision making under deep uncertainty.

Mr. Schwarz explained the concept:  “Each one of these things has a distribution to it, and so we try to understand how the system responds to different forcings from each of these variables:  riverine, inflow, tides, and sea level rise.  Then we model the system repeatedly; it’s a stress test approach to see which ones lead us to bad results, like where the water level gets too high and goes over the top of the levee.  Then we can count the outcomes that occur.  If each of the distributions is correct, we will end up with a probabilistic understanding of how likely it is that we will see flooding in each of these places.”

As for the data used, he noted that the Kopp et al. 2014 paper is used for basically all sea level rise guidance in the state of California right now.  The watershed hydrology is from the Central Valley hydrologic study, which is consistent with the Central Valley Flood Protection Plan.  The tide and surge distribution come from the USGS Golden Gate gauge.

Mr. Schwarz acknowledged there are some drawbacks to this method:

  • First of all, it requires millions of runs to capture the co-occurrence of low-frequency events and to really understand the probabilities. However, once you have accomplished the million model runs, you have a very wide understanding of the approach.
  • The data need to be accurate; there shouldn’t be any bias in the model. However, he noted that it is less about the exact precision of whether the water level is 14.1 feet or 14.2 feet; as long as it’s relatively accurate, it will all wash out in the end because of the millions of model runs.
  • It requires good information on the probability distributions for model inputs, such as, is our assumption about sea level rise accurate?
  • There are really no existing analytical tools that can do this for the Delta, so they had to build a new model.
  • Some compromises had to be made due to complexity.

What we get out of this is, overall, a much better risk profile across all major modes of levee overtopping in the Delta; that is, we can understand all the different ways that flooding might occur driven from the San Joaquin River driven by the Sacramento River or sea level rise, or surge, or whatever it is, and in which area of the Delta,” he said. 

The information produced provides the decision-maker with a fully fleshed-out risk management set of data that they can use to test what different levee improvements might do to alter risk throughout the Delta. In addition, the approach is modular in that assumptions and data can be easily updated as more data is collected on climate change impacts.  He also noted it’s open-source and publicly available.

The diagram on the bottom left shows the modeling process for the study.  Mr. Schwarz noted that it’s a complicated process with lots of inputs and processing steps. There is also a tidal datum analysis that looks at what would be future mean high or high water or mean low or low water mean sea level in the future with sea level rise.

For the flood hazard scenarios, they looked at current conditions, 2030, 2050, and 2085 under a range of sea level rise conditions because these are distributions that are being sampled. 

This is the 95% confidence interval for each of these time periods,” Mr. Schwarz explained.  “The watershed hydrology is sampled from across the GCM range. And the historical hydrology scaled based on factors from the climate change projections to get at what will the future inflows look like to the Delta.”

He then presented a slide showing the flood risk exposure maps produced from the study, noting that it isn’t a flooded/not flooded scenario, but instead it’s probabilistic, meaning this area will be flooded X percent of the time, or have an X percent of chance of flooding in any individual year at this time period.

The shade of blue denotes the flood risk for that region, with the areas shown in dark blue having the highest with a 10% annual chance of flooding to the lightest blue, which has less than .5% annual chance of flooding, which is similar to a 200-year flood event. 

Under current conditions, the Suisun Marsh has the highest flood risk, and areas of the south Delta have some additional flood risk, but most of the Delta is in pretty good condition.  However, moving out to the mid and late century, there are very high levels of flooding across the Delta on very frequent recurrence intervals – so 10- and 50-year recurrence intervals will be all it takes to flood many of these islands throughout the South and Central Delta, he said.

The North Delta still looks pretty darn good, all the way out to the end of the century, even with no improvement in the levees, assuming they stay at their current level of elevations,” said Mr. Schwarz.  “That’s really attributable mostly to the Yolo Bypass and the investment that was made there in the flood system to give big flows a place to go during big events. And that just doesn’t exist on the San Joaquin system. And so those big flows come into the Delta and cause a lot of flooding throughout the Delta.

He then presented a figure showing the areas of strongest climate change influence throughout the Delta. 

This is one of the most informative graphics that we’ve been able to tease out of the modeling that we’ve done on water surface elevations and how things will change throughout the Delta as a function of climate change,” said Mr. Schwarz.  “What this is showing is that for the areas with the red dots, when flooding occurs, the high water levels are driven by the riverine system, whereas in the green areas, it’s driven by sea level rise.

What’s important here is is that this is telling us that our adaptation strategies are really going to have to be different for different parts of the Delta,” he continued.  “It is not one Delta; it is many different regions of the Delta, and they’re going to need different strategies. So strategies for the red dotted areas could be things that happen up in the upper watershed with reservoir reoperations or floodplain operations; that’s just not going to help as much for those areas of the green dots. And so we’re going to have to consider that.”

This slide shows an overall flood exposure infographic from the report that shows the value of exposed assets and the number of residents who live in highly socially vulnerable areas.

One of the key additions that Delta Adapts flood risk analysis really takes into account are the people who are living in highly socially vulnerable conditions,” he said.  “It’s a suite of different factors, but it could include not having transportation of your own, not speaking English, or having medical problems that would make it more difficult to be communicated with or rescued or rescue yourself during an event.

For more information:


QUESTION: Did the modeling consider the correlation between the impacts of river flow and sea level rise? Because both are greater in scenarios with higher carbon pathways.

Andrew Schwarz:  “The distributions are essentially independent of each other, except for the way that we sampled from mid-century for both, so you’re getting a sampling of climate change from consistent future periods.  The flooding event is a combination of what climate change is doing to intensify the hydrologic cycle and make it more likely to get a big event, and then we also have a very variable system in California. So even if we get a minimal amount of climate change, we can still get a very large flood event coming from the river system. So we could have a small amount of sea level rise, and it’s still a very large storm. But you’re right. If we get a very high GHG pathway, we would probably be more likely to get a worse result than what we’ve modeled. Although the river hydrology is all from the 8.5 pathway. So that’s more of a more pessimistic pathway.”

QUESTION: Are other federal, state, or local agencies expressing interest or incorporating this type of modeling into their existing models?

Andrew Schwarz:We’ve had different conversations with different parts of the federal government; obviously, the Corps of Engineers and FEMA are two that are most connected to this type of effort. The Army Corps of Engineers is starting to look at revising the 100-year flow estimates for the Delta, and we’ve had some conversations with them about taking a similar approach to this. And we did work with them on some of the high-end flow hydrology issues. So I think this is gaining traction, as FEMA and the federal Flood Insurance Program are looking at more probabilistic approaches to flooding and thinking about mitigating risks. So I think that there is growing interest in this type of approach.”

Print Friendly, PDF & Email
%d bloggers like this: