BAY-DELTA SCIENCE CONFERENCE: Exports & Outflows: Recalibrating Water Year Types in a Changing Climate

The Sacramento and San Joaquin water-year type indices are fundamental to the State Water Project and Central Valley Project operations as every year, how much water is allocated to whom depends on whether the hydrology is expected to be wet, above normal, below normal, dry, or critically dry.  The water-year classifications were last revised in 1991 and were indexed to historical climatic conditions; however, as California’s precipitation patterns and rain/snow mix continue to change, this framework may need to be revised.

At the 2021 Bay-Delta Science Conference, Wyatt Arnold, an engineer with the Department of Water Resources Climate Change Program, discussed a study he did with colleague Romain Maendly looking at how water year types could potentially be recalibrated and the implications of that for Delta water management.

The study had two objectives:

• To model the effects of revising the Sacramento and San Joaquin Water Year Type definitions to maintain the historical frequency of occurrence under rising temperatures and quantify the effect on Delta water management objectives, such as exports, salinity, and outflows.
• And then quantify the interplay between rising temperatures and sea level rise on Delta water management in connection with those water year type changes.

The study area is delineated by the runoff systems for the State Water Project and the Central Valley Project that convey water from Northern California, across the Delta, and down to the Central Valley and southward to urban areas in Southern California.  The Sacramento River provides three-fifths of the unimpaired inflow volume to the Delta; the San Joaquin River provides one-fifth. 

The slide shows the modeling workflow for the study.  Basically, the observed record was combined with the reconstructed runoff records from paleo tree ring records going back over a thousand years (900-2011), creating a much longer record of long-term inter-annual variability.  That dataset was then used to simulate streamflow and other information under incrementally warmer temperatures; CalLite 3.0 then simulated reservoir and conveyance operations for the State Water Project and the Central Valley Project under different scenarios. The model outputs included reservoir storage, State Water Project deliveries, Delta exports, and others.  Iterations of the model become a water system stress test under different incremental changes of climate.

A brief history of the ‘water year type’

In 1978, the State Water Board’s water right Decision 1485 first linked fish, agricultural, M&I, flow, and water quality objectives to the Sacramento Valley index.  

Those were revised about a decade later by a multi-agency workgroup to better represent water availability for three main objectives:

• the water year type determined in the spring matched as close as possible to the water year type that was ultimately realized by the end of the water year;
• it accounted for system storage and flood constraints,
• it accounted for the seasonal importance of runoff during the year.

The revised water year construction was adopted into Bay-Delta Plans beginning in 1991. It subsequently became part of Water Right Decision 1641, as well as some of the triggers in the biological opinions.

How the water year type index is constructed

The slide shows the equation used to calculate the water year index for the Sacramento Valley (SVI) and the San Joaquin Index (SJI).  X and Y are seasonal runoff components, and Z is the previous year’s index.  A seasonal coefficient of 40% (April through July) /30% (October through March) /30% (remainder) for the Sacramento Valley is applied to the output; for the San Joaquin, those coefficients are 60/20/20 using the same time periods.

The index volume then becomes a threshold that is the dividing line between the water year types.  There are five for the Sacramento and San Joaquin:  critically dry, dry, below normal, above normal, and wet.

The effect of warming temperatures on the hydrograph

The slide shows the seasonal monthly hydrograph for the Sacramento and the San Joaquin simulations going back 1100 years.  

“What you can see is that with every incremental rise in temperature, the hydrograph shifts earlier in the year,” said Mr. Arnold.  “That’s the runoff centroid shifting from the March-April period earlier into the winter. There’s also an average water year runoff decline; it’s about 1 to 2% decline per degree for the Sacramento and about a 2% decline in the average annual runoff for the San Joaquin.”

The coefficients play a role here by amplifying this decline.  “For example, the San Joaquin temperature is increased by two degrees,” he explained.  “What we see from the hydrological model results is that there’s a 4% average annual runoff decline.  According to how the index is calculated, that same two-degree temperature rise results in a 12% average annual index decline.  The reason is that as the runoff shifts earlier into the October through March period, from April through July, it’s now getting only a 20% weight, whereas previously, it had a 60% weight in terms of its seasonal coefficient.”

The slide shows the cumulative distribution of the water index calculated for every year in the 1100 year record; the shaded bands are 1001 windows, each a length of 100 years.  The red is the same trace but with a three-degree temperature rise, which illustrates the variance and the natural internal variability in the index.

“For example, the above normal to wet threshold of 9.2 can correspond anywhere to 42% of years are below that threshold, all the way up to 75% of years are below that threshold, depending on which 100 year period you’re looking at in the Paleo record,” said Mr. Arnold.  “With temperature rise, basically the overall response is for the index to drop across these thresholds. And so for example, the central tendency, which is the cumulative distribution – if you take all that 100 years together at the same time, you see that the percentile which corresponds to that threshold goes from 10% under the Paleo historical to 15%, with a three-degree temperature rise.  So there is a higher frequency of critical and below normal years, lower frequency of wet and above normal years under rising temperatures.”

The effect of recalibrating the water year types

So what if we tried to maintain the frequency distribution of the water year types that we’ve been managing to over the past several decades?  One way to do that is to drop the thresholds to maintain the percentiles based on the observed record.

“As the water year type frequencies change, that will also change what regulatory actions are being triggered according to the water type that’s been designated,” said Mr. Arnold.

The slide on the upper left shows the recalibrated water year type thresholds that maintain historical water year type distribution under changes in average annual temperature.  The slide on the upper right shows which regulatory action in the Delta is tied to water year type; each also has a time window during which they are imposed on the system.

“So when we’re modifying the water year type of frequency, we’re looking to see how these regulatory actions change flows through the Delta and exports south of the Delta,” said Mr. Arnold.

They did two CalLite runs, with and without the water year type adjustment, for each of the incremental temperature changes, as well as model runs with no sea level rise and 60 centimeters of sea level rise.

On the chart, ‘base’ is the change in performance under climate change with no adjustment to the water year type thresholds; ‘ΔAdap’ is the change in performance relative to the base with an adjustment to water year type thresholds, or the additional change on top of the base change under a given climate change condition.

There are two axes of perturbation:  Temperature rise of +1 to +4 degrees, and sea level rise at 60cm.  On the plot, red indicates a performance decline in that metric; blue indicates performance improvement.  The lower-left corner of the response plot represents the historical statistic for that metric. The contour lines correspond to the degree of change that’s interpolated across the response to different climates.

“Basically, we found that there’s a trade-off about 10,000 acre-feet of average annual flow that either contributes to South of Delta exports, or to an increase net Delta outflow, and that’s associated with whether you modify those historical water year type thresholds or not,” said Mr. Arnold.  “So that 10,000 acre-feet is the approximate volume that’s associated with changing the thresholds.”

As for the additional 10,000 acre-feet of average annual flow, the CalLite model showed that, on an average annual basis for increased salinity at Rock Slough, there is about a 30 to 40% mitigation of the increased salinity due to climate change.

“The increased salinity is still there, but there’s some sort of reduction or mitigation of that increased salinity,” he said.  “This also translates to things like X2, which is expected, and especially in years following above normal and wet years, which are crucial for a very specific biological opinion trigger to increase fall flows for fish. So the increase in about a kilometer per year of X2 under those specific year types can actually be reduced by about another half to .7 kilometers. So there’s a much higher mitigation of the increased X2 by just modifying the water year type thresholds.”

In conclusion …

Mr. Arnold then gave his takeaways:

Rising temperatures cause the water year centroid to shift and annual runoff to decline, which ultimately leads to fewer ‘wet’ years and more ‘critical’ years for the Sacramento Valley Index, and much more so for the San Joaquin Index, even when total annual runoff has not changed as severely as the index itself

After modifying the Sacramento Valley Index and the San Joaquin Index water year type thresholds to maintain historical frequency, the CalLite model suggests a 10 TAF / °C trade-off in average annual Delta Exports versus Net Delta Outflow which carries across sea level rise up to 60cm.

The long-term average response to this trade-off of key Delta water management metrics tied to WYT include:

–A <10% mitigation of X2 increases annually (which diminishes with sea level rise)

–A 30% mitigation of Fall X2 increases after wetter years (which diminishes slightly with sea level rise)

QUESTIONS & ANSWERS

QUESTION: What performance metric were you using to compare the base and the ΔAdap plots?

“Basically, we looked at South of Delta exports on an average annual basis, so that’s both SWP-CVP together; so that was one performance metric,” said Mr. Arnold.  “The second one is net Delta outflow. From that, we’ve identified a volume trade-off of 10,000 acre-feet. Now, as far as salinity, we looked at Rock Slough station, and we looked at X2, and then compared the change in performance of those two variables under each of those scenario constructions.”

QUESTION: What was the most surprising thing from your results?

“That sea level rise can actually tend to improve certain things based on the management of the system,” said Mr. Arnold.  “Because there’s an increased delivery of flows to maintain salinity, you get less of a benefit from changing the water year type thresholds than you might imagine because that sea level rise is already driving the necessary increase of flows, regardless of whether the water year type is dry or wet, or whatever. So that was a bit surprising.”

“The other is  (although this is not shown, because it’s just too much information to handle), the changes in precipitation. And whether you’re on the upper envelope of that 100-year shaded band that I showed for the cumulative frequency or in the lower window of that band can have a lot more influence on the changes in the Delta than water year type threshold changes themselves. So the variance in the water year types, and their distribution, just according to natural variability, is so wide that these temperature increases, while they’re substantial for influencing the San Joaquin water year type because it’s very dominated by the Sacramento, those changes are kind of muted in that band of natural variability.’

QUESTION: If Water Fix were fully implemented, how would you foresee that impacting the water year thresholds?

“You’d have to run the model with some sort of tunnel volume and tunnel operation scheme. We didn’t do that, so I really have no clue. And as far as I’m aware, there’s no connection between what is going into the tunnel environmental impact report with changes in water year type and the thresholds. This is just an exploratory effort that it really has nothing to do with the tunnels assessment.”