Water whiplash in California was predicted in a 2012 study by Sarah Null and Josh Viers, funded by the California Energy Commission’s PIER program. The study shows the importance of independent research.
At the time of Null and Viers’ CEC study, the Department of Water Resources’ climate change simulation methods used Global Circulation Model (GCM) outputs to modify the historical record of wet & dry years. But DWR’s water resources engineers were concerned that this could miss shifts in the distribution of wet and dry years. In Using Future Climate Projections to Support Water Resources Decision Making in California, DWR’s water resources engineers noted:
In water resources planning, it is often assumed that future hydrologic variability will be similar to historical variability, which is an assumption of a statistically stationary hydrology. This assumption no longer holds true under climate change where the hydrological variability is non‐stationary. Recent scientific research indicates that future hydrologic patterns are likely to be significantly different from historical patterns, which is also described as an assumption of a statistically non‐stationary hydrology. In an article in Science, Milly et al. (2008) stated that “Stationarity is dead” and that “finding a suitable successor is crucial for human adaptation to changing climate.”
In contrast to DWR, Null and Viers’ study did not use the historical record. Instead they used GCM outputs that had been fed into a Variable Infiltration Capacity model to get streamflow. They did statistical tests to check that the outputs for 1951-2000 were consistent with the historical record. They used two emissions scenarios, A2 (severe climate change), and B1 (moderate) for the study.
When the inputs were used directly, all of the GCM models projected a significant increase in dry and critically dry years by the latter half of the century, with a corresponding decrease in wet and above normal years. (see chart below.) The chart below shows the results of the study.
SVI stands for Sacramento Valley Index, is a measure of the runoff to the Sacramento Basin from the Sacramento, Feather, Yuba, and American rivers. Table 6 shows an increase of 9.7% in critically dry years, and 11.7% in dry years under the A2 scenario.
SJI stands for San Joaquin Valley Index. It is a measure of the runoff to the San Joaquin Basin from the San Joaquin, Stanislaus, Merced, and Tuolumne rivers. Table 6 shows an increase of 34.9% in critically dry years, and a decrease of 4.8% in dry years under the A2 scenario.
Note that the changes that were projected as happening after 2050 are happening now, and the water whiplash may be even more severe than these early models predicted.
Good post, we all need to be thinking about this.
I am not trying to claim precedence for this (because I don’t see any place where I called it out in the text), but based on your post, I went back and had a look at a paper we published way back in 2004 (all but 20 yrs ago now), to see whether we’d noticed this whiplash tendency back then. This 2004 paper was really the first time that some of us got our hands on daily-level output from a respectable (then) climate model for use as inputs for hydrologic simulations in the Sierra.
https://c5424e72-0cac-4aad-87be-4f75d371c0c6.filesusr.com/ugd/3b5c57_8ffdc64e1d184ae989dbcf7078685991.pdf
In that paper, we DID call out the fact that nearly all of the change in future precipitation was concentrated in the largest storms, a finding that have only expanded through time as we’ve gotten our hands on more and more climate-model outputs etc (fig. 2). But in terms of annual-level whiplash, although its pretty clear that it was already there in projections and simulations from fully 20 yrs ago (figs. 3 and 5), we didn’t acknowledge it or (maybe) trust it so much back then. But in hindsight this is clearly one of those things that have been in the projections all along. It just took the community a while to recognize it (not the only such thing, in hind sight). These patterns that have been in nearly all of the models and projections, pretty much forever, are some of the things that I am inclined to have most confidence in now. Overall those years, the models and methods have evolved drastically, so the findings that have hung in there all along generally reflect the most fundamental properties of the climate and hydroclimatic systems and demonstrably least model-dependent..thus most to be trusted.
Kudos to Null and Viers for catching this so long ago, and to Swain for identifying and illustrating so clearly now!
Thanks so much for making this comment. I have blog post discussing the process of reaching scientific confluence. Bradshaw et. al. said “[s]cientific confluence is reached by curiosity, rigorous testing of assumptions, and search for contradictions, leading to many — sometimes counter-intuitive or even conflicting — insights about how the world works.” I think this is important to unpack and think about. https://cah2oresearch.com/2021/09/20/on-scientific-confluence-and-the-challenges-of-avoiding-a-ghastly-future-in-california-water/