From Random Forests to Flood Forecasts: A Research to Operations Success Story — American Meteorological Society

Students in Sam Ng’s Field Observation of Severe Weather class hit the road every spring to observe storm structures, like this mesocyclone in Imperial, Nebraska. Photo by Sam Ng via Metropolitan State University of Denver

Click here to access the paper (Russ S. Schumacher, Aaron J. Hill, Mark Klein, James A. Nelson, Michael J. Erickson, Sarah M. Trojniak, and Gregory R. Herman). Here’s the abstract:

Excessive rainfall is difficult to forecast, and there is a need for tools to aid Weather Prediction Center (WPC) forecasters when generating Excessive Rainfall Outlooks (EROs), which are issued for the contiguous United States at lead times of 1–3 days. To address this need, a probabilistic forecast system for excessive rainfall, known as the Colorado State University-Machine Learning Probabilities (CSU-MLP) system, was developed based on ensemble reforecasts, precipitation observations, and machine learning algorithms, specifically random forests. The CSU-MLP forecasts were designed to emulate the EROs, with the goal being a tool that forecasters can use as a “first guess” in the ERO forecast process. Resulting from close collaboration between CSU and WPC and evaluation at the Flash Flood and Intense Rainfall experiment, iterative improvements were made to the forecast system and it was transitioned into operational use at WPC. Quantitative evaluation shows that the CSU-MLP forecasts are skillful and reliable, and they are now being used as a part of the WPC forecast process. This project represents an example of a successful research-to-operations transition, and highlights the potential for machine learning and other post-processing techniques to improve operational predictions.

*Corresponding author: Russ Schumacher,

#Paonia enters voluntary water restrictions — The Delta County Independent #GunnisonRiver #ColoradoRiver #COriver #aridification

Photo credit: Jonathan Thompson

From The Delta County Independent (Lisa Young):

The Town of Paonia entered into Stage One water voluntary restrictions at the recommendation of Mayor Mary Bachran during the May 11 meeting…

The mayor hit the highlights of Resolution 2020-17 including that stage one restrictions are voluntary; does not apply to drip systems and use of hand watering containers; reduction of irrigation — no irrigating when wind gusts or sustained winds, in order to reduce evaporation; outreach on water use and fixing leaks; limited gardening-car washing, pond and pool filling.

Town Administrator Corrine Ferguson said the town would go to Stage Two if and when water demands exceed supply and they are no longer spilling excess water at the treatment plant.

Stage two restrictions would limit watering on even and odd days with no watering on Saturdays while watering of the town park/properties would be limited to trees and planters.

Gunnison River Basin. By Shannon1 – Own work, CC BY-SA 4.0,

Why #Wildfire Burn Scars Become #FlashFlood Zones — And What To Do If You’re In One — #Colorado Public Radio

The Colorado River divides Glenwood Canyon slurry on the ridge from the Grizzly Creek Fire on Monday, August 24, 2020. (Kelsey Brunner/The Aspen Times via Aspen Journalism)

From Colorado Public Radio (Michael Elizabeth Sakas):

The burn scars around places like Glenwood Canyon, Estes Park and Grand Lake are now dangerous in a different way: Areas downhill and downstream from these burned regions are now highly susceptible to flash flooding.

“A third of an inch of rain in 15 minutes is all it’s going to take to start the low-end part of that flooding,” said Greg Hanson, a meteorologist with the National Weather Service in Boulder.

The Boulder office has already issued multiple flood advisories for burn areas. The Colorado Climate Center warned in a tweet that the threat of flash flooding in burn scars will be a recurring issue.

Hanson hopes there won’t be many flash flood warnings this summer, but he believes it’s inevitable…

Why are wildfire burn areas at a higher risk of flooding?

After a wildfire moves through an area, it burns up most of the plants and material that would absorb a lot of that rain, like the tree canopy and the leaves on the ground. Burned vegetation also coats soil with a wax substance that can cause soil to become hydrophobic, meaning it will repel water instead of absorbing it…

The first two years after a fire is when the worry is highest, Hanson said. But the flash flood risk often remains for much longer…

Glenwood Canyon/Colorado River. Photo credit: Allen Best/The Mountain Town News

One area to watch out for? Glenwood Canyon

The stretch of Interstate 70 through Glenwood Canyon was closed for more than a week in the aftermath of the 2020 Grizzly Creek fire. The burn scar that remains is vulnerable to a greater chance for flash floods and debris flows that rush down the canyon walls and affect the drivers and people recreating below.

CDOT is prepared for mudslides and rockfall in Glenwood Canyon, which are “very likely” if there is moderate to extreme rainfall, said agency spokesperson Elise Thatcher.

There’s a safety closure protocol in place to evacuate the canyon, Thatcher said. If there’s a certain amount of rain in the 24-hour forecast or a Flash Flood Watch, rest areas and recreation paths will be evacuated and closed.

CDOT will evacuate all traffic from the canyon if there is a Flash Flood Warning and stage crews to be on standby to clear the road of debris and assess the damage before reopening, Thatcher said.

Managing #ColoradoRiver risk: #ClimateChange is water change — Science Magazine #COriver #aridification

Eugene Clyde LaRue measuring the flow in Nankoweap Creek, 1923. Photo credit: USGS via Editorials360

From Science Magazine (John Fleck and Brad Udall):

In the 1920s, E. C. LaRue, a hydrologist at the United States Geological Survey, did an analysis of the Colorado River Basin that revealed the river could not reliably meet future water demands. No one heeded his warning. One hundred years later, water flow through the Colorado River is down by 20% and the basin’s Lake Powell and Lake Mead—the nation’s two largest reservoirs—are projected to be only 29% full by 2023. This river system, upon which 40 million North Americans in the United States and Mexico depend, is in trouble. But there is an opportunity to manage this crisis. Water allocation agreements from 2007 and 2019, designed to deal with a shrinking river, will be renegotiated over the next 4 years. Will decision-makers and politicians follow the science?

It has been said that climate change is water change. Globally, the effects on rivers vary widely, from increased risk of flooding in some places, to short-run increases in river flows in others as glaciers melt and catastrophes ensue once the glaciers are gone. The only constant is change, and our inability to rely on the way rivers used to flow. Like many snowmelt-fed rivers, for the Colorado this translates into less water for cities, farms, and the environment.

Research published over the past 5 years makes the threat clear. Run-off efficiency—the percentage of rain and snow that ends up as river water—is down, with half the decline since 2000 attributed to greenhouse-driven warming. For every 1°C of warming, researchers expect another 9% decline in the Colorado’s flow. This year’s snowpack was 80% of average but is delivering less than 30% of average river flows. Hot, dry summers bake soils, reducing flows the following year. The Colorado is not unusual. Researchers have identified similar patterns in other North American rivers, as well as in Europe, Asia, Africa, and Australia.

Colorado River water management has a long and uneasy relationship with science. LaRue’s analysis of the early 20th century was brushed aside in favor of larger, more aspirational estimates of the river’s flow made by bureaucrats who wanted to build dams. Scientists who agreed with LaRue—there were many—were ignored. This left the river overallocated and put the basin at risk.

Fortunately, there has since been progress in forging water management plans on the basis of science. For example, the US Bureau of Reclamation has been incorporating climate change into its analyses for more than a decade. Admirably, it overcame some of the political and technical challenges of incorporating the effects of climate change in the water allocation rules adopted in 2019. Models used to support decision-making were adapted to incorporate the 21st-century’s declining flows. Computer simulations showing emptying reservoirs were enough to convince decision-makers of the need to cut back. But have the modelers gone far enough?

The scientific challenges are formidable. Although the direction of change—a shift toward less river water—is clear, the details can be murky. This is a challenge for the handoff from science to the world of policy and politics. But we cannot allow that murkiness to stand in the way of taking seriously what the climate science is telling us.

As the basin’s water management community prepares for a new round of negotiations over the water allocation rules, how bad of a “worst case scenario” should be considered and who will get less water as a result? It is tempting to use today’s 20% flow decline as the new baseline—that is, modeling future reductions on the basis of what has already been observed. But only by planning for even greater declines can we manage the real economic, social, and environmental risks of running low on a critical resource upon which 40 million North Americans depend.

The United States and Mexico—not just America’s West and Southwest—can’t afford to get this wrong. There are still political challenges that harken back to the struggles of E. C. LaRue a century ago—namely, as political boosters chose overoptimistic estimates of the river’s flows to make their jobs easier. Climate science indicates that there will likely be less water in the Colorado River than many had hoped. This is inconvenient for 21st-century decision-makers, and overcoming their resistance may be the hardest challenge of all.

Colorado River Basin map via the Babbit Center for Land and Water Policy/Lincoln Institute of Land Policy

The river of lost flows: Is the #AnimasRiver drying up? A look back at the dwindling last century says yes — The #DurangoTelegraph #SanJuanRiver #ColoradoRiver #COriver #aridification

View of Denver and Rio Grande (Silverton Branch) Railroad tracks and the Animas River in San Juan County, Colorado; shows the Needle Mountains. Summer, 1911. Denver Public Library Special Collections

From The Durango Telegraph (John van Becay):

The American West is stricken by drought. That is not news. For the past 20 years, moisture levels have been abnormally low in the region. Shriveling crops, raging wild fires and diseased forests are just some of the fallout. Now scientists fear we may be on the verge of a megadrought – a long-lasting period of greatly reduced moisture. The last megadrought was in the 1500s and lasted some 50 years. Past megadroughts were naturally occurring. Their causes are complex and not completely understood. What scientists do understand is that the present drought has been worsened by anthropogenic (human caused) warming. That’s right, our old friend global warming has worsened our present drought by 30-50 percent. How bad could it get? The past 20 years contends with any 20-year period in the last 1,200 years for the severity of drought. That is some bad news.

The San Juan Basin drainage has been especially hard hit. The Animas River shows decreased flow not for the past 20 years, but for the past 30 years, since the early 1990s. In fact, discounting the abnormally wet 1980s, the Animas River has been steadily losing flow since the beginning of record keeping in the 1890s. The history of a single river graphed on a chart is notoriously hard to plumb for trends. Too many years buck the norm presenting a confusing welter of peaks and troughs. But dive deep enough and the evidence is there. The Animas River is not doing well.

Durango flood of 1911 river scene. Photo credit Center of Southwest Studies, Fort Lewis College.

On Oct. 5, 1911, after days of heavy rain, the Animas River came roaring out of its banks. It peaked at 25,000 cubic feet per second (cfs) which was 11 feet deep measured at the station near the power plant. The water reached the top of the arches on the Main Avenue Bridge. The monster runoff of 1911 set the record –since records have been kept – for peak run off of the Animas. Jump ahead to 1927 – a freakish combination of late snow melt and an early monsoon rain June 27, 1927, lashed the river to its second-highest historical level: 20,000 cfs and 9.65 feet…

Peak flow is flashy. It makes stunning photographs. It puts people on the river and pictures in the paper. But it’s a flighty thing, much like spring itself. Peak flow does not water crops or nurture cities. That measurement is annual discharge – which is measured in thousand acre feet, or k/a/f…from 1913-25, the average annual discharge was 750 k/a/f, that’s three-quarters of a million acre feet per year.

Our second period of increased annual discharge is 1941-1950 when the average annual discharge was 665 k/a/f. In our third period, 1979-88, average annual discharge ticked upward to 700 k/a/f, a period that has not been matched for the last 80 years.

For comparison, from 2010-19 the average annual flow of the Animas dropped to 496 k/a/f, a staggering 34 percent decline from the average flow a century ago. If scientists are correct – and we are entering another megadrought, aided and abetted by global warming – the future of the Animas appears grim.

In the past Durango survived the loss of its smelter. A present controversy over the coal-fired train continues to tear at the town. What is Durango’s future with an Animas River reduced to the size and ferocity of an irrigation ditch? How does that impact Durango’s self-image? Where then will lie the soul of el Rio de las Animas Perdidas? Along a dry river bed?

San Juan River Basin. Graphic credit Wikipedia.

Water Leadership program builds skills, confidence to tackle escalating #water challenges — @WaterEdCO

Water Education Colorado’s 2021 Water Leaders May 2021.

Here’s the release from Water Education Colorado (Jayla Poppleton):

Water Education Colorado kicks off its 15th Water Leaders Program May 27, 2021].

Sixteen up-and-coming water leaders from a diverse range of communities and water sectors across Colorado have been selected to participate in this intensive personal and professional development opportunity. They come from both private and public sector organizations, from state agencies, water districts and nongovernmental organizations.

“This is an invaluable investment that we are making, that our participants are making, to ready themselves for the incredible challenges that we as a state are facing around water,” said Jayla Poppleton, executive director for Water Education Colorado. “We are equipping leaders with the confidence and skills to effect change, to work collaboratively across interest areas, and to feel rewarded in what they do as they lead their teams to innovate and craft water solutions.”

Established in 2006, the Water Leaders Program has produced nearly 200 graduates.

Several notable alumni include current Colorado Ag Commissioner Kate Greenberg, Esther Vincent, director of environmental services at Northern Water, and Matt Lindburg, managing principal for Brown and Caldwell who is supporting the state on its 2022 update to the Colorado Water Plan.

Participants are selected based on proven commitment to Colorado water and demonstrated potential for increased leadership roles. Many class members are civically active, serving in a wide range of volunteer roles such as on boards and commissions, in addition to their day jobs.

Molson Coors is a title sponsor of the 2021 program. The company has been brewing beer using Colorado water since 1873 and invests in variety of water sustainability initiatives, ranging from improved efficiency to protection of source watersheds.

“Molson Coors is proud to sponsor the Water Leaders program and ensure a bright future for Colorado by supporting our future water leaders,” said Kayla Garcia, community affairs manager for the company.

Throughout the four-month program, which culminates with a graduation ceremony on Sept. 24, the group will undergo a variety of self-assessments as well as an external, 360-degree feedback review from peers, supervisors, and direct reports. Facilitators will challenge participants to be vulnerable with their hopes and aspirations as well as their fears and perceived limitations as they confront complex issues for Colorado water management and protection in the face of climate change, population growth, and widely diverse community values around resource protection and use across Colorado.

“We have seen individuals break out of their shell and find their calling in the water community in ways that they never even thought possible,” said Stephanie Scott, leadership programs manager for Water Education Colorado. “Seeing graduates stretch beyond their wildest dreams, both personally and professionally, is the true magic of the Water Leaders Program.”

For more information about the program visit:

About Water Education Colorado

Water Education Colorado is a 501c3 nonprofit providing policy-neutral news and informational resources, engaging learning experiences, and empowering leadership programs. We work statewide to ensure Coloradans are knowledgeable about key water issues and equipped to make smart decisions for a sustainable water future.

WEco offers a variety of digital content and also produces two print publications: Headwaters magazine and the Citizen’s Guide reference series. These publications are distributed to policy makers, water professionals, agricultural and environmental organizations, university students, business leaders, and community groups.

In addition to the Water Leaders Program, WEco runs two other leadership programs: 1) Water Fluency – a comprehensive water literacy course for decision makers without a professional water background, and 2) the Water Educator Network – an affiliate program for fellow water educators to improve their practice.

WEco also provides a variety of other educational and outreach opportunities, including tours, forums, workshops, and the annual Sustaining Colorado Watersheds conference, held in October each year in Avon, Colo.

In June 2018, WEco launched Fresh Water News, a nonprofit news initiative dedicated to providing nonpartisan news coverage of the water issues that define Colorado and the American West.

Double-dipping: Why does #LaNiña often occur in consecutive winters? — NOAA #ENSO

From NOAA (Nat Johnson):

As the 2020–21 La Niña has come to an end, leaving us with neutral conditions in the tropical Pacific, we now wonder if we have seen the last of La Niña for a while or if we will see another dip into La Niña conditions by next fall. In the world of the El Niño-Southern Oscillation (ENSO), double-dipping is not a party foul—it’s actually quite common for La Niña to occur in consecutive winters (not El Niño, though). If you’re wondering why, then this is the blog post for you!

Mirror, mirror on the wall
To understand why La Niña commonly double dips, we first need some basic understanding of ENSO asymmetry. Often, we think of El Niño and La Niña as mirror opposites—for example, that the warmer-than-average east-central tropical Pacific conditions during El Niño are exactly matched by the cooler-than-average conditions of La Niña. This mirror-opposites perspective is a pretty good approximation, but it’s not perfect!

Average November – January sea surface temperature anomalies (°C) for the top 10 strongest (top) El Niño and (bottom) La Niña events since 1950 based on the November – January Niño3.4 index. Anomalies are calculated with respect to 30-year base periods updated every 5 years. NOAA figure with ERSSTv5 data obtained from the NOAA Physical Sciences Laboratory.

Upon closer inspection, we see some subtle but important differences between El Niño and La Niña in terms of their patterns and behavior. First, the sea surface temperature anomalies (difference from the long-term average) during El Niño tend to be centered farther east than for La Niña, especially for the stronger El Niño episodes. Second, the strongest El Niños tend to be stronger than the strongest La Niñas. We can see these differences in the maps above. The warm anomalies in the eastern Pacific for the El Niño composite are between 2.25 and 2.75 degrees Celsius (4 and 5 degrees Fahrenheit), while the cool anomalies in the east-central Pacific for the La Niña composite top out at 2.25 degrees C.

Monthly sea surface temperature anomalies in the Niño 3.4 region of the tropical Pacific for (top) all first-year El Niño and (bottom) first-year La Niña winters since 1950, starting in the March before the first-year peak. graph based on ERSSTv5 temperature data.

We also see notable differences in how sea surface temperatures in the Niño 3.4 region of the Pacific change over time for first-year El Niños (meaning, the previous winter did not feature El Niño) and first-year La Niñas. After the peak of El Niño, which typically occurs in late fall or winter, Niño3.4 surface temperatures usually decline rapidly to neutral conditions by the following spring. In the ensuing summer through winter, ENSO neutral conditions continue, or La Nina develops, but the occurrence of El Niño in a second consecutive winter is uncommon.

For La Niña, in contrast, the return to ENSO neutral after the late fall peak is usually more gradual, and, as Emily recently reminded us, the transition to El Niño after the first winter of La Niña rarely occurs. Instead, we generally see ENSO-neutral or a transition back into La Niña conditions by the following fall, as shown by that second dip in average Niño3.4 temperatures in the plot above. To reiterate what Emily already noted, of the 12 first-year La Niña events on record, 8 were followed by La Niña the next winter, 2 by neutral, and 2 by El Niño.

All things being unequal
Things would be simpler if El Niño and La Niña behaved similarly, so why are the transitions out of El Niño and La Niña so different? The answers lie in the processes that cause El Niño and La Niña to strengthen and weaken. Several months ago, Michelle introduced us to the essential processes for ENSO, the fun-to-say Bjerknes feedbacks. Essentially, Bjerknes feedbacks describe the reinforcing interactions (positive feedbacks) between the ocean and atmosphere that cause El Niño and La Niña to grow: changes in the tropical Pacific ocean temperatures cause changes in the overlying trade winds, which then cause additional, reinforcing changes in the ocean temperature (please see Michelle’s post for more of the details!).

May–September monthly zonal (east-west direction) wind stress anomalies against the monthly sea surface temperature anomalies in the Niño 3.4 region from 1980-2016. The wind stress anomalies corresponding to La Niña conditions (blue dots) are averaged over the central tropical Pacific from 5°S – 5°N and 167°E – 153°W. The wind stress anomalies corresponding to El Niño conditions (red dots) are averaged farther east, 5°S – 5°N and 179° – 139°W. These regions have the strongest relationship with the Niño 3.4 index and are chosen following Choi et al. (2013). The steeper slope of the linear fit to the red dots relative to the linear fit to the blue dots indicates a stronger zonal wind stress response under El Niño conditions. NOAA figure based on MERRA Reanalysis wind stress data and ERRSSTv5 sea surface temperature data.

If all essential processes that comprise this feedback process were equal but opposite for El Niño and La Niña, then we might expect mirror opposite patterns. However, this isn’t the case. For example, the response of the trade winds to the Niño3.4 surface temperatures is unequal between warmer and cooler surface conditions. We see this above in the scatter plot of monthly May–September zonal (east-west direction) wind stress anomalies (1) in the equatorial Pacific versus the Niño3.4 surface temperature anomalies. As we expect, warmer Niño3.4 surface temperatures that are tied to El Niño conditions bring weaker trade winds and a weaker Walker circulation (positive wind stress anomalies), whereas cooler Niño3.4 conditions bring stronger trade winds (negative wind stress anomalies) connected with a stronger Walker circulation.

However, the plot also reveals that the relationship between the warm (El Niño) and cool (La Niña) sea surface temperature anomalies and the wind anomalies they generate is not equal. El Niño’s warm water anomalies generally produce wind stress anomalies that are stronger and farther east than those produced by cool La Niña anomalies of equal strength. The stronger response is demonstrated by the steeper slope of the red line versus the blue line. This difference may seem a bit subtle, but it can have big consequences for ENSO asymmetry (2).

The strength of the coupling between the winds and upper ocean matters because it not only strengthens El Niño and La Niña, but it also sets the wheels in motion for the ultimate demise of each event. The strong coupling between ocean and atmosphere and the more eastward wind stress anomalies during El Niño contributes to a robust poleward discharge of equatorial Pacific upper-ocean heat, typically by early spring, that ends the El Niño event and sets the stage for the next La Niña. The weaker coupling and more westward wind stress anomalies during La Niña means that the corresponding “recharge” of heat is also weaker, and so the ocean is not as primed for a transition to El Niño. Instead, if we experience the right sequence of tropical weather, a second winter of La Niña may return instead (3).

The scenario I have described is far from complete, as there are many other atmospheric and oceanic (and even biological!) processes that likely contribute to ENSO asymmetry and the tendency for La Niña to persist for more than one winter (4). There is still considerable debate about which processes are most important for these El Niño/La Niña differences (still so much unsettled ENSO science!).

A heads up?
Can we predict when La Niña will double-dip? Some recent studies suggest that we may be able to predict (in a probabilistic sense) more than a year in advance the likelihood of two-year La Niña based on the strength of the preceding El Niño and poleward discharge of equatorial heat (5). However, the most recent La Niña is a bit unusual because El Niño did not immediately precede it, and so it is difficult to identify any clear indicators that La Niña will return next fall at this time. Although we must await further guidance to get a better handle on the forecast, we at least can say that both history and the current ENSO forecast suggest that El Niño is unlikely to return in the near future.

Special thanks to Dr. Andrew Wittenberg for guidance and helpful comments while preparing this blog post!

(1) Wind stress measures the lateral force per unit area that the wind exerts on the ocean surface. A “positive zonal wind stress anomaly” corresponds to an anomalous eastward force on the ocean surface, exerted by a stronger-than-average eastward component of the winds; this weakens the normally westward force exerted by the trade winds. Conversely, negative zonal wind stress anomalies indicate a stronger westward force on the surface ocean, exerted by stronger-than-average westward trade winds.

(2) The zonal wind stress analysis of this post is modeled after Choi et al. (2013). That study provides conceptual and theoretical support for the hypothesis that the asymmetry in zonal wind stress response to tropical Pacific sea surface temperatures, like what is shown in the scatter plot of this post, is sufficient for explaining many of the asymmetries between El Niño and La Niña, including the tendency for La Niña to persist for more than one winter (see the next footnote for additional details).

Choi, K.-Y., G. A. Vecchi, and A. T. Wittenberg, 2013: ENSO transition, duration, and amplitude asymmetries: Role of the nonlinear wind stress coupling in a conceptual model. Journal of Climate, 36, 9462-9476.

(3) For those of you who are interested in more of the scientific details about how this wind stress asymmetry between El Niño and La Niña may contribute to the tendency for La Niña to double-dip, this footnote provides additional, more technical discussion, courtesy of Dr. Andrew Wittenberg.

ENSO events are sparked by a disequilibrium between two coupled ocean/atmosphere time scales in the response to wind changes: a fast equatorial adjustment (oceanic equatorial Kelvin waves + Bjerknes feedbacks), and a slow off-equatorial adjustment (curl-induced oceanic Rossby waves and delayed negative feedbacks).

Equatorial wind anomalies that are located farther east (as during a strong El Niño) induce more transient growth, disequilibrium, and overshoot into the opposite phase, for 2 key reasons. (1) Their resulting zonal current anomalies & upwelling anomalies are located closer to the warm pool edge in the central Pacific and shallow thermocline of the east Pacific. This strengthens the transient growth via fast equatorial processes (zonal advective feedback and Ekman feedback, two of the important Bjerknes feedbacks), which amps the wind anomalies so much that they can over-discharge the equatorial thermocline several months later. (2) When the off-equatorial wind stress curl anomalies are located farther east, they’re also farther from the western boundary, and so their resulting off-equatorial oceanic Rossby wave trains have farther/longer to travel before they can reflect back onto the equator and start the turnabout of the ENSO event. This enables the equatorial disequilibrium to grow more strongly before being impeded & eventually reversed by the delayed negative feedback. Thus, the more eastward-shifted zonal wind stress anomalies during strong El Niños enable stronger transient growth & disequilibrium, increasing the over-discharging and subsequent overshoot.

In short, under the “zonal wind stress asymmetry hypothesis” of Choi et al. (2013), La Niña is more likely to double-dip because its winds stress anomaly is so far from the “sea surface temperature-ticklish” zone of the central/eastern Pacific, and so close to the western boundary “transfer station” for the off-equatorial feedbacks, that La Niña just can’t get as much of a disequilibrium going – and hence can’t over-charge the equator enough to guarantee an overshoot into El Niño.

(4) For a comprehensive review of existing hypotheses for ENSO asymmetry, I recommend An et al. (2020).

An, S.-I., E. Tziperman, Y. Okumura, and T. Li, 2020: ENSO irregularity and asymmetry. In A. Santoso, M. McPhaden & W. Cai (Eds.), El Niño Southern Oscillation in a changing climate (pp. 153– 172). John Wiley & Sons.

(5) For example, a few recent studies, including two led by Dr. Pedro DiNezio, suggest that the probability of a multi-year La Niña may be skillfully predicted 18-24 months in advance, given a preceding El Niño. The sources of skill are rooted in the strength of the El Niño and the magnitude of poleward heat discharge 6 months after the peak of El Niño.

DiNezio, P. N., C. Deser, Y. M. Okumura, and A. Karspeck, 2017a: Predictability of 2-year La Niña events in a coupled general circulation model. Climate Dyn., 49, 4237–4261,

DiNezio, P. N., C. Deser, A. Karspeck, S. Yeager, Y. Okumura, G. Danabasoglu, N. Rosenbloom, J. Caron, and G. A. Meehl, 2017b: A 2 year forecast for a 60–80% chance of La Niña in 2017–2018. Geophys. Res. Lett., 44, 11 624–11 635,

Wu, X., Y. K. Okumura, C. Deser, and P. N. DiNezio, 2021: Two-year dynamical predictions of ENSO event duration during 1954–2015. J. Climate, 34, 4069-4087,