NOAA ENSO December 2024 Update: Weak La Niña Forecast Packs Surprising Punch

This post first appeared on the climate.gov ENSO blog and was written by Nat Johnson

Although it’s taking its sweet old time to get going, a weak La Niña, the cool phase of the El Niño/Southern Oscillation (ENSO), is expected to develop this winter. Much like I did last November, I thought it would be a good time to delve into how these expected tropical Pacific conditions might influence the pattern of precipitation over North America this winter. After all, ENSO—the entire El Niño and La Niña system—is the single most predictable influence on year-to-year differences in average winter precipitation across the United States. Without further ado, let’s get to it!

• Related: [UPDATED: NOVEMBER 22] NOAA Winter 2024-25 Forecast: La Niña Returns and What That Means For the Ski Season

Break out the models!

Last November we were anticipating a big El Niño to dominate the winter precipitation forecast. It’s a much different story this year, as ENSO has been transitioning to the opposite, cool phase. Moreover, this winter’s developing La Niña is expected to be quite weak, so we might also expect less of an ENSO influence on the winter precipitation pattern. Do our state-of-the-art seasonal forecast models agree? Let’s turn to the North American Multi-Model Ensemble (NMME) for guidance. The map below shows the December-February average of all precipitation anomaly forecasts from all available NMME Models (1). (Anomaly = deviation from the 1991-2020 average, with green shading showing wetter conditions and brown shading showing drier conditions.)

The NMME forecast for this upcoming winter does, in fact, resemble the traditional La Niña precipitation pattern quite closely. In particular, we see the classic reduction in precipitation throughout the southern U.S. and Mexico with enhanced precipitation in the Pacific Northwest and Ohio and Upper Mississippi Valley regions, a pattern that also resembles NOAA’s U.S. Winter 2024/25 Precipitation Outlook issued by the Climate Prediction Center. This pattern is nearly a mirror opposite to the forecast pattern I showed last November—not a big surprise, given that the forecast has shifted from El Niño last year to a potential La Niña this year.

Upon closer inspection…

Although the forecasted precipitation pattern doesn’t raise any eyebrows, a rather surprising feature is just how strong the forecasted precipitation anomaly is, given that we are only expecting a marginal La Niña, at best (there are still 1 in 4 odds that this winter will remain ENSO-neutral). Eagle-eyed readers might even notice that the precipitation anomalies in this winter’s NMME forecast generally look stronger than last year’s, when we were entering one of the strongest El Niños of the past 40 years! That’s pretty unexpected.

To ensure that our lyin’ eyes aren’t deceiving us, we can add some quantitative muscle to this observation that something unusual is going on this year. First, I created a time series (index) that describes the strength of the NMME’s forecasted winter precipitation anomaly (2). Then I compared it to the forecasted ENSO strength for that winter. The strength of the forecasted precipitation anomaly generally tracks the strength of the ENSO forecast very closely throughout the NMME record (3), but this relationship breaks down this winter. In fact, by this measure, this winter’s precipitation forecast is the 6th strongest predicted anomaly in the 34-year NMME record despite an ENSO forecast of only near-average strength (ranked 18^th out of 34) (4).

So, the data backs up the eye test. It is awfully weird that we would see such a strong La Niña-like precipitation pattern given the weak forecasted La Niña.

It’s all relative

Well then, what is so special about this year that’s boosting the model winter precipitation forecast? One possible culprit that we have raised repeatedly, including in Emily’s latest update, is all the warmth in the tropical oceans outside of the central-eastern Pacific. That means that the current ocean temperatures in the Niño-3.4 region (the main area for monitoring the El Niño/Southern Oscillation) may not be that cool in relation to the historical record of Niño-3.4 region temperatures, but they are quite cool relative to the rest of the global tropics at the present time.

It turns out that this relative temperature in central-eastern Pacific may be more important than the actual temperature for triggering a response in the tropical precipitation and winds that lead to the cascading global impacts that follow. Fortunately, as Michelle eloquently described a few years ago, there is a simple and elegant way to modify the traditional Niño-3.4 index to account for this tropics-wide warming: calculating the local anomalies but then also subtracting the average sea surface temperature of the entire tropics. This subtraction creates a relative Niño-3.4 index (or Relative Oceanic Niño Index, RONI, for seasonal averages) (5).

Because the global oceans have been warming continually, we have been seeing a growing difference between the traditional and relative Niño-3.4 index. If we focus specifically on the NMME forecasts (6), we see that winter of 2024/25 has the largest forecast difference between the two indexes, with the traditional index nearly 0.5° C (0.9° F) warmer than the relative index.

But what does this mean for the precipitation forecast?

Now let’s see what all this could mean for North American precipitation. Below, I compare the actual NMME winter precipitation forecast with what we would expect just based on the historical relationships with the forecast Niño-3.4 index, both traditional and relative. The precipitation forecasts in the two right panels are “reconstructed” from the Niño-3.4 index forecasts and the precipitation patterns that the model historically predicted when those Niño-3.4 index forecasts were made (7). These reconstructions match the pattern of the actual NMME forecast (left panel) quite well, confirming that ENSO, once again, is a dominant influence on the winter precipitation forecast in the models. This is unsurprising.

The surprising part is that we also notice important differences in the strength of the patterns – the precipitation forecast reconstructed from the relative Niño-3.4 index (right panel) generally has larger precipitation anomalies than the forecast reconstructed from the traditional Niño-3.4 index (central panel). The forecast reconstructed from the relative Niño-3.4 index matches the actual forecast strength (left panel) better than the one from the traditional Niño-3.4 index. In other words, the relative index describes the predicted winter pattern better than our traditional indicator of ENSO! This gives support for the claim that (1) the large amount of tropical warmth is enhancing the winter precipitation forecast anomaly, and (2) the relative Niño-3.4 index is a better indicator of the models’ precipitation forecast this winter than the traditional Niño-3.4 index (8).

Caveats

In this post, I have focused on the average of many individual precipitation forecasts, which filters out most of the unpredictable, chaotic weather (i.e., internal variability) that could make or break a single forecast. Therefore, we cannot rule out a very different outcome from what is forecast, but the “ensemble averaged” forecast gives an indication of where the odds of a wetter or drier forecast are enhanced. The current NMME precipitation forecast indicates that the probability of a La Niña-like winter precipitation pattern is higher than we might expect based on the traditional Niño-3.4 index. Importantly, this forecast also suggests that even if we do not meet the official criteria for La Niña this winter and end up ENSO-neutral, the tropical Pacific still might give a La Niña-like push to the climate over North America anyway. If that comes to pass, I would keep an eye on the relative Niño-3.4 index and see if that warmth throughout the tropics might be the reason why.

Footnotes

1. As I did last year, I averaged all the forecasts produced in September, October, and November of this year from 7 different models (one small difference is that I used the two recently updated versions of the Environment Canada models this year). Each model has a set of forecasts (ranging from 10 to 30) with slightly different initial conditions to sample the different possible realizations of chaotic weather variability. By taking the average of all forecasts, we tend to smooth out the effects of chaotic weather variability and isolate the predictable seasonal forecast signal. The NMME precipitation map was produced by averaging 372 individual forecast maps (124 for each of September, October, and November). It’s important to remember that each of those 372 individual forecast maps can differ substantially from the average map due the chaos of weather, just as chaotic weather can make the winter we actually experience deviate substantially from the average forecast.
2. There is no single, established method for doing this, but I came up with a precipitation forecast strength index, which involves (1) taking the square of all the ensemble mean forecast precipitation anomalies at each grid point (so all square values are positive regardless of the sign of the anomaly), (2) taking the area average over all North American land grid points, and (3) taking the square root of the resulting average (so that it’s a root mean square measure). I did this calculation for all available NMME precipitation forecasts from 1991-2024.
3. Specifically, the Pearson correlation coefficient between my precipitation forecast strength index and the ENSO forecast strength, defined as the absolute value of the average NMME forecast December – February Niño-3.4 index forecast, is 0.84. This correlation is quite high (maximum value for a correlation coefficient is 1), which means that the precipitation forecast strength index is usually a pretty good proxy for ENSO forecast strength. However, just because a relationship usually works doesn’t mean that it will continue to work, as this post demonstrates. It’s important to understand what physical mechanisms might cause a relationship to suddenly break down.
4. The average of all individual NMME Niño-3.4 index forecasts for this winter is -0.95°C, so the ENSO forecast strength is 0.95°C.
5. Usually, such simple modifications to an index or metric don’t work so well in the climate sciences. Why is subtraction of the tropical average sea surface temperature a good enough way to account for what’s going on outside of the central-eastern Pacific? It’s a little too complicated to go into all the details here, but the bottom line is that the temperature in the free troposphere (that layer in the atmosphere above about 5,000 feet and extending up to about 50,000 feet) does not vary horizontally very much in the deep tropics. This simplifies many calculations, and it means that the temperature of the entire tropical free troposphere varies closely with the average tropical sea surface temperature. Therefore, the average tropical sea surface temperature accounts for much of the atmospheric temperature variability that affects where and how much precipitation falls. The main point is that this is not just an ad hoc modification to the Niño-3.4 index – there is a lot of science that goes behind it!
6. One technical detail about my NMME calculations of the Niño-3.4 index: unlike with the official ONI, which uses base periods that are updated every five years, I used a single 1991-2020 base period to calculate the climatology. I chose this single base period because I do not have forecast data before 1991 to calculate updated base periods. However, as with the NMME forecasts, we also see a growing divide between the traditional and relative ONI in observations, even with the updated climatological base periods.
7. The “expected” precipitation forecast based on the Niño-3.4 index is the simple linear regression of December – February precipitation anomalies at each grid point with the forecast Niño-3.4 index (traditional or relative) as the predictor. The regression model is trained with all ensemble-averaged NMME forecasts from 1991-2024.
8. If you paid close attention, you may have noticed that I mentioned that this winter’s NMME precipitation forecast looks even stronger than last winter’s forecast, and that last winter also had a very large difference between the traditional and relative Niño-3.4 index. So, does it appear that the relative Niño-3.4 index can explain why last winter’s precipitation forecast was weaker than we might expect, given how strong the El Niño was by traditional measures? The answer is yes! At least, in part. In fact, from a relative Niño-3.4 index perspective, last year’s El Niño forecast was of similar strength to this year’s La Niña forecast.