donsutherland1

For those who want a highly simplified but easy-to-understand description of sensitivity analysis, one can reference the following: https://thedecisionlab.com/reference-guide/statistics/sensitivity-analysis. And for those who want insight into the estimated probability, applying multiple variations to the model value, assessing the probability against a benchmark (1991-2020 normal value) using a normal distribution.

The January 15-16 event was highly unlikely to be a big snowstorm from the onset even if some social media accounts warned of a "big one." Aside from occasional operational GFS runs (unsupported by the other guidance and the various ensembles), no big event was on the table. Three important reasons apply: 1) The AO-/PNA+ pattern is just getting established. The trough was not likely to be sufficiently sharp for January 15-16. 2) There are numerous areas of vorticity competing with one another. The probability that the various areas of vorticity will develop into a single consolidated low that takes an ideal track for a classic NE snowstorm is low. Large splashes of color on vorticity maps don't always translate into surface potential. Neither do deep 500 mb lows. There's a lot involved for the upper levels and surface to become aligned. 3) The realistic scenarios of a light impact had been available for some time, even as the exact solution was outside the range of the guidance regarding the ability to resolve synoptic details. During the 1/11 0z EPS cycle, a single ensemble member had 6"+ snow (none had 10"+) and during the 1/11 12z EPS cycle, no ensemble members had 6"+ snowfall in New York City. For Boston, the figures were 6 and 4 members respectively. Bottom line: the probability of a significant NE snowstorm was low; the probability of a major NE snowstorm was extremely low. Based on the above three reasons, I never incorporated the idea of a significant or major snowstorm into my daily posts regarding the January 15-16 timeframe. If one is constructing a checklist, one needs both a favorable pattern and favorable synoptic details to get a KU-type snowstorm. If either is lacking, the probability of such an outcome is low. When both are lacking (as is the case regarding January 15-16), the odds are close to zero.

For those who are not familiar with it, sensitivity analysis incorporates uncertainty into an assessment. It is not deterministic in nature. It is not a point estimate. After the first 11 days of the month, the probability of a colder than normal January was just 55%. That means that there is still a lot of uncertainty. One sees it in the varying ensembles and weeklies and the lack of run-to-run continuity in cases. That's the key takeaway from any sensitivity analysis. It's capturing the large amount of uncertainty quite well, assuming one knows what one is looking at with the probabilities. For perspective, back in December, the estimated probability of a below normal monthly anomaly was 89% on December 7th. That was about as close to a "slam dunk" as one can get at such a long lead time. The estimated monthly mean temperature then was 34.0° (December finished at 33.8°). The major weakness is that the analysis assumes a Gaussian curve (normal distribution). Not everything is normally distributed. This assumption is necessary, because the kind of detailed model verification data that would be needed to construct a more robust curve is not publicly available. Thus, the simplifying assumption of a normal distribution is used. Finally, I include the estimated monthly mean temperature only because it's nice to have. But one should know what one is looking at. The estimated mean is similar to an ensemble mean, while the probability of an outcome plays the role akin to an ensemble spread (low probability reflects a large spread in outcomes due to high uncertainty). IMO, the probability of an outcome (I just use below or above normal) is really the more useful element of the piece concerning sensitivity analysis, especially at long lead times.

After 34 consecutive days during which it was negative, the PNA went positive today (+0.173). Whether a persistent PNA+ regime develops will have important medium- and long-range implications. Following this evenings snow squall and snow showers, modestly cooler air is returning to the region. Through midweek, highs will generally reach the 40s during the daytime and 30s for lows in New York City. Somewhat colder readings are likely outside the City and in areas where strong radiational cooling takes place. After the middle of next week, temperatures will "step down" with highs mainly in the middle and upper 30s in New York City and lows in the middle and upper 20s. Some light precipitation is possible on Friday. Rain showers could transition to a period of snow. A light accumulation is possible. Additional precipitation could arrive on Sunday. No significant Arctic blasts or significant snowfalls are likely through at least mid-January. Afterward, conditions might become more favorable for both cold and snowfall, especially if the PNA remains predominantly positive, as has often occurred following the breakdown of long-duration PNA- regimes. PNA-related developments would have larger implications for snowfall. A persistently positive PNA would have above climatological risk of moderate or significant snowfalls. A mainly negative PNA would favor mainly small snowfalls. It will likely be another day or two before the guidance reaches the high-skill timeframe for teleconnection forecasts. The ENSO Region 1+2 anomaly was -0.7°C and the Region 3.4 anomaly was -0.5°C for the week centered around December 31. For the past six weeks, the ENSO Region 1+2 anomaly has averaged -0.37°C and the ENSO Region 3.4 anomaly has averaged -0.63°C. La Niña conditions will likely continue into at least late winter. The SOI was +20.69 today. The preliminary Arctic Oscillation (AO) was -1.499 today. Based on sensitivity analysis applied to the latest guidance, there is an implied near 55% probability that New York City will have a cooler than normal January (1991-2020 normal). January will likely finish with a mean temperature near 33.5° (-0.2° below normal). Supplemental Information: The projected mean would be 0.9° above the 1981-2010 normal monthly value.

Three photos from Armonk during this evening's snow squall and a short video:

Beware of the latest ill-informed Social Media hype: No "big" snowstorm is likely for the Northeast during January 15-16. Key factors argue against it: 1) The AO-/PNA+ pattern is just getting established. The trough is not likely to be sufficiently sharp. 2) There are numerous areas of vorticity competing with one another. The probability that the various areas of vorticity will develop into a single consolidated low that takes an ideal track for a classic NE snowstorm is low. Notice that only the operational GFS was shown. Moreover, large splashes of color on vorticity maps don't always translate into surface potential. There's a lot involved for the upper levels and surface to become aligned. 3) The realistic scenarios have been available for some time, even as the exact solution cannot yet be pinned down. During the 1/11 0z EPS cycle, a single ensemble member had 6"+ snow (none had 10"+) and during the 1/11 12z EPS cycle, no ensemble members had 6"+ snowfall in New York City. For Boston, the figures were 6 and 4 members respectively. Bottom line: the probability of a significant NE snowstorm is low; the probability of a major NE snowstorm is extremely low. What's far more likely than a big snowstorm is the idea that there will be periods of precipitation that transition to periods of snow, as some areas could start as light rain. A light accumulation is plausible in the New York City area. Details are uncertain, but a 1"-3"/2"-4" type snowfall would be vastly more likely than a 6"+ one. Southeast New England has a better chance of seeing a moderate accumulation. Even if some areas of 6" or above snows develop over a portion of SE New England, the areal coverage of such amounts will likely be limited. By no definition will this be a classic NE snowstorm. It won't be something that would be placed in the iconic KU Northeast Snowstorms books. Best guess from this far out: At most, maybe one of the following locations will see 6" or above snowfall during January 15-16: Baltimore, Boston, New York City, Newark, Philadelphia, Washington, DC. Most will see < 3" of snow. That's not a "big one." By the way, that's the same outfit that ignored the guidance and also failed to understand how a PNA- teleconnection translates downstream (SE ridge/warmth in the SE), and said that places like Atlanta would have a much colder than normal January. January 1-10 has a mean anomaly just above 12° above normal in Atlanta. Just for January 2026 to reach normal, Atlanta would need to experience its coldest January 11-31 since 1985! Again, as noted ad nauseum, anyone can play forecaster on social media. There, "clicks" and "views," not knowledge, skill, or accuracy are the currency of value. Entertainment value not quality of information is promoted and monetized. Hence, it's no surprise whatsover that what might become a moderate storm, if things work out (still not assured), is being hyped as a potential "big one."

A snow squall has moved into the Binghamton area. Rain and snow showers remain possible across the NYC area later this afternoon or early this evening with the frontal passage.

The model should use the 4dVAR initialization scheme used by the leading models (ECMWF, GGEM, UKMET). It's difficult to justify not using the best-in-field initialization at this point in time. The inferior initialization puts the model at a disadvantage right from the start. Once the GGEM moved to the 4dVAR initialization, along with other changes that were made, it rocketed to the #2 model in terms of key verification metrics.

Barring some big changes in the guidance, it continues to appear that the Thursday-Friday period will see some precipitation, but probably < 0.50". A series of waves should move along a slow-moving front, but the consolidated low will likely develop too far offshore for a more significant impact. Larger impacts remain more likely across southeastern and eastern New England. Currently, fewer than 20% of EPS ensemble members show 0.50" or more QPF for the New York City area. The 1/11 0z EPS ensemble mean QPF was pretty close to the 16z NBM output. Overall, there remains little evidence that a pattern conducive to large snowstorms coupled with the ingredients necessary will be in place by the January 15-16 timeframe. Nevertheless, rain showers could transition to snow showers or even some periods snow around the New York City area and its immediate suburbs. A light accumulation of snow is plausible. More immediately, a line of rain and/or snow showers accompanying a frontal passage appears likely to cross the region late this afternoon or early this evening coupled with gusty winds. No accumulation is likely in and around NYC, but temperatures should fall from the middle 40s into the upper 30s or lower 40s during the frontal passage.

Amidst all the discussion of a possible strong WWB this month, the development of a pretty strong EWB slipped under the radar. In fact, the past two days have seen the SOI at or above +20. Interestingly enough, a slightly weaker EWB (just below +20) preceded the record-January WWB of 2017 by about a week.

Yesterday, JFK Airport, LaGuardia Airport, and Newark all had highs of 53°. That followed their having identical highs of 54° a day earlier. The last time all three sites had identical highs on two consecutive days was November 11-12, 2025 when the highs were 42° and 51° respectively.

Right now, the NAO still appears likely to be mainly positive. It wouldn't be a "deal-breaker."

The PNA has gone positive with a value of +0.173. As noted previously, all 10 long-duration winter PNA- regimes (25 or more consecutive days) have been followed by a regime change to positive with more than two-thirds of the succeeding 30 days being above 0.

Here you go. Note: I don't have data for Morristown.

An orderly pattern progression remains underway. The warm anomalies peaked yesterday with temperatures running 12.6° above normal in Central Park. In the extended range, that progression should culminate in a return to colder weather with persistently below normal temperatures. Rain and showers will end early tomorrow. Rainfall remains on track for a storm-total 0.50"-1.00" rainfall across much of the region by the time the last showers depart early tomorrow. Through the middle of next week, highs will generally reach the 40s during the daytime and 30s for lows in New York City. Somewhat colder readings are likely outside the City and in areas where strong radiational cooling takes place. After the middle of next week, temperatures will "step down" with highs mainly in the middle and upper 30s in New York City and lows in the middle and upper 20s. Some light precipitation is possible on Friday. No significant Arctic blasts or significant snowfalls are likely through at least mid-January. Afterward, conditions might become more favorable for both cold and snowfall, especially if the PNA goes positive. PNA-related developments would have larger implications for snowfall. A persistently positive PNA would have above climatological risk of moderate or significant snowfalls. A mainly negative PNA would favor mainly small snowfalls. It will likely be another day or two before the guidance reaches the high-skill timeframe for teleconnection forecasts. The ENSO Region 1+2 anomaly was -0.7°C and the Region 3.4 anomaly was -0.5°C for the week centered around December 31. For the past six weeks, the ENSO Region 1+2 anomaly has averaged -0.37°C and the ENSO Region 3.4 anomaly has averaged -0.63°C. La Niña conditions will likely continue into at least late winter. The SOI was +21.91 today. The preliminary Arctic Oscillation (AO) was -2.082 today. Based on sensitivity analysis applied to the latest guidance, there is an implied near 58% probability that New York City will have a cooler than normal January (1991-2020 normal). January will likely finish with a mean temperature near 33.3° (-0.4° below normal). Supplemental Information: The projected mean would be 0.7° above the 1981-2010 normal monthly value.