Report: Weather Forecasts Indicating Natural Gas May Not Have Yet Bottomed

Below is a freely available report we have compiled for Bespoke Investment Group, a research group we have been increasingly working with.  A summary of the report has been posted on their website here, with the full report listed below.  SWCT/NY Weather Premium clients were made aware of this report last evening.

On November 21st, we wrote a report here indicating that the weather would warm significantly enough through most of the month of December so that natural gas demand would be limited and a previously-set short-term high would be locked in. We emphasized that drawdowns would be below average and supply would grow due to this lack of demand. The report analyzed a number of weather models, showing why we had very high confidence that arctic air would be bottled up in Alaska and northern Canada, allowing mild Pacific air to flood the continental United States. At that time, the February contract was trading around $4.60. On January 5th, that contract settled below $2.90 and has set even lower lows. Much has happened in between, and in this time the natural gas market has begun to separate greatly from pure weather forecasts and demand-side expectations. From reviewing the history of trading in the last 2 months and analyzing weather trends for the coming month, we are able to see that, despite calls that this downward trend may be over, the near-term bottom in natural gas has likely still not been set.

In late October and early November, natural gas was trading very well in line with weather forecasts – after all, winter weather is the number one demand driver for the home-heater. A run-up in early November was in-line with forecasts released to Premium clients on; the run-down in late November into December was also in line with what was released to clients. Around the 10th of December, things began to change. With the front-end contract price around $3.60, we began advising clients that, on a weather-basis, natural gas could be preferred long versus short, as the pattern looked to turn colder for the eastern half of the nation for the end of December into January. On December 14th, natural gas significantly gapped up 15 cents, but sold off during the physical trading on the 15th. From there, it never looked back, selling off all the way down to $2.80 despite increasingly cold forecasts. That heart of that cold is settling into the nation now, even as prices keep falling. The weather forecast was spot on; the market forecast was not.

A number of factors could be blamed for this sell-off; natural gas production has been much higher and stockpiles have significantly increased due to the mild December, just as oil has continued to sell-off. What could not be blamed for the second round of the sell-off were the extreme cold forecasts for January. But now, even the weather could be turning bearish, or at least less bullish, lowering the chances of a recovery in natural gas prices through the middle of January.

We want to begin with a forecast map for late this coming Friday night, seen here. The huge blue portion represents negative anomalies at 500mb in the atmosphere, and the large U shape across the east coast represents an upper level trough, which is injecting frigid arctic air straight into the northeast. This is a textbook cold outbreak setting, as you can even see warmer temperatures in western Canada/Alaska helping pump the much colder air down into the east coast. It was this extreme cold that was observed in mid to late December, and was expected to provide some support for the natural gas market. It still may provide some support in the way of very large EIA drawdowns, but it certainly has not provided much support thus far.

In meteorology, we see two types of forecasting: medium-range computer forecasts and longer-range pattern analysis. Medium-range you can think of being based primarily off of images and data like what is shown above. Longer-term patterns may be established not off of computer weather guidance, which can rarely accurately forecast past 14 days out, but instead off of key weather indicators and historical comparisons.

First, we will start with the forecast weather models. They have been bearish for mid-January for a while, though there has been a shift in them so that they now support significantly colder temperatures through January 17th. That trend began last Thursday, and we reported to our Premium clients Friday that models would trend colder in that timeframe over the weekend, providing short-term natural gas support. They did, and that likely contributed to a natural gas gap up pre-market Sunday night. The sell-off Monday, however, may have been slightly predicated in what we are going to outline here.

In the chart here is the 6z Tuesday GFS ensemble weather model guidance for Friday, January 16th. You can see the similarities to the pattern setting up this weekend, with the negative anomalies over the east coast and the positive anomalies over the west coast. This is the cold shot that was unexpected, although our models indicate this will only last 2 to maybe 3 days in any one location. It will result in very impressive cold over northern New England, but models are skeptical that cold will extend far back into the Midwest or be nearly as strong further Southeast for a sustained amount of time. Both the timing and location are very important to the forecast, because that means that the pattern can be considered “progressive.” In a progressive pattern, colder air moves in and out quickly; there are no prolonged cold outbreaks, and oscillations between above and below average temperatures can happen quicker. Cold shots can still come, but they will more often than not be offset by milder spells as well.

It is this GFS model chart that has many very concerned about mid-month warmth. For almost a week and a half we have been warning clients that the middle of January could be significantly warmer than the beginning, and models continue to agree on this despite trending colder January 15-17. We show this weather model because it is actually the coldest of the four main pieces of model guidance we use. In the pattern shown above, the negative anomalies over the east coast are kicked further east over eastern Canada, and the ridging over the west coast helping pump cold air down into the east has shifted further east, now pumping warmer air up into the Midwest and, in future frames, eventually into the east coast. Through the end of the run, the model shows positive heights returning back to the west coast with positive heights still along the east coast, setting up what is often referred to as a “gradient pattern.” In such a pattern, the northern half of the US can be slightly below average, though southern areas of the country remain near or above average. While not as mild as December, this is a pattern significantly more seasonable than the one we are in now. More concerning could be the next chart.

This is the Canadian weather model guidance for a similar timeframe as the American GFS weather model guidance. Yet again, you can see very strong agreement with the negative heights moving into Alaska and the positive heights moving into the east coast. This establishes stronger agreement between the models on this warming for the east coast past January 17th, and though it may take a few days to kick out the colder temperatures, the polar vortex branch is seen being kicked back into the north pole on this model (it is not even visible as an individual circle), making it harder for sustained cold like we are seeing this week. The European weather model (ECMWF) may be the warmest of all 3, showing a similarly mild pattern past January 17/18. And while a consensus did break down temporarily around January 15/16 on mild weather moving into the east, being replaced by the brief cold shot, models continue to indicate that any cold shot is transient and that seasonable, if not even above average temperatures, could be seen by as early as January 18th. The cold, meanwhile, is unlikely to rival the cold seen this week for most of the nation.

A fundamental question to this is analysis, though, is WHY are models showing this easing of the cold, especially after the extreme cold to enter the month? There are a number of reasons, which play into the second part of forecasting: the long-range forecasts based off of pattern analysis and raw data. Obviously, this is the more complicated analysis, but it is necessary in order to verify a computer consensus more than 10 or 12 days out. Overall, indicators that we are observing are increasingly supporting this idea of seasonal weather dominating through the middle and likely end of January.

The first indicator we use is the Quasi-Biennial Oscillation, or direction and strength of stratospheric winds. Though research on this is still being conducted, it can often correlate to strength of the Pacific jet and overall northern hemispheric pattern; a stronger Pacific jet can allow mild Pacific air to flood into the US, which was what happened in December and is what the concern is for the end of January (though it would not be nearly as extreme). The latest image of the 30mb QBO is attached here.

There’s a lot going on here, but we want to draw attention to the bottom graph. The new data was released yesterday, and it showed the QBO continuing to fall into December at the 30mb (red) level, an indication that strong westerly winds were still dominating in the stratosphere. The 50mb level continued falling too, but it lags the 30mb level as it takes time for changes in the 30mb level to propagate downward into the 50mb level. Our expected December 30mb QBO value was anywhere from -17 to -20, as it was expected to fall from the November value -23.65. The actual value was -25.35, the lowest value ever recorded for December, and this is a sign that there will be a strong influence from the Pacific Ocean on the pattern, resulting in potentially more seasonable conditions for the eastern half of the US into the second half of January. This confirms what our computer guidance shows.

Another indicator we use is the Madden-Julien Oscillation. We could spend awhile explaining exactly what it is, but essentially it is a measure of convection (thunderstorms, storms, etc.) in the western Pacific. A strong correlation between weather there and weather 2-3 weeks later in the US has been established and captured in an index. When the MJO is in phases 3-7 it tends to be warmer for the eastern US; 7, 8, 1, and 2 tend to be colder and snowier. The forecast from the European computer model for the MJO is shown here.

As seen, the MJO is currently in phase 5 and going into phase 6. Enough other factors came together to create the cold this week, overriding the MJO while it was in its warm phases, and the cold later next week may be driven by it approaching some of the colder phases, as it looks likely to pass near or through phase 7. What we want to explain about this here is not actually the fact that it will be driving warmth. Rather, it is important that the MJO will NOT be driving cold. In previous forecasts, many were expecting the index to swing around through quadrant 7 and into quadrants 8 and 1, the coldest and snowiest correlating quadrants. With latest forecasts indicating instead it may fall back into the center circle (in which there is no strong influence of storm activity in the western Pacific with North American weather) and then maybe moving back into the warm quadrants, there is little evidence for this index being able to provide a colder and snowier pattern for the continental United States. It is worth noting that MJO forecasts are notoriously unreliable, with our computer guidance and top forecasters rarely able to provide exact forecasts, but trends such as the one we explained above away from certain quadrants are just one more factor away from continued severe cold.

These are just a couple of the numerous indicators we look at to help determine weather in the 2-5 week range, which is one of the hardest timeframes to forecast for, especially in the winter months. But they do line up very well with the Premium winter forecast we released to clients in early October, which read:

“What we draw from all of this is a bias against significant snows this winter…In general, recent analogs support that February, in terms of anomaly; will be the coldest to normal of the months, with every analog since 1979 actually having at or above average temperatures for the month of January…we will have a winter dominated more in the January and February time frame than in the November and December time frame.”

In order to achieve above average temperatures in January, or at least get temperatures in the average range, we would need a warm-up after the front-end cold this month. The forecast is slightly higher confidence as it comes after a high verification for a very mild December. The cold snap in November was not identified until about two weeks out, but in any broad winter forecast released months before actual occurrence identification of broader trends, such as the sustained December warmth and lack of significant US snow cover, contributes more to actual natural gas demand.

Taking all of this together, we have an expectation that after a very cold start to January through the 17th, the next 10-12 days will average around to above average through the east coast. There is likely to be a continued battle between colder and warmer air, but there will also be abundant moderation, especially in the Midwest, which has seen some of the worst of the cold. This means that natural gas demand, while quite elevated to start the month, will begin to wane, and likely not crimp supply nearly as much as the month goes on. With record production, another sustained extreme winter would be necessary to cut into current supply, and despite potentially record-setting cold this week, it looks like, at least through the month of January, the cold will just not be sustained enough to support record prices. Price spikes will remain possible with sizable EIA drawdowns expected in the coming 2-3 Thursday reports, but downward pressure should remain.

So for prices, we do not yet believe that the short-term bottom has been set. Natural gas has been seeing extreme volatility recently as traders attempted to weigh the record production and potential long-term warmth against the extreme short and medium term cold. But with the worst of the cold only having around a week left, it is likely that you gradually see the Climate Prediction Center charts trend warmer in the longer-term over the coming 7-10 days unless weather models and forecasting indicators significantly change (like they did back in early November). Lessening the weather support could allow prices to move one leg lower, and even if in the next few days or the next week there is a run due to the excess demand in the short-term, this long-term easing of severely cold weather conditions should allow prices to re-test and likely break recent lows. The forecast all winter (again, released in early October to clients) has been for February to be the only month where cold and/or snow could be the sustained story, and it remains possible that some of these long-range forecasting indicators begin to shift as we approach that month to indicate that cold and snow. However, the operative word there was “could;” the new data shows it remains possible the milder weather carries over into February as well. Until February, and certainly for the front-end February natural gas contract, the easing of the severe cold for the majority of the country should contribute to price weakness in a market that has already seen extreme downward pressure.

Finally, while weather forecasts this time of year are notoriously volatile and low confidence, this one has enough going for it to likely influence the market. Natural gas has not correlated as well with recent weather forecasts and expectations, so before acting on any advice it is important to consider any of the other number of factors contributing to the recent decline as well. And with some of our models not in perfect agreement, whether this more seasonable trend is a couple weeks or a couple month trend is too early to say, and thus it is too early to determine how much weakness it brings to prices. But barring a major change in these weather models and indices over the coming days and weeks, at least one more run down of natural gas future prices looks likely before any potential for a demand-driven sustained recovery.

For any questions about this report or for additional information, please email Jacob Meisel at


Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>