During the past few days, a few bookies and odd-setters have contacted me, asking my take on the weather associated with the Super Bowl on Feb. 2 at the MetLife Stadium in Meadowlands Sports Complex in Rutherford, NJ, located just to the west of New York City.
In this blog, I will take a look at the forecast outlook for the game, making it clear what is possible to say nearly two weeks out from the big event. Since my blog is only read by Seahawks' fans, perhaps some of the information can be useful to the team and the 12th men/women that might attend.
Why is the weather important?
Sure, really bad weather might cancel or delay the fun, but the real question folks want to know is which team will be favored by the weather. There are all kinds of speculations in the media and online conversations:
(1) Many suggest that rainy and windy weather would be a big advantage for the Seahawks, an "all weather" team comfortable with the elements in Seattle. The Denver Broncos are used to far drier and sunnier weather. It turns out the local meteorologist Nick Bond has done an
analysis of the Seahawks win-loss records versus weather conditions, finding:
- When it rains or snows during Seahawks home games, the team’s record
during the past decade is 17-4, a winning percentage of 0.81. The
average point differential is almost 12.
- Without precipitation, the Seahawks record in home games is 42-25, a
winning percentage of 0.63. The average point differential is about 5.
Rain and storms are no problem
(2) Cold is seen as an ally for Seattle, since Bronco's quarterback Peyton Manning is said to be cold sensitive. As noted in an
authoritative article by CBS News:
When the temperature is 32 degrees or colder at kickoff, Peyton
Manning's career record is 4-7. That's a 36.4 winning percentage. In all
other games -- meaning, warmer games -- his winning percentage is 69.3
percent (174-77).
Reputably cold sensitive
So if the temperature is below 32F or if rain/snow is in the air, expect weather to advantage the Seahawks. Good weather, meteorology should not be a factor.
Can we forecast the weather for Superbowl Sunday a week or more ahead?
Although weather forecasting technology has improved dramatically over the past few decades, forecast skill--the ability to predict specific weather features-- fades in time. At longer forecast leads, generally greater than 7-8 days out, one is usually better using climatology.
So as any good forecaster does, lets start with the climatology (or historical average conditions) for Feb. 2 from nearby Newark, New Jersey:
Average high temperature is 40F
Average low temperature is 25F
The probability of getting at least .01 inch of rain over the day is 35%.
But we can do much better than that.
The game starts at 6:30 PM. The New Jersey State Climatologist Office
has computed the relevant statistics during game time. Start with temperature at kick off:
Averaged over roughly 80 years, the kick off (6:30 PM) temperature averaged 34F. But it is clear there has been warming over the past century, so perhaps 36F would be better, eyeballing the chart above.. However, this is for Newark and the stadium is in the Meadowlands, which is is usually a few degrees cooler than the highly developed Newark Airport area, so let's stick with 34F. So the expected temperature would be very close to the levels that would start giving Peyton Manning trouble. It is also highly probable that this will be the coldest kick-off in Super Bowl history: the current record is the 39-degree kickoff temperature at Super Bowl VI in 1972 in New Orleans’ Tulane Stadium.
Precipitation during the game? Any particular hour has about a 15% chance of precipitation.
This is a good start.
But let's get back to the issue of forecasting. The National Weather Service and some other operational forecasting agencies (like the European Center for Medium Range Forecasts, the Canadian Meteorological Center, the UKMET office) run their models out several weeks and current runs are now getting out to Super Bowl Sunday. But as I noted above, on average forecast skill degrades in time and generally offers no skill over climatology after aboiut a week. The top figure below shows how forecast skill for sea level pressure of major weather forecasting modeling systems declines with time up to 10 days (1 is a perfect forecast, GFS is the U.S. model). Once you get to around .7, climatology is as good as the model forecasts--about 168 hours out (7 days).
So you can't trust the detailed forecasts
in general out more than about 7 days. But sometimes there is skill extending out beyond 7 days (like with Superstorm Sandy) and there is a tool for determining when such days exists:
ensemble forecasts. But more on that later.
First, let me show you the forecast for 7 PM on Super Bowl Sunday
that you can't trust because they are out roughly 10 days, forecasts made by the National Weather Service GFS model. High pressure over the game site and no precipitation. Temperatures would probably be a bit colder than normal under such conditions.
But can we trust this solution? As noted above, a major weather prediction technology uses many forecasts valid at the same time:
ensemble prediction. Each forecast starts slightly differently or makes slightly different physical assumptions. If the forecasts stay together at the longer ranges, forecast confidence is higher. Also the average of the ensembles tends to be more reliable than a single forecast.
(Paul Allen connection: several of us have submitted a proposal to Paul Allen's foundation to use this technology to greatly enhance regional climate modeling, using an ensemble of climate models. And then we would use the large scale ensemble to predict regional details under climate change. A really promising technology with lots of good science.)
Here is the output of an ensemble made up of U.S. and Canadian global models (called NAEFS, the North American Ensemble Forecasting System) for 7 PM on Super Bowl Sunday. The solid lines are ensemble mean temperature (degrees C) and shaded areas tell you how uncertain the forecasts are (actually the variability of the various forecast model simulations).
Temperature around freezing (0C), and moderate uncertainty.
What about precipitation? Here is the output from the ensemble system (solid lines are precipitation (in mm) and shading is uncertainty. Light precipitation at most.
Finally, take a look at the ensemble predictions for one nearby point: New York City. The median (most probable) value is the black horizontal line. The yellow boxes tell you the range for half the ensembles around the median. The brackets show you the extremes of the various ensemble members. You notice there is less uncertainty (the ensembles are all on the same page) for the short term forecasts.
This ensemble output suggest that it isi probable that the temperatures will be warmer than currently (they are very cold now), but there is a lot of uncertainty. Precipitation should be light, but a few ensemble members are fairly wet. Winds are modest.
So what is the bottom line of all this? If you were a Seahawks strategist, you should expect upper 20s to low 30s and a lack of heavy precipitation.
You also have to keep your eyes on the forecasts because there is considerable uncertainty ten days out. Forecasts will lock in as we get 2-3 days from the big game. The Seahawks will probably not have the heavy precipitation advantage, but the temperatures have a good chance of being cold enough to give Peyton Manning some problems.
And with the above revelations the bookies will stop pestering me....
The kind of 12th Man that we could use on Feb. 2. He brings his own storms.