Saturday, February 17, 2007
Post storm assessment: Storm was well-forecast
Jason Samenow @ 3:30 PM
Despite considerable chatter about how complex last week's storm was and that the forecast was sub-par, a re-examination leaves me with a different impression: the storm wasn't all that complex and the forecast was generally good.
Last Wednesday, February 7 CapitalWeather.com, in its Snow Lover's Crystal Ball first identified the potential for a storm:
As early as last Saturday, we mentioned the possibility of a wintry mix entering the equation. We conveyed the forecast uncertainty indicating "a storm track closer to the coast or inland might result in a changeover to mixed precipitation or rain along."
On Sunday morning, we communicated with increased confidence "that an important storm with significant impacts from the Mid Atlantic to the Northeast will develop" but cautioned "for areas along and east of I-95, a changeover to mixed precipitation is a possibility during the second half of Tuesday." By Sunday late afternoon (more than 36 hours out), we came to the realization that an ice storm had become more likely:
We issued our first accumulation forecast late Sunday night calling for generally 1-3" of snow +ice (2-4" snow +ice in the NW suburbs)-- a forecast which we stuck to throughout the event and which largely verified for most of the metro area. We provided nowcasts and updates throughout the storm, and our overall message did not change -- that significant icing was likely particularly north and west of DC.
Of course, we erred in underplaying ice possibilities in DC and to the south and east as did most forecasters (except the National Weather Service who predicted subfreezing temperatures for the bulk of the storm consistently for most areas). And we updated our forecast when it became clear temperatures were not going to rise.
The models, which we use as tools for making predictions, generally forecast this event quite well. Initially, they incorrectly suggested we'd have a snowy event. This solution resulted from their placement of cold high pressure far enough east to hold enough cold air in place for snow. But gradually, the models realized this area of high pressure would actually be positioned further west, allowing the storm and its warm air to move further north. So, it became increasingly obvious that mixed precipitation was more likely than all snow. The models also did a reasonably good job indicating that the cold air at the surface would be difficult to dislodge due to persistent northeast winds (which they accurately simulated). Some forecasters ignored this, but to their detriment.
The was a fairly well-forecast storm -- by both computer models and human forecasters. Mistakes were made -- particularly underestimating (to varying degrees by different forecasters) the staying power of the cold air at the surface overnight Tuesday into Wednesday morning. But from the standpoint of providing long lead time in predicting the storm's impact, most forecasters did well -- aided by the generally strong performance of forecast models.
Pictured: The Washington Monument (top) and the Lincoln Memorial (middle) at dawn today. By CapitalWeather.com photographer Kevin Ambrose.
The Evolution of the Forecast
Last Wednesday, February 7 CapitalWeather.com, in its Snow Lover's Crystal Ball first identified the potential for a storm: With some surprising consistency this far out, models are hinting at the possibility of a substantial winter storm for the East Coast during the Tuesday-Wednesday period of next week.We also mentioned the likelihood that the situation could change/evolve and that the models might give us conflicting guidance (i.e. referring to the "the model waffle dance" the next day).
As early as last Saturday, we mentioned the possibility of a wintry mix entering the equation. We conveyed the forecast uncertainty indicating "a storm track closer to the coast or inland might result in a changeover to mixed precipitation or rain along."
On Sunday morning, we communicated with increased confidence "that an important storm with significant impacts from the Mid Atlantic to the Northeast will develop" but cautioned "for areas along and east of I-95, a changeover to mixed precipitation is a possibility during the second half of Tuesday." By Sunday late afternoon (more than 36 hours out), we came to the realization that an ice storm had become more likely:
...significant icing has emerged as a more significant threat. This decreases the probability of heavy snowfall amounts of 6" or more, but does not reduce the storm's potential seriousness and overall impact.
We issued our first accumulation forecast late Sunday night calling for generally 1-3" of snow +ice (2-4" snow +ice in the NW suburbs)-- a forecast which we stuck to throughout the event and which largely verified for most of the metro area. We provided nowcasts and updates throughout the storm, and our overall message did not change -- that significant icing was likely particularly north and west of DC. Of course, we erred in underplaying ice possibilities in DC and to the south and east as did most forecasters (except the National Weather Service who predicted subfreezing temperatures for the bulk of the storm consistently for most areas). And we updated our forecast when it became clear temperatures were not going to rise.
A Word About the Computer Models
The models, which we use as tools for making predictions, generally forecast this event quite well. Initially, they incorrectly suggested we'd have a snowy event. This solution resulted from their placement of cold high pressure far enough east to hold enough cold air in place for snow. But gradually, the models realized this area of high pressure would actually be positioned further west, allowing the storm and its warm air to move further north. So, it became increasingly obvious that mixed precipitation was more likely than all snow. The models also did a reasonably good job indicating that the cold air at the surface would be difficult to dislodge due to persistent northeast winds (which they accurately simulated). Some forecasters ignored this, but to their detriment.
Summary
The was a fairly well-forecast storm -- by both computer models and human forecasters. Mistakes were made -- particularly underestimating (to varying degrees by different forecasters) the staying power of the cold air at the surface overnight Tuesday into Wednesday morning. But from the standpoint of providing long lead time in predicting the storm's impact, most forecasters did well -- aided by the generally strong performance of forecast models.
Pictured: The Washington Monument (top) and the Lincoln Memorial (middle) at dawn today. By CapitalWeather.com photographer Kevin Ambrose.
Comments are closed for this archived entry | Link | 
Email this post
Email this post
