Thursday, January 4, 2018

A Remarkable Forecast of a Beast of a Storm

Intense frontal cyclones don't come much prettier than the nor'easter rampaging up the east coast today.  Satellite imagery and radar for 1200 UTC shows the beautifully wrapped up system with the low center just a shade ESE of Virginia Beach.  Precipitation was heaviest near or just offshore (radar imagery well offshore is nonexistent). 


If we slap on the RAP 925-mb analysis (roughly 750 m above sea level and a good level for seeing the frontal structure) we see the classic structure of an intense frontal cyclone predicted by yesterday's NAM (see prior post) with the cold and occluded/warm fronts oriented at right angles, weakening of the temperature contrast associated with the cold front near the occluded/warm fronts (known as the frontal fracture), the frontal temperature contrast associated with the occluded front maximizing near and west of the low center, and the occluded front extending through the low center as a back-bent occlusion.  One can also see a near cutoff pocket of warm air near the low center (a.k.a. the warm-core seclusion) as cold air encircles the system.  The area in purple shows an intense low-level jet with winds in excess of 40 m/s (80 knots, light purple) wrapping cyclonically from the west to south of the low center, culminating in a maximum in excess of 45 m/s (90 knots, dark purple) known as the poisonous tail of the back-bent occlusion.  In this case, the wind maximum likely represents a sting jet, a local wind maximum near the tip of the comma cloud head that is produced by the descent of strong winds from aloft.  For more on this subject, see What is a Sting Jet?


The 30 hour NAM forecasts that we presented yesterday were quite remarkable and I've reproduced them below for comparison with the RAP analyses above.  The frontal structure, low center position, and low center intensity are very well captured.  The NAM forecast central pressure of 961 mb is a bit overdone compared to the RAP analysis 967 mb, but analysts at the National Weather Service Weather Prediction center put it at 960 mb, so the NAM forecast is certainly within the uncertainty. 




Such a forecast is a remarkable scientific achievement.  We shouldn't take it for granted.  Through the mid 1980s, operational numerical modeling systems frequently failed to predict intense frontal cyclone development of this type.  Scientific papers describe errors in central pressure forecasts of as large as 55 mb.  You read that right.  55 mb.  Basically, a complete and total failure to predict the cyclone development. 

Today, it's almost impossible to believe numerical forecasts could be that bad, but they were, and many on the high seas paid with their lives.   It is only through advances in understanding, observing systems, computing infrastructure, and numerical modeling techniques that we knew a storm like the one above was coming (in fact, many days in advance).  Surely there will be some issues with details of the local forecasts that require further research and model improvements, but forecasts of intense frontal cyclones have come a long long ways. 

3 comments:

  1. Back in the early 80's the most powerful computer had only 64 million bytes of memory (your typical PC has ~100 times more) and ran 400 million floating point operations per second - theoretical top speed. With such small memory, it limits horizontal (and vertical) resolution and the amount and quality of physical processes one can simulate. So, pretty cool how computer technology is the "base" that Atmospheric Science has built upon to make these amazing forecasts.

    ReplyDelete
  2. The "base" still has to be the data. The conditions in the field. The computer tech and sophisticated modeling is amazing, especially how fast its advancing. But accurate field data is the "base". Same old GIGO principle.

    ReplyDelete
  3. Absolutely! Thanks for pointing out the importance of improvements in model initialization. Back in the 80's the field was just starting to investigate using satellites for data assimilation. I thought about my omission overnight. :)

    ReplyDelete