wpostServer: http://css.washingtonpost.com/wpost

The Post Most: Local

Posted at 12:58 PM ET, 07/03/2012

Could forecasters have better predicted the June 29 derecho?


A mature convective system known as a derecho impacts the Washington, D.C. area late evening on June 29, 2012 as seen from infrared satellite. (NOAA)
We know a number of facts about the June 29 derecho now. As many as 5 million lost power from near Chicago to the East Coast, at least 22 were killed, and it caused a yet uncounted millions or billions in damage. Yet, as of 24 hours prior and even the morning of, very little in the way of mass indication that it was coming.

While the United States moves toward becoming a weather ready nation, how could something like this happen with half a day or less of warning? Is it destined to happen again?

The story starts with a somewhat atypical weather pattern for this area, but one known to produce large-scale severe weather events from the Plains to the East Coast. Possibly the most important part of the pattern is a jet stream flow nearby, often traveling around the north end a big ridge of high pressure producing significant heat.

The “ring of fire,” where these air masses clash, is a well known location of summertime thunderstorms.

Related: Derecho: Behind Washington, D.C.’s destructive thunderstorm outbreak, June 29, 2012

But, what made this event special all the way to the East Coast, and were any early clues ignored or overlooked?


Derecho climatology in the United States. Note: the main grid study here did not include the mid-Atlantic states east of the Appalachians, so return rate may be high in our region. Also, this return rate includes considerably smaller events (compared to June 29) that were still classified derechos by meeting criteria of path in which damaging winds occurred. (NOAA SPC)

Again, derechos are not that common, particularly in this area. In other words, this was anything but the typical storm.

It’s extremely hard to nail down specifics about a derecho’s track and intensity well ahead of time. Even those derechos regarded as among the most powerful on record start from benign activity. It’s only once storms and their associated cold pools of air roll into a supportive downstream environment that development and expansion occurs.

Climatology states that the D.C. area can expect a derecho (defined as a storm complex with a wind swath of 240+ miles in path) about every four years. This gives no regard to maximum intensity once suprassing those criteria, nor extreme path lengths or north/south coverage. Our area last saw something similar on June 4, 2008 — that event had a much smaller footprint of damaging winds, but impacted the D.C. area severely nonetheless.


Panels showing jet stream configuration and energy maximums (yellow/orange at top and red at bottom) at the 500mb level. The area where the lines compress is the jet stream.The top image is a forecast from the Global Forecast System during the evening of June 27 (0z 6/28) for Friday, June 29. An area from the Midwest to our region was at the base of the stronger westerlies associated with the jet and at the top of a large ridge of high pressure to the south. Comparisons to past events (“more trees down derecho” in 1980 and June 4, 2008) that produced widespread wind events in the D.C. area are shown below the forecast map from the GFS. (NOAA NCEP and PSU EWALL NARR)
To get long-track wind events such as this, the D.C. area almost always needs a fast jet stream flow aloft from the west to be located just to the north. In the week leading up to the event, models began to advertise that the area would fall into that type of pattern. Other ingredients, those of instability, moisture and heat (in this case high to extreme on all counts) also started to fall into place.

What was lacking on many model runs was a consistent signal for precipitation or a precipitation trigger. Some models had one or the other, then they didn’t. Others never really had storms at all.

But, a closer examination of potential energy (at jet stream level — 500mb) showed consistent inklings that small impulses would ride along in the northwest flow at the edge of the upper-level heat dome to the south. Questions revolved around timing, and if they’d be strong enough to touch off an event.

Basically, the atmosphere was headed toward tinderbox levels, and all that was needed was one little disturbance to act as the proverbial match. That eventually got underway Friday morning in the Midwest.


Warnings left no spot untouched on June 29. They came faster than the line moved, and early enough for action as long as people were tuned in. Yellow polygons are severe thunderstorm warnings, red are tornado warnings and purple are marine warnings. (IEM Local Storm Report App)
Warnings during the event itself were very good. The National Weather Service, as they’ve become known for, issued long-lead storm storm warnings with accuracy and precision. Before that, by late afternoon and evening, the idea that something big lurked over the horizon was well communicated across both media and government sources as well.

But, in almost all cases it appeared everyone was behind the ball on acknowledging the potential threat before it got underway.

Let’s say a high-end tropical storm was barreling into the area. Folks in the path would probably want more than a few hours heads up and even less time of actual warning. Right?

Many or all morning forecasts on June 29, even ours mentioning the risk of late-day storms, did not really hint at the extreme potential ahead. And in all likelihood that was not because no one saw that potential. Instead, the uncertainty of a pattern with no discernible large-scale trigger for storms. That made actually forecasting an event to occur a tricky proposition.


Forecast convective available potential energy (CAPE) for Friday evening from the Global Forecast System over four days prior to Friday. Very high to extreme levels of CAPE are shown across much of the region ultimately hit by the derecho. Forecast soundings indicated a Plains-like elevated mixed layer helping to create a potentially volatile environment. (TwisterData.com)

Compare the derecho lead up to that of the Maryland tornado outbreak of June 1. Warnings about the tornado event came early and often. It was well-modeled and well-communicated that a storm would happen. While the system eventually produced enough weak tornadoes to put it up on the list of the most prolific local outbreaks in recent times, the number of people actually impacted by severe weather on June 1 was a tiny fraction of the number impacted June 29.

So, did the derecho of June 29 reveal a missing link in the chain of meteorological success across the board in recent history? In some ways, it might have.

As CWG readers know, weather is best described in probabilities. But, in reality, storms either happen or they don’t. The issue is there are days like Friday, when storms might not happen, but if they do, they will be explosive.

Is the meteorological community better erring on the side that storms won’t manage to reach potential? Or err on the side of caution and risk false alarms?

Link: Overcoming the tornado false alarm problem


An index for severe weather, “Craven Brooks Significant Severe,” shows anomalously high levels from near Chicago through the Ohio Valley. Lesser, but still notable, values show up across the mid-Atlantic. This is from the Short Range Ensemble Forecast run by the Storm Prediction Center 84 hours prior to Friday afternoon. (NOAA SPC SREF)

The Storm Prediction Center’s (SPC’s) outlooks in the days leading up to the event showed a corridor of possible thunderstorm activity in fairly close proximity to the actual event. Based on their text, confidence was low that thunderstorms would form. This lack of confidence continued to plague their threat level forecast into June 29. Their early-morning verification (how accurate their forecast was) that day (shown below) was not so good.

Little doubt the preeminent severe weather forecasters in the country at SPC saw the players on the field for more than they were officially forecasting. But uncertainty in a trigger likely kept them from sounding an alert. Uncertainty rules, even with the unusual potential, heavily provided by an air mass characterized by an advection of an elevated mixed layer toward the area, more typically found over the severe weather capital of the Great Plains.

I’m not singling SPC out for any reason other than it’s easy to track their forecasts. To their credit, they did quickly ramp up the threat level (or risk) as the day progressed. But, generally they were a little slow in sounding the alarm.

The upgrades to the threat levels came after the formative derecho had already begun to produce widespread damage reports. And the serious upgrade in threat for the high-population centers of the East Coast did not come until the 9 p.m. update, as the line was already moving into the area.


The progression of severe weather outlooks from the Storm Prediction Center regarding June 29. Day 3 (top left) was issued June 27 and Day 2 (top right) was issued June 28. Neither outlook went even as far as to issue a slight risk of severe thunderstorms in the D.C. area (which is very common here and elsewhere). The day 1 forecast (bottom) was first issued shortly after midnight (6z) on June 29. In each case, odds for severe weather were indicated to be low from Ohio through the mid-Atlantic. In the 6z outlook the day of, the risk for damaging winds was less than 5% from a specified point (uncontoured) in the D.C. area. (NOAA SPC)

The intensity of the storms also raises the question as to whether or not Potentially Dangerous Situation thunderstorm watches should have been issued. In retrospect, the answer is almost certainly yes.

It’s important to point out earlier warnings probably wouldn’t have significantly altered the destruction caused by this event. Though the power companies are always easy scapegoats, it’s unlikely they could have prepared in a way that would have prevented mass outages.

On the other hand, earlier warnings would’ve enabled more people to better secure outdoor items, charge their portable devices and seek shelter. Fewer would’ve have been caught off-guard by the storm’s sudden fury and in harms way.


Hi res model runs of the WRF ran the night prior to the storms show varying levels of storminess in similar shapes to what ended up occurring in the D.C. area. To their (dis)credit, these models have a tendency to be erratic and overforecast storm activity. (NOAA NCEP)

The bottom line is that it’s simply not possible, despite dire warnings, to properly inform the public at large in a three-to-six hour timeframe, even with instant mass communications we have today. Longer lead times are needed.

Numerous questions revolve around how the community can properly communicate a “derecho pattern” and share the knowledge that the pattern does not always produce. Then there is the problem of explaining the fact that if it does occur, figuring out where the storms might track poses another dilemma, one in which there may be no answer other than to wait and watch.

Regardless of the difficulties involved, without overwarning to the point of fatigue, certain circumstances - like those of June 29 - require extra attention and better communication of risk. The analyses sprinkled throughout this post show the potential for a dangerous weather event over a large, heavily populated area was present, even if not clear.

Allowing days like this to appear like a general summer thunderstorm day, the type that many have come to ignore, is not a practice that should be accepted - particularly in cases when so much is in the line of fire. And we must get better about learning how to most effectively communicate the threat level with as much lead time as possible.


A compilation of maximum radar reflectivity as the derecho moved from the Midwest to the East Coast. (IEM)

Post script, 7/5/2012: The above discussion did not fully describe the full extent of advanced warning provided by the Storm Prediction Center in the several hours leading up to the derecho. It issued a special mesoscale discussion at 6:03 p.m. highlighting the severe weather threat for our region which stated: “WITH AN EXTREMELY UNSTABLE DOWNSTREAM AIRMASS WHERE SURFACE TEMPERATURES HAVE HEATED INTO THE LOW 100S IN MANY LOCATIONS...EXPECT THE DERECHO TO CONTINUE”. At 6:35 p.m., importantly, it issued a Severe Thunderstorm Watch which stated thunderstorm wind gusts to 80 mph were possible.

By  |  12:58 PM ET, 07/03/2012

Categories:  Latest, Thunderstorms

 
Read what others are saying
     

    © 2011 The Washington Post Company