Friday, September 9, 2011

A thousand year rain in DC?

What an incredible rainfall event over the MidAtlantic states!  The remnants of tropical storm "Lee" nearly matched the ultimate rainfall producer in the 20th Century in these parts called "Agnes" in 1972.  Rainfall exceeded 20" in a few places in southern St. Charles county in Maryland while amounts exceeding 15" were reported near Harrisburg, PA and over a foot was widespread from Virginia north into New York.  Binghamton, NY actually exceeded 10" and broke their all time rainfall record for a single day. The Hydrometeorological Prediction Center has a nice compilation of the huge rain totals from "Lee".


Many areas have likely gotten rainfall that they haven't seen in decades including the Washington DC area.  What is really amazing is that a few sites saw three to six hour rainfalls that exceeding even a 1000 year return interval.  Even by tropical cyclone standards this is a pretty incredible total.

One case in point is the rainfall observed at Ft. Belvoir, VA.   This site was nearly in the middle of a huge rainfall maximum of 9 - 12" that mostly accumulated on September 8, or the second day of heavy rain in the area (figure 1).  A majority of this rain fell during the afternoon of the 8th as thunderstorms erupted in a north to south axis right across the western half of the Washington DC metro area.

Figure 1.  12 hour gauge bias-corrected radar-estimated rainfall centered over Washington DC and ending at 09 September 2011 01 UTC.   The arrow points to Ft. Belvoir (KDAA).  This data is available at http://www.srh.noaa.gov/ridge2/RFC_Precip/.
What I am really impressed with is the intensity of this deluge.  Let's take the same type of graphic above  but for the three hour rainfall (figure 2).  The rainfall exceeded 7" along I-95 south of the Beltway while more than 5" of rain extended up toward Arlington, VA.  No doubt this was big rainfall and the flash flooding impacts were huge namely because of the dense population.  

Figure 2.  Similar to figure 1 except for three hour rainfall ending at 01 UTC.

Consider the rainfall accumulation at Ft. Belvoir (KDAA) as it compares to the expected time interval in which these kinds of rains are expected to occur.  This is near the epicenter of the heaviest rainfall (figure 3).  The measured hourly rainfall was quite heavy, peaking at 2.65" ending at 22 UTC but this kind of rainfall can be expected to occur once every 20 - 25 years.  It's rare but nowhere near as rare as seeing this kind of rainfall rate persist past one hour to three hours!  But that's what happened at Ft. Belvoir.  They saw greater than 2"/hour rainfall rates last three hours yielding over seven inches!  That's where we see that rainfall entering into the 1000 year return interval.   Finally the thunderstorms let up and the rainfall rates diminished.  Notice that the 12 hour rainfall of 7.9" is around a once in 200 - 300 year event and the 24 hour total of 9.3" is expected to occur once in perhaps 150 years.

Figure 3.  A plot of rainfall vs period of rainfall accumulation for Ft. Belvoir, VA (black trace).   In addition,  this plot shows the maximum expected rainfall as a function of time for several return periods ranging from one year to 1000 years.  Notice that the three hour rainfall at Ft. Belvoir exceeded the 1000 year return period!  Rain return intervals acquired from NWS Hydrological Design Studies Center.
Consider that the nature of the land becomes accustomed to a typical rainfall pattern where the return intervals are fairly small.  By saying accustomed, I'm implying that the nature of the stream banks are shaped by frequent periods of rainfall and subsequent runoffs.  The soil accumulates in a certain way in response to normal rainfall and runoff too.  The vegetation grows in a similar way from the grass to the trees that line stream banks and other low spots.  The root systems grab on to the soil only as tightly as needed to hold on in normal rainfall and runoff patterns.  And we build our infrastructure in response to what we consider the normal range of events from the height of bridges to where we consider the boundaries of a flood zone.  Certainly in the age where we build cheaply, we're even more dependent on staying within what we consider normal events.

So can you imagine what happens in a once in a thousand year rainfall event?  Perhaps these pictures compiled from the Capital Weather Gang may provide some justice to the impacts.  Needless to say the impacts were big in a negative way with multiple high water rescues, washed out roads, flooded buildings, and transportation halted in general.   Numerous school districts were closed the following days.  Residents in some areas were isolated due to damaged roads.  And there were four fatalities.

So in summary, it's not just the rainfall amounts but also the rarity of these events that's important.   Even though a thousand year rainfall event is basically a statistically-based extrapolation, the point is made that we should expect major societal impacts when the event is rare.  The three hour rain that fell in Ft. Belvoir would even match a 100 year event along the upper Gulf Coast of the US according to the Rainfall Frequency Atlas.

That's all for now, however I'm going to look at what it takes to get a three hour rainfall of that magnitude from a meteorological point of view.

Sunday, September 4, 2011

Fires and intense updrafts in east Norman

While working on a winter weather course, a colleague of mine shouted about a fire visible to the east of the National Weather Center.  Naturally this was something I had to see. This was the second time a major fire was visible from the Center, the last one being a major apartment complex fire a mile to the east.  Upon reaching the observation deck, I was treated to an impressive column of smoke to the southeast that appeared extend almost overhead.  

The smoke column extending upward from the Noble fire around 2:30 pm CDT.
Flames were easily visible at the bottom of the column as you can see from the video below.  Some of these flames were most likely over 100' tall as numerous dry Eastern Red Cedars caught fire.  Columns of darker smoke erupted as trees or groves of Cedars ignited.


This fire wasn't just one of those single acre type grass fires that we see on roadsides though it certainly started small.  No, the potential of this fire to go out of control and consume thousands of acres was real.  We were already in an exceptional long period drought and the Oklahoma Mesonet Fire danger model was showing very high values throughout eastern Cleveland county up to the edge of the Cross-Timber woods.  Groves of dense Eastern Red Cedar mixed with completely dormant grass fields combined to make a volatile combination.

The Oklahoma Fire Danger Model explained here.  The arrow points to the Noble fire location.  Notice the sharp gradient to lower values (more containable fires) in eastern Cleveland County within the Cross Timbers.
The weather was typical for this summer of record breaking heat and dryness.  Temperatures were in the low 100's F and relative humidities were in the low 20 percentage range and south winds gusting over 20 kts.  SPC already issued a nearly critical fire danger outlook and the state of Oklahoma issued a burn ban. During the early afternoon the expected conditions materialized.

The SPC day 1 fire outlook.  Central Oklahoma's on the edge of a critical risk.

The fire danger was certainly high but not unprecedented in Oklahoma.  During the winter time we'll frequently see these conditions appear and we'll get rapidly moving fires.  But what set these summer conditions apart from the winter was the huge depth of the nearly dry adiabatic lapse rates, on this day, almost 3 km.
The 00 UTC Norman sounding shortly after the fire was put out.

These conditions we had were typically found in the western mountains where the airmass was of classic desert continental tropical origin, sometimes advected eastward driven by strong synoptic forcing into the Plains States following the dryline.  But now they were prevalent all throughout the southern Plains as we've become the new source of the continental tropical airmass through months of uninterrupted baking of the ground.  Time was the only factor needed for the continuous baking to cure huge swaths of vegetation into a combustable fuel which now included even parts of the Cross-Timbers.

Could fire tornadoes or even a firestorm have developed?

Anybody unfortunate to be ahead of a big fire would be justified in calling out that they experienced a firestorm.  However I hear amongst fire specialists that they refer to a firestorm to describe a certain extreme behavior.  That is a firestorm generates a velocity structure that helps it to intensify in a positive feedback loop.  Enhanced inflow at low-levels feeds the fire Oxygen from ahead and the lateral flanks.  Vortices forming along the flanks also help to concentrate the heat at the head (downwind) end of the fire.  The updraft plume is typically deep and separated from the ground in a single column consisting of short-period pulses.  This updraft structure is effective at not just generating the internal fire-induced circulations but also launching embers in typically erratic directions but often well ahead of the fire.  These types of fires are said to be plume dominated as opposed to the wind-dominated fires that we see in central Oklahoma during a pre-dryline strong south wind events of the cool season.  Wind dominated updraft plumes tend to hug the ground.

The picture below shows the type of fire plume so reminiscent of an incipient firestorm we've heard and seen further west.  It erupted immediately off the ground in an updraft column that was roughly uninterrupted until it reached its LCL at nearly 3 km AGL as shown by the sounding above.  Fire induced updrafts often contain a bit more water vapor than implied by the sounding due to the combusted fuels and so the LCL may have been a bit lower.  Nevertheless, this 300 acre fire was capable of producing an upright updraft 3 km deep.  The video above even shows some anticylonic vorticity within the west side of the broad updraft as it interacted with the environmental shear just above ground.  I wouldn't be surprised if this fire modified the low-level flow creating a calm wake to the north and accelerated flow around the west and east flanks creating a low-level broad vortex pair as it was tilted by the updraft in a similar way to one mechanism by which a supercell forms.


A vertical fire induced updraft with pyrocumulus at the top.  

A little more intensity of the fire and we may have had significant fire vortices erupt out of the flanks of the fire to propagate downwind.  I may exaggerate but compare the pictures and video above and the time lapse by John Hart with the fire vortex simulations available on the Visualization and Enabling Technologies webpage of NCAR.  I chose one to highlight below that shows a fire with surface wind vectors, three-dimensional heating and vorticity.  The similarities are pretty striking.  



These vortices would be large, not the small ones sometimes visible within small flames.  For an example, see the fire induced vortex from a forest fire in eastern Colorado here.  The main fire updraft plume is to the left of the vortex.  However the vortex connects with the main plume aloft.  A very dynamically similar analog occurs with heated water induced updraft plume around the entry point of lava into the Pacific Ocean in the big island of Hawaii.  There are multiple videos online showing this such as here, here, and here.  Perhaps the intensity of the Noble fire is far short of that of the lava-induced plumes or the great fires out west but both heat sources had erect plumes separating from the ground.  

Were there other plume-driven fires during this outbreak that could've been better candidates to produce a firestorm with large fire vortices?  As it turned out, there was one candidate to the southwest of us in the Wichita Mountains.  The Meers, OK fire as it was called, started about the same time as the Noble fire and both plumes can be seen in the GOES-E visible imagery at 1930 UTC.  The Noble fire was at its peak and the Meers fire was just getting started perhaps both having burned similar areas.

The GOES-E visible imagery at 1930 UTC captured from the NCAR RAP website showing two fire-induced smoke plumes, one near Noble (upper right arrow) and another near Meers, OK (lower left arrow).
While the Noble fire was aggressively suppressed, the remote rough terrain around the Meers, OK fire possibly inhibited access by firefighters and the so the fire grew.  By 2302 UTC, the GOES visible imagery showed a classic wedge-shaped anvil with a singular strong updraft plume anchored on the south side.  There was even a hint of an overshooting top.  The size of this fire surely put the Noble fire to shame and yet the Noble fire also exhibited a similarly shaped smoke plume suggesting a plume-driven fire. Even more striking was the obvious anvil-layer divergence signature at the top of the fire.


The GOES-E visible image from 2302 UTC showing the anvil and intense updraft plume from the Meers, OK fire.  Two other fires to the south in TX exhibited much weaker updraft behavior.

The KTLX 2250 UTC 0.5 deg scan of the Meers, OK fire.  On the left is the reflectivity and the right, radial base velocity.  The green and red arrows highlights where you can see the anvil-layer divergence over the top of the smoke plume.

The Meers, OK fire grew to over 20,000 acres as opposed to the 380 acre spread of the Noble fire as reported by news9.com.  At any one time the Meers, OK fire was likely much more intense too.  The radar data from the Noble fire did not indicate this kind of diverging anvil and its updraft was much weaker.  So if the Noble fire plume exhibited signs of a vortex pair, I would expect that the Meers, OK fire was much more capable of producing intense vortices.  However no radar could adequately sample the lower levels of the plume where the vortices would most likely reside.

Fire intensity and Cedars.

How did the ground conditions affect the fire northeast of the Noble Highschool?  An overhead view of the area near the fire initiation pretty much shows patches of dense Eastern Red Cedars.  These trees were so close together that their crowns were touching.  Considering how dry we've been, the fuel moisture level must've been incredibly low in these trees.  But these trees can become torches even without such dry conditions.  I also saw numerous trees whether or not they were burned, that showed branches right down next to the ground.  The stage was set where a ground fire (as it started considering that a ditch digger started the fire) could easily have spread into the crowns of the trees.  The video above and from real-time media clearly showed numerous crown fires.  I'm not sure if these crown fires started to spread independently of the ground fire based on what the video above showed.  Considering the sporadic crown eruptions, I suspect the crowns torching in lockstep with the advance of ground fire and not spreading on its own.  Nevertheless, where crown fires existed, the fires were intense and everything quickly burned as the picture below so eloquently shows.

I might add that in addition to the crown fires suggestive of a front, I could see that there were numerous spot fires ahead of the main fire also sending up their own plumes.  A prominent one showed up on the pictures above.  The nature of this plume-dominated fire probably contributed to producing these spot fires.

A Google Earth overhead view of the Noble Fire initiation area (represented by the red swath).  The inset shows a red arrow where the picture below was taken.


A picture of intense burning in Eastern Red Cedars to the east of Noble Highschool.  One of the Cedar trunks showed active flames within a hollow (at center).


Finally, I'm amazed at how fast this fire was quickly brought under control considering the hostile conditions.  There were two firefighters that suffered some burns, and another with a shoulder separation.  In addition, the blackhawk helicopter crew faced a hostile male when they lowered the bucket into a nearby pond.
A Blackhawk firefighting helicopter lifting another bucket of water from a pond to dump on the remains of the Noble fire.

References to read:

Fire whirls and vortices simulation

COMET's extreme fire behavior course
http://www.meted.ucar.edu/fire/s290/unit11/
this is part of a larger course on fire weather forecasting.

Oklahoma Fire Danger Model
http://agweather.mesonet.org/models/fire/description.html

WIldland Fire Assessment System
http://okfire.mesonet.org/public/?cat=links