Numerical Operation Model Verification...
"Historical Blizzard of December 2009"
(Courtesy of Americanwx.com)
A historical synoptic setup appeared evident nearly a week in advance of this storm system...
My long term section issued on December 12...
As I mentioned yesterday the east-based negative NAO will be retrograding towards a more favorable west-based negative NAO with above normal geopotential heights over the northern Atlantic including Greenland. With rex blocking forming towards the Hudson Bay in northern Canada, the 500mb jet pattern looks favorable going towards the week of Christmas. During my special Middle Atlantic Winter blog last winter I mentioned quite about the favorable trend for a negative to positive trending NAO causing large storms systems over the east coast. This trend may slightly be available around the 20-22 of the month. Another interesting teleconnection is the current negative AO anomaly which is nosing towards -3. This makes it one of the lowest AO anomalies in many years and I have seen statistics proving a negative AO this low in December would allow for a colder than normal winters as patterns similar to this are very difficult to budge under a winter-time regime. With favorable upstream blocking conditions and deep troughing over the east coast under a trough axis over the east coast, I think it is suffice to say the potential for a KU storm over the region is significantly higher than normal.
Well currently the AO has dipped today to the lowest on record for the month of December with -5.668. Link. The previous lowest on record occurred in 1976. Interestingly enough, once again the winter of 1976-1977 reoccurs on my list of analogs. Anyways with a record low negative AO, the NAO is also trending towards positive the classic acknowledgment of cyclogenesis along the east coast. What was evident early up the week in advance of the system was the impressive jet streak up the coast with a favorable trough axis allowing the cyclone to travel up the coast and/or near the 40/70 benchmark. The considering factor keeping me at bay for the week's model cycles was the PV position, which began to show trends of retrograding north and west. On December 16, I began to explore the close analog of February 1983 Blizzard...
This is a very bold statement considering my conservative stance usually... This has the potential to be a major east coast snowstorm or completely swing out to sea. There does not really appear to be a middle ground in this scenario. We really have everything we could ever ask for considering a Miller A creeping up the coast very slowly with a PV to the north and a high pressure with a decent positive PNA. But it all goes back to the phases of the shortwaves to our west and the location of the PV. Trends are encouraging, but I am not jumping the gun yet too much. I am surprised there are not more people tracking this as it is definitely the greatest threat for a major snowstorm we have seen in years even though guidance currently does not portray it. 0z NAM is a huge run in my opinion, we shall see trends, so I will post another modified update then. But I will be around this evening too track it too.
0z NAM on the 17th of December began to show the classic trend northwest. Classic KU storms always trended northwest within 24-72hr range, especially PDII. 12z ECMWF on the 17th also edged significantly towards a major snowstorm...
After these few model runs, everyone began to jump on the bandwagon of a historical snowstorm that would slowly climb the coast courtesy of anomalous blocking to the north. In this research blog I will explore the mesoscale features that allowed for unique banding and localized higher snow totals, also the second half of the blog will take a look at model verification.
Mesoscale Banding vs. Dry Air (Virga)...
Mesoscale bands are unique to significant storms along the East Coast in particular due to unique moisture 700mb transfers from physical features such as the Gulf Stream from the Atlantic Ocean and the Gulf of Mexico along with the Great Lakes to the northwest. Mesoscale banding is usually associated with localized features such as surface fronts and inverted troughs. Surface fronts are typically responsible for the mesoscale banding present in KU storms and are usually located along isolated baroclinic zones. They are anomalous easterly 700mb moisture transfers associated with warmer air and PWATs ascending over the northeast flow with a 1030mb+ high to north at the surface to over 800mb aloft. Surface fronts typically run from either south to north or southwest to northeast allowing bands to remain stationary or pivot slightly northwest. These convergence zones allow for heavy precipitation totals and can usually be discerned not only by NEXRAD radar, but also by pressure falls of near 15hPa/3hr. Low level warm air advection and frontogenesis can also intensify these precipitation rates.
Around 4pm, radar composites evidently showed several of these unique mesoscale bands across the Middle Atlantic localizing snowfall rates of up to 3in/1hr. As noted above, a gravity wave is shown, which are very common in classic significant winter storms. These gravity waves can enhance precipitation rates north of the warm front near the interaction of the troposphere jet streak. The gravity waves appear as bands of convective activity allowing also for unusual cloud formations and pressure fluctuations. Unusual wind direction aloft interaction and combined rapid surface pressure troughs can be a catalyst for gravity waves in the more significant of winter storms. Gravity waves typically propagate northward. In this storm in particular the noted gravity wave caused an enhanced series of bands from Richmond to Washington DC with TSSN being reported in Richmond, VA.
Another interesting mesoscale feature of note in this historic blizzard, was the convergence zone around an inverted trough towards the New Jersey, northern Delaware, and Philadelphia region.
This inverted trough formed in response to the approach upper level trough and axis of lower surface pressure falls north of the main cyclongenesis towards the southern Delmarva. This allowed for enhanced mesoscale banding and higher QPF. Several models including the ancient MM5 and modern high resolution RUC and ARW were able to pick up on this feature a bit in advance...
(Courtesy of Americanwx.com)
Current operation models, especially global models do not have high enough resolution to pick up on the mesoscale features as listed above. This poses forecasting headaches for pinpointing where the enhanced snow totals will occur. Other synoptical forecast problems exist with precipitation gradients on the north and northwest sides of systems...
Two highly noticeable snowfall gradients occurred in this system with one locally noted around the Pennsylvania I-76 turnpike and the other occurring about 25mi south of I-80. The first cutoff was between the mesoscale banding of a surface front in northern Maryland and southern Pennsylvania allowing for heavy snow in the banding region, but only moderate totals a few miles to the north. This is evident in just one quick example of a 15mi difference in Lancaster County, Pennsylvania...
LITITZ 9.0 457 PM 12/19 SPOTTER
QUARRYVILLE 20.0 724 PM 12/19 SPOTTER
Northern snowfall boundaries exist courtesy of sharp, rising western boundaries of cold conveyor belts with high pressure allows for strong confluence. This was highly evident this was going to exist in this scenario in response to the Polar Vortex located over upstate New York and southern Quebec. As will be noted below several higher resolution models were able to detect the sharp cutoff to the north. Also, while trying to pinpoint snow total forecasts, I took the stance of a sharp-cutoff to the north, which paid out to be highly accurate. The only issue with my snow map below was the detection of the surface front mesoscale banding over Franklin, Adams, Lancaster, York, Chester, Philadelphia, and Delaware County. I though, was able to note the inverted trough band over Montgomery and Berks County. In general I am very pleased with my official forecast map for Pennsylvania. Here are the plotted snow totals vs. my forecast...
Also important to note are several upper level features that help distinct the location of the heavy snow axis. These old rules of thumb were actually used quite heavily before the plethora of models we have now. What is interesting to note is several model runs including the NAM and GFS suffered convective feedback problems earlier in the week, but using these techniques I was able to see the surface guidance was likely progged wrong allowing for the heavy snow axis to be farther north and west. The presence and position of 500mb chart features is critical to the location of the surface low track, 850hPa low track, jet streak location, maximum vorticity, and QPF location.
The H5 chart above details December 19 and the location of several of these important factors. 300mb and 500mb charts depict winds aloft including the jet stream allowing for us to be able to see the position of troughing, ridging, etc. First what must be present is a split flow with a dominate subtropical jet allowing for the formation of the low, and a polar jet keeping arctic air on the north side of the United States into Canada. The subtropical jet is typically the catalyst for development allowing for cyclogenesis either in the Gulf of Mexico for Miller A systems or near the southern North Carolina coast for Miller B systems. In this instance it occurred in the Gulf of Mexico deepening to near 1000mb crossing the Florida panhandle. Next on the checklist is for the presence of a shortwave embedded in the flow allowing for an increase in amplitude in the trough. In this instance you can note above is was located in Missouri. Looking at analogs a similar S/W placement occurred for the nor'easter of 19 January 1987 also located in southern Missouri. In any case this impulse is important as it spawns the primary low over the Gulf of Mexico or Atlantic Ocean and allows for increase wavelength and amplitude of the trough, hence the term amplifying trough. Also important is the presence of a closed low in either Newfoundland or across the Canadian Maritimes. This acts as an anticyclone allowing cold air advection to continue to filter down the east coast in a conveyor belt. In this instance above you can note it was located near Nova Scotia.
Also of importance is a building negative EPO ridge/positive PNA. This helps to increase wavelength and amplitude of the trough. Also typically of importance is the presence of a cutoff low in the desert southwest, but it was absent in this storm very similar to the zonal flow in the southwest during February of 1983. The term jet streak is also of importance as it allows for a higher distribution of maximum vorticity, maximum confluence, and allow for left exit regions of the cyclone. Heavy snow corridors are always confined parallel or to the west of the jet streak. The jet streak is also noticed as acting as a low level thermal field coinciding with heavy snowfall patterns along the corridor. As the trough begins to turn negatively tilted with jet stream winds near 100knots up the coastal plain (jet streak), the vorticity maximum is noted typically tracking through the Ohio Valley up through southern Virginia advecting off towards the Delmarva. In this instance it traveled a bit farther northward towards Washington DC up through central Delaware to located just south of Long Island. The location of the 500hPA vorticity is important as far as precipitation types go with locations to the south by 50mi or so likely experience precipitation changeovers from snow to rain or sleet/freezing rain. But locations from 50mi south of vortex on northward usually allow for snowfall. This was evident in the December 19th event as location about 25mi changed to rain briefly and then back to snow. Also of importance and noted above is the position of a cutoff trough over southern Canada towards Ontario. As you can see in that 500mb chart, there is an anticyclone noted just north of Lake Superior. This blocking allows for confluence over the Northeast, which acts as cold air advection in to the II quadrant of the cyclone.
Of likely most significant importance is the track of the 850hPa low...
As noted above for December 19, this tracked south of Virginia across northern North Carolina and tracked near the 40/70 benchmark. Tracks of 850hPa lows typically track east to northeast along the Middle Atlantic to New England coast allowing for ideal snow forecasting methods. December 19th allowed for geopotential heights to indicate a strengthening 850hPa low tracking as followed above. In most instances the heaviest snowfall occurs 50km-300km to the north or northeast of the 850hPa low track. This corresponds to in this instance the heavy snow falling over southern Pennsylvania to as far south as northern Virginia correlating to the I-95 corridor on eastward being impacted with the heaviest snow up the coast. The track of the 850hPa low was just too far eastward for areas over central Pennsylvania to southern New York State up through western Massachusetts to get impressive snowfall totals.
(Courtesy of Americanwx.com)
Using these techniques I was able to make a generalization forecast without really looking at surface numerical model maps such as QPF fields. These tools are general rules of thumb including the above mesoscale section of synoptic forecasting. So now that we looked at the forecasting w/o model aspect, let’s look at model verification...
Numerical Model Verification...
Model verification is highly important during the winter season in future synoptic precipitation events. Being able to pinpoint common biases will be beneficial in the long term. All model maps are courtesy of the archives located at Americanwx.com...
GFS at 84hrs...
Noted above are several errors in the surface map. First the location of the closed low in Canada was slightly progged too far east and needed to be reassessed farther towards the west coast of Nova Scotia. The idea of the precipitation sharp cutoff in quadrant II of the storm was correct in this cycle all the location was several hundred miles too far east. Important to note at this time frame is a southeast bias in the GFS. At first talk in the meteorological community was the typical northwest trend was occurring because of the Southeast ridge during the previous -SO years, but with the absence of the La Nina and Southeast ridge, we continue to see a correction of the GFS to the northwest in this time frame. What is critical to note is the abnormal use of the GFS. Most local meteorologists use the GFS almost 80% for their forecasting, this includes other weather agencies. So the reason for the lack of jumping on board for this storm for most was due to the GFS track offshore. This is interesting as in relatively the GFS has some of the worst verification scores out of the global models, but due to its easy maps to read and readily availability, it makes it very popular.
GGEM at 126hrs...
Also at 3-5 days out the GGEM had the surface low relatively far east tracking outside of the 40/70 benchmark. But interesting to note is the position of several inverted trough features over New Jersey which actually verified closer to the event. In general this GGEM run did a much better job at the medium range forecasting this system.
NAM at 78hrs...
The NAM was much more interesting at this ranging progging the low to be a good 50-100mi closer to the coast to the GFS. But using the rule of thumb techniques mentioned above, I was able to pinpoint that the NAM was suffering surface convective feedback problems allowing for the sharp-precipitation cutoff line near the Mason-Dixon Line and for the odd left exit region the cyclone took.
Closer towards the short term forecasting aspect of storms systems such as this, it is important to make use of several higher resolution models, one of which performed amazingly accurate during this storm.
SREF at 60hrs...
Despite the NWS offices infatuation with the Short Range Ensemble Forecasts, I have never quite been a fan of the mean resolution run. I find several individual ensembles are usually too extreme or opposite ends of the spectrum skewing the final results. In any case the SREF QPF appeared a bit too low in this time frame.
NAM at 27hrs...
One of the most accurate models once again was the NAM which scored highly in verification. While the sharp northern precipitation cutoff was about 25mi too far south, it did note several important mesoscale features. One of them included the abnormally dry dew points over the southern Poconos into the Lehigh Valley and Schuylkill Valley. Also it noted the surface front frontogenic forced mesoscale band over northern Maryland with QPF totals near 2in. Interestingly enough the NAM is often regarded as a wet biased model, but it performed much better in the QPF shield and amounts than the GFS did.
ETA at 51hrs...
Surprisingly enough one of the best models at this range was the ETA, which is quite unusual since the NCEP retired this model years ago. But its QPF placement was really spot on along with several mesoscale features being noted actually verifying. The ETA is the model often regarded as forecasting some of the strongest storms of the century including the 1993 Superstorm, PDII, and the Blizzard of 1996.
GFS at 6hrs...
One of the worst models even at 6hrs out was the GFS. It did a terrible job forecasting the low track, which verified closer to the coast and it also had the QPF precipitation shield well too far northwest. Again interesting to note is the GFS as being used by so many meteorological services, that is one reason the NWS issued warnings for locations such as Williamsport, PA up the Poughkeepsie, NY in which those locations saw T-1in. Lesson learned from the past three winter events I think is to stray away from the GFS, although the local meteorologists will likely never do so.
HIRES ARW at 48hrs...
The most accurate model progging of this storm verified at the 48hr forecast from the HIRES ARW run by the NCEP. Interestingly enough its brother the HIRES NMM did not verify so well, but the ARW did a wonderful job at QPF verification especially over northern Maryland through southern Pennsylvania and southern New Jersey. It also noted the sharp-cutoff just south of the I-80 corridor.
In conclusion once again in general the NAM scored the best verification out through 84hrs with only a few slightly convective feedback problems along the way through the model run cycles. Long range models, the GGEM scored the highest for medium range followed by the ECMWF, with the good ol' American GFS coming in last. Keep in mind supposedly the GFS just went under a few upgrades. Anyways as far as high resolution models, the HIRES ARW performed the highest with near spot on QPF totals and surface low track. My goal through this research blog was to expand one's horizon above just numerical model forecasting, but to look at a synoptical approach. When I analyze my forecasting methods, I do not just use what the GFS 850mb Temp/Heights/QPF looks like, but I verify to see if it looks accurate compared to the progged H5 and H3 charts. I try to use a blend of guidance to come up with a reasonable solution, with my reasonable solution typical based of some sort of analog.
(Courtesy of Americanwx.com)
"Index of /winter0910/storms/Dec1909." New York MetroWeather. Web. 22 Dec. 2009.
Kocin, P. J., and L. W. Uccellini, 2004: Northeast Snowstorms, Volume I: Overview. Meteor. Monogr., Vol. 32, No. 54, Amer. Meteor. Soc., 1-296. Print.
"Here northeast of Harrisburg 2009-2010 winter statistics"
Current Snow Cover- 1.50in
Monthly Total- 13.0in
Seasonal Total- 13.0in
October Total- 0.0in
November Total- Trace
December Total 13.0in
Winter Weather Advisories- 3
Winter Storm Warnings- 1
Ice Storm Warnings- 0
Blizzard Warnings- 0
Freezing Rain Advisories- 1
Winter Storm Watches- 1
Lowest High Temperature- 28.0F
Lowest Low Temperature- 14.7F
Wind Chill Advisories- 0
Wind Chill Warnings- 0
(Winter Storms Stats)
Dec 5 - 1.5in - First accumulating snow of season
Dec 8-9 - 2.5in - Snow changed to plain rain
Dec 13 - .1in - Freezing rain
Dec 19 - 9.0in - Heavy snow, higher amounts to south
Update 4:30pm 29 December 2009...
Good afternoon!!! In a few days I will be approaching my third year on Wunderground. When I approached this site in December of 2007, I expected creating a blog only forecasting basically my opinions for the Harrisburg geographic region. To some it may be a surprise, my age 17, or perhaps not in some regard. I kept my age hidden in part because of the dangers of the internet, but also the reaction. How in the world would adults respect a teenager forecasting the weather and telling them information? Knowing the acceptance level across the meteorological community for teenagers being regarded as weather weenies, I was apprehensive and kept this minuscule fact hidden. Upon the creation of my blog, I was going to go above and beyond a mediocre performance.
My passion for weather is correlated back to a very young age and my interest consumes not just the actual tangible thunderstorm presence, but for the "physics" behind the mechanics of the atmosphere. I built a foundation in general with the presence of books, in my meteorological library. But as I advanced in books from "The Weather Channel Basic Comprehension Weather Book" I moved into meteorological textbooks. This correlated with my current curriculum in high school of college level honors Calculus and Physics allowed me to bridge the transition from pure weather enthusiasts and weenie to actually understanding the process of lets say "the reason driving a car on a clear night will allow for violent temperature fluctuations driving over hills and into valleys." Having pursued personal interviews with meteorologists, taking online courses, visiting different meteorological communities, and using the resources of the 21st century; I grew as a young adult. The sails of my ship were visible at a young age, which is for the definite pursuance of meteorology and this goal has not left the tip of my eyes ever since.
Information in my actual physical blog and comments have always been and will always be of my own. While I learn from others, as asking a question is key for education, I believe in credit given where it is due. Being apprehensive about being accepted on here was critical and has always been. Understanding perfectly, I can see adults being unable to accept me for being anything but a weather weenie. While yes I love the idea of monstrosity of a blizzard affecting my region; I also enjoy the physics behind the due process of the low pressure bombogenesis. All I am asking is really for no changes on the blog as really nothing is changing other than this announcement. I have been on this blog for three years and I have gained a solid reputation and a plethora of experience. Yes I love vegetable gardening, photography, PennDOT jokes, etc... I am what you would call perhaps unusual for one's age, hahaha. But my peers, family, associates, community, have continued to always encourage my passion for the science of meteorology. There is a cliche saying "you learn something new every day," this is true more than ever in the process of education. It is a constant transition and/or push or pull between educating others with adults teaching students, and students teaching adults.
In general I ask this, approach this with an open mind and not with a biased opinion and conclusion. Wunderground remains a mature and respecting community, so expect nothing to change here. While I may come off a bit conceded or mellow dramatic, this is all to prove my point that my blog is all collaboration of my knowledge and not that of others in some twisted form of plagiarism. One can consume as much education as the next if you put your mind to it. So go about this in strife as we have been the last three years here on Wunderground. Thank you to all in support of my blog over the years.
Here is to another few years!!!