Ph.D. Student - Earth System Science (UC Irvine), B.Sc. - Atmospheric Sciences (Cornell University)
By: Zachary Labe , 9:49 PM GMT on February 11, 2014
Posted: 11 February 2014
Another busy weather period is likely over the next 48 hours as a strong low pressure moves up the east coast. It's Gulf of Mexico origins will allow it to bring significant moisture and bring winter weather impacts from Florida to Maine. The forecast, however, is not overly difficult. The synoptic setup is pretty straight forward. For those who own Paul Kocin's, "Northeast Snowstorms Volume 1/2" books, this is very characteristic for a Miller A system. A track along the immediate coast will focus in heavy snows just northwest of the I-95 corridor with warmer conditions closer to the coast. High winds under a tight pressure gradient are likely in addition to increasing surf. Lack of upstream blocking, we keep this storm from achieving its full potential, so nothing overly historic in this regard. Highest storm total snowfall amounts will be limited to mostly under 20" due to the rapid movement of this system.
(Courtesy of the WPC)
This blog will take a different approach to my usual forecasts. It will instead look at more of the scientific reasons that certain features of this event will occur with meteorological and mathematical support. I will provide a snow map during the day Wednesday and post it on the blog.
Diving into the meteorological goodness of modeling, there are some clear concerns for those looking for a consensus.
The above images are the GFS and ECMWF respectively for sea level pressure (SLP) at 48 hours on the 12z, 11 February 2014 run. We can already see some differences in track. It is important to note that while the differences seem minor in a graphical representation, they can have major impacts on a forecast for a corridor that contains a large majority of the US population.
Below is a pretty interesting tool using NCEP verification charts...
It is important to note several things about this chart. First, we can see that temperatures have been running several anomalies too cold for H5 temperatures. This is confirmed by both the NAM and GFS. We also see that in general there is no overwhelmingly accurate choice between the GFS and NAM for 500mb analysis. Recently, I have seen many inaccurate statements regarding biases on the computer models. For a full list of observed biases on the computer models see this Link. Thermal profiles have actually been doing fairly well this year on the GFS/NAM. The recent upgrade to the GFS has eliminated its old biases of the typical unamplified southeast low track we used to find for nor'easters. It's cold bias has also been eliminated. The biggest concern for this event would be over-estimation of QPF, which is a readily observed bias on the GFS and NAM. This is confirmed by several ground reports cases this winter. The latest statements from the NCEP confirm that the GGEM is the western outlier and the NAM is the eastern outlier. While they cannot be discarded entirely, it is usually a pretty good idea to eliminate their use for the majority of a forecast.
Most of the upper air data sites across the Pacific have been heavily and completed sounded for this event. Only minor nuisances exist between the computer models, and most of them are due to discrepancies in different parameterizations and algorithms. It is also important to note that this storm setup is significantly different than the common Miller B systems we have been seeing the last few winters. In our setup, we have a moisture-fed low pressure developing in the Gulf of Mexico and moving northeast. Cold air is also lacking in this event. A retreating 1032mb high pressure will be quickly scooting off the coast of Maine. This is definitely a less than ideal position for widespread I-95 snows. Also across the Great Lakes, we have an interesting feature. A potent upper level low (ULL) is swinging through Michigan. This will act to 'kick' the storm off to the northeast and prevent it from moving too far west. It will also act to increase a confluent flow of dry air creating a sharp western precipitation gradient. Surprise. Surprise.
As mentioned earlier, the synoptic setup is pretty simple. The main features have already been described. It is a matter of the exact track of this system that will play a significant role in impacts. Multiple forecasting rules of thumb can be utilized in this case to go above and beyond model guidance.
Ice crystal formation is a result of several processes: deposition, mixed-phasing, riming, and ice nucleation. The differences in the processes are often a result of thermal boundaries, aerosols, and moisture gradients in the clouds. As ice crystals form, they begin to fall through a vertical column of the atmosphere. To look at that end result of these ice crystals, we must look at a vertical profile of the atmosphere. On a large-scale, a good rule of thumb is to look at this infamous 1000-500mb thickness level. We always hear about the 540dm line being magical for a rain/snow line, but however it is very inaccurate.
Mathematically, a thickness layer is proportional to the mean virtual temperature over two heights. In this case that is being 1000mb (assumed surface of Earth) and 500mb (approximately 18000ft). To provide a little math goodness to this blog, we can estimate a location thickness using the hypsometric equation:
Z2-Z1 = ((Rd*T)/g)*ln(P1/P2)
Z2-Z1 is the thickness in meters
Rd = dry air gas constant of 287
T = virtual mean temperature (kelvins)
P1 = pressure of 1000mb
P2 = pressure of 500mb
Note that the virtual temperature is the theoretical temperature a dry air parcel would have to total pressure and density equal to that of a moist air parcel. Anyways, when you run the actual numbers, you can see that freezing is around 540dm. That is how you get the actual derivation for thickness calculations. You find the thickness over any two pressure levels along with the mean virtual temperature. This does not mean a whole heck of a lot for forecasting precipitation types because of one key idea. That thickness utilizes an average of the temperature over 18000ft. We clearly know from analyzing vertical profiles, that there are temperature nuisances in that vertical approach with possible warm layers.
It is necessary to see if there are any above freezing layers over that thickness. Using balloon soundings, if any layers are +3C or warmer, than complete melting of the snowflake will occur. Also if the warm layer is less than 750m thick, it is likely snow will occur. The depth of the warm layer is the separation between precipitation types.
So lets take this and apply it to this forecast. I am utilizing the 4km NAM only due to its higher resolution maps.
The upper left panel includes surface temperatures, while the upper right is a temperature at 850mb (~5000ft). The temperature at H85 is not a thickness, but a specific temperature at one height. It is another tool in forecasting rain/snow types. The 0C lines are the bolded, colored lines in both panels. It is generally assumed using forecasting rule of thumbs that the rain/snow line is around 15-30 miles or so to the north of the H85 0C line. This is not always the case, but in our setup it should work. Analyzing Skew T plots will also help us determine where the rain/snow/ice lines will be. It is the actual vertical slice of the atmosphere.
For our purposes just focus on the horizontal axis. Those are temperatures at the surface (~1000mb). Follow the black contours that are straight at an angle going to the right of the plot. The right red line is the actual air temperature. The left black dashed line is the dew point temperature. The above Skew T is for NYC at the height of the storm. We can see that there is a layer from 2500-5000ft above freezing. That is right around our melting threshold of 750mb (2500ft). Despite a H85 right around 0C and 1000-500mb thickness of sub 540dm, we would think it would be a snow profile. But looking at the actual temperatures aloft, we can see that the snow will melt and produce sleet at the surface.
Of course we can have atmospheric dynamic and mesoscale processes override this concept, but that is getting a bit too detailed for our purposes. I want to show that there is quite a bit beyond looking at a few images produced by graphical operational and ensemble modeling. It requires further analysis.
We can eliminate forecast models, once we have established a basis idea. A couple of rules of thumb I have been utilizing for this event include...
1) Heaviest axis of snows are usually 50-100mb north and west of the 850hPa low track. This works pretty well, especially for nor'easters where cold-conveyor belt/slanted convection (CCB) bands are fairly common. Despite, what model QPF predictions show, I always utilize this idea.
2) Forecasting western precipitation gradients is probably, in my opinion, the most difficult aspect of these events. I have been utilizing a new rule of thumb that works pretty well... Find the model consensus 0.5" QPF line. Expect that line to be the no snow/snow line. In sharp ~5" or more versus 0" cutoffs above small distances, this works pretty well.
3) Nor'easters, Miller A in particular, often feature very little in the way of freezing rain due to their thermal profiles. There can be an area of sleet, but freezing rain is unlikely at these latitudes. Also in our case for this event, an unfavorable high system to the north will keep these probabilities low. It will mainly be a sharp rain/snow line.
4) Find the axis of most impressive frontogenesis, omega, and vertical vorticity (again usually 50-100mi northwest of the H85 low). This axis will see the most mesoscale banding. Snow ratios will also be most impressive in this area.
5) Snow ratios are generally low for systems out of the Gulf of Mexico due to the deep saturated layer aloft. Maximum snow ratios will only be 12:1 or so. And given the antecedent air mass is pretty stale, it looks like wet snow bomb is in order.
6) Note the area of sharpest SST gradient between shallow land waters and deeper waters. This is a natural area of baroclinicity where a low in this type of setup is most likely to track. It is often very close to the coast, if not even farther inland.
7) Back-end deformation snows are always forecast, but never come to fruition. This is often courtesy of a drying, downsloping mechanism that eliminates this QPF. I expect no back-end snow in this event.
8) The heaviest snow axis usually falls further northwest than model guidance shows. This is an observed bias by the WPC and NCEP that needs to be taking into account when making a forecast.
9) Dynamically induced snowfall is always a threat despite what may seem like a rain profile. This area is difficult to predict, but occurs in the axis of highest frontogenesis on the 'warm' side of the storm.
10) Lack of deep mixing of the atmosphere usually prevents surface winds from reaching full potential for inland areas. The coast is an exception to this rule.
So what does all of this mean...
Essentially, a strong low pressure will track from eastern North Carolina and parallel the coast. It will likely track north-northeast before turning northeast around the mouth of the Chesapeake. It will track inside the 40/70 benchmark. Excessive moisture in a large precipitation shield will overspread the region Wednesday night through Thursday.
Precipitation will start as snow for all areas with surface temperatures in the lower 30s. Current 2m T prognostics keep surface temperatures around 30F even for the far northwest locations during the majority of the event. As the low center tracks along New Jersey, warm southeasterly winds will allow precipitation to change to a heavy rain. Sleet will also be around 30 miles ahead of this line. Given climatological forecasts and statistical methods, I am expecting this mix line to push into the I-95 corridor. The duration is a bit uncertain. However, I expect all I-95 metropolitan regions to receive at least winter storm warning criteria snows of 6" or more before changing to any mixed precipitation. Coastal areas will see significantly less.
Heaviest snows will fall from along a line of Frederick, MD - Lancaster, PA - Allentown, PA - Morristown, NJ - Newburgh, NY - Worcester, MA - Concord NH - Bangor, ME. Accumulations may reach 12 inches or more along this axis. As mentioned earlier, due to the impressive speed of this storm system, accumulations of more than 20" are not overly likely. Or at least not widespread.
Wind gusts of 40mph are likely along the coast with high surf. Heavy rain is expected in parts of the Delmarva and New Jersey in excess of 1-2 inches.
Selected City Accumulations for the Northeast...
Hagerstown, MD- 14-18"
Baltimore, MD- 9-14"
Salisbury, MD- 2-4"
Washington, DC- 7-13"
Wilmington, DE- 4-8"
Dover, DE- 3-6"
Pittsburgh, PA- 0-3"
State College PA- 2-4"
Williamsport, PA- 2-4"
Altoona, PA- 4-8"
Harrisburg, PA- 8-12"
Lancaster, PA- 14-18"
Philadelphia, PA- 5-10"
Allentown, PA- 12-16"
Scranton, PA- 5-10"
Trenton, NJ- 5-10"
New York City, NY- 5-10"
Poughkeepsie, NY- 12-16"
Binghamton, NY- 4-8"
Ithaca, NY- 1-3"
Albany, NY- 8-12"
Hartford, CT- 6-10"
Concord, NH- 10-14"
Worcester, MA- 7-12"
Boston, MA- 4-7"
Nantucket, MA- 0-1"
Hyannis, MA- 0-3"
Burlington, VT- 5-9"
Portland, ME- 5-10"
Bangor, ME- 10-14"
There is a storm system of interest for the upcoming weekend, but I am going to jump ahead into the longer term. It is pretty evident based on latest guidance and global wavelengths that a warm temperature pattern will be impacting the Northeast during the end of February. The anomaly or length of this above normal temperature period is uncertain and will greatly affect whether flooding and ice jams are a threat. A Great Lakes cutter with heavy synoptic rains and an anomalous warm front would certainly cause major problems over the Northeast. The 10 February 2014 0z ECWMF shows this exact setup around days 8/9. The GFS ensembles are not as supportive and temper this warm spell for only a five day period or so. The EPO is heading positive along with other teleconnections in favor of upper level ridging over the east. Warmer weather is definitely coming. As for how long, there are varying signs... Given past analogous cold weather winters this can often be a reloading period for some more snow/cold to impact the region in March. See 1994. I am currently hedging toward this type of setup, but many questions remain. In any case, I am expecting these next seven days to be the final cold/snow in this ongoing spell since mid January.
Follow my 24hr forecasts on Twitter... Link and Facebook... Link.
Lower Susquehanna Valley Doppler...
(Courtesy of WGAL)
"10mi northeast of Harrisburg 2013-2014 Winter Statistics"
Monthly Total (October)- 0.0"
Monthly Total (November)- Dusting
Monthly Total (December)- 9.6"
Monthly Total (January)- 10.3"
Monthly Total (February)- 25.0"
Seasonal Total- 44.9"
(Snow Storms Stats)
Trace - November 8 - First trace of snow - Lake effect snow shower
Dusting - November 12 - First snow on the ground - Anafront
1.5" - December 8 - First inch of snow - WAA double barrel low
4.3" - December 14 - Miller B - Changed to freezing rain/sleet
1.3" - December 17 - Alberta Clipper
2.0" - December 26 - Surprise squall/clipper
4.8" - January 2-3 - Miller B Coastal
1.5" - January 10 - SWFE all snow
3.1" - January 21 - Redeveloping clipper with heavy snow along I-95
6.0" - February 3 - Wet snow from coastal low
1.5" - February 5 - All sleet accumulation with 0.3" of freezing rain
1.5" - February 9 - Light snow with Alberta Clipper
10.5" - February 12-13 - Nor'easter
3.5" - February 15 - Alberta Clipper redevelopment
2.0" - February 18 - Clipper/snow squalls
Winter Weather Advisories- 11
Winter Storm Warnings- 4
Ice Storm Warnings- 0
Blizzard Warnings- 0
Freezing Rain Advisories- 3
Winter Storm Watches- 5
Lowest High Temperature- 9.6F on 1/7/2014
Lowest Low Temperature- -3.1F on 1/7/2014
Wind Chill Advisories- 3
Wind Chill Warnings- 0
(Cornell University (950ft elev.) Snow Stats)
Monthly Total (October)- 0.0"
Monthly Total (November)- 3.7"
Monthly Total (December)- 16.4"
Monthly Total (January)- 18.5"
Monthly Total (February)- 19.0"
Seasonal Total- 57.6"
Comments will take a few seconds to appear.