North American Mesoscale Model Verification on January 26-27...
The North American Mesoscale Model (NAM) is a high resolution model courtesy of the National Center for Environmental Prediction. Another very common and referenced name for the model is the Weather Research and Forecasting Model (WRF). They are the same model and run out to 84 hours. The purpose of the model is to run a higher resolution determining mesoscale features that often cannot be picked up by the global models due to their larger scale. The NAM can be run on an 80km resolution with a broad view of the United States, or 40km which allows the viewer to zoom in on a localized region to put up on mesoscale features. Also a very high resolution form of the NAM exists using a 12km parametric and can zoom into different states. The model is released four times a day at 0z, 6z, 12z, and 18z. Keep in mind that time is zulu or more commonly known as greenwich mean time.
Computer models are critical to forecasting the weather as they use parametric and other mathematical equations to derive the current and predicted state of the atmosphere using a physical and chemical explanation as the background for the predictions. The NCEP is associated with the National Oceanic Atmospheric Administration (NOAA) to be the developer of these computer models. Each meteorological organization worlwide uses their own developed computer models to serve as a basis for the forecasts. Without computer models, we are not able to derive predictions on the weather greater than 24 hours out.
Like all computer models, the NAM shows all layers of the atmosphere in maps determining positioning and speed of the jet stream, surface precipitation amounts (quantitive precipitation forecast QPF), temperatures throughout the entire core of the atmosphere, convective indices, simulated radars, etc. Due to the higher resolution of the NAM, several problems do exist. It often produces convective elements in many middle latitude cyclones and these are known as convective feedback problems. This is where it developed these unstable regions and often associates them as surface lows causing the surface depiction to be distorted. This was a problem in the recent prediction of the January 26-27 winter storm and will be noted below. Also the NAM has a tendency to produce higher than normal precipitation amounts. I typically find myself cutting totals nearly by 30% as shown on the NAM. Finally the NAM has a bias to overamplify a low pressure and close them off to early particularily on eastern United States cyclogenesis off the coast.
The January 26-27 winter storm featured many model headaches do to some interesting variables. Early on it appeared likely for a low pressure to track up inland along the coastal plain. There was a lacking 50/50 low, unfavorable position of the western trough axis, neutral NAO, natural baroclinicity along the coastline, and stale antecedent cold air mass. This would produce rain along I-95 with heavy snows inland. The 500mb synoptic pattern showed very strong signals for this time of setup with most computer model guidance also in support. The GGEM and ECMWF led the pack with the heavy snows from I-81 on westward. The GFS suffered major problems with varying solutions for each run. But then the computer models began to delay the storm. In fact it was delayed nearly 48-60 hours from the original starting time. This caused a different scenario to unfold. The lacking high pressure to the north was still a problem as the anticyclone zoomed east-northeast, but now a high pressure and associated shortwave out ahead of it allowed sunk east-southeast across the Midwest and western Great Lakes. This acted as a 'kicker' helping to push the cyclonegenesis farther off the coast. This in turn allowed for a colder scenario along with precipitation to occur farther east. Therefore a turn of events allowed for a major I-95 snowstorm with 6in+ totals from Washington DC to Boston. This was very fortunate for snow lovers in that corridor due to pure luck given the poor synoptic setup.
Given the NAM's high resolution, it often overanalyzes prognostics post hour 60. It tends to enhance QPF, overamplify lows, and pick up on mesoscale features that really do not exist. Therefore I typically throw out hours 60-84. It would be a rare event where you would find the NAM 84 hour verifying anywhere close to accurate. But in the near term range, the NAM does an excellent job locating temperature thermals, QPF ranges, and picking up on mesoscale features; coastal fronts, enhanced convection, deformation bands, etc. But in this recent storm, it suffered a plethora of problems and even the 6 and 12 hour surface maps had poor verification especially in the QPF department.
Let us first look at the actual accumulated precipitation totals for the storm.
Given this is a 24 hour accumulated precipitation amount, about .01-.1in of additional precipitation fell south of the Mason-Dixon line in the previous 24 hours.
Here are the preceeding NAM total QPF forecasts...
(January 24; 18z) (January 25; 6z)
(January 25; 12z) (January 25; 18z)
(January 26; 0z) (January 26; 6z)
As you can see, the NAM had a lot of variance with the northwestern sharp precipitation gradients. These tight gradients this year have been caused by the rapid intensification of the coastal lows allowing the heaviest moisture to be confined closer to the center of circulation. Also in this instance, a very cold and dry air mass along with associated cold front was quickly advancing southeast across the Great Lakes and was even picked up on by the 700mb RH charts. This allowed the flow out of the northwest to dry up some moisture for areas more inland.
The NAM simulated radar vs. the forecast QPF did not match up. Often the NAM simulated radar showed the heavier mesoscale bands lining up in northern Maryland and southern Pennsylvania where as it only showed total QPF to be .25in-.5in. In fact looking at total verification, the NAM did very well for its simulated radar.
(Actual NEXRAD National Radar) (6z NAM January 26 Simulated Radar)
The NAM did seem to have a hold on the 500mb map showing the negatively tilted trough producing the coastal low along with the placement of the upper level low and associated shortwave kicker just to the west.
(18z NAM January 24; 500mb) (0z NAM January 26; 500mb)
In general the differences in the 500mb maps were very subtle with just a general strengthening in the closed 500mb low, which verified a tad north of the January 24 18z model run.
The NAM did a very excellent job in identifying mesoscale band using the UUV/700mb RH charts.
(6z NAM January 26; 700mb)
It indentifed the enhanced snow growth over southeastern Pennsylvania up through New Jersey and New York City. The problems with the NAM generally existed in the QPF fields. The model likely suffered a very convective feedback issues in QPF totals. This is why it is important to note other maps than surface maps to help locate the heaviest precipitation. The 700mb map screamed that snow totals would be farther inland with the enhanced deformation band and UUV rates. In general the high resolution models handled this the best with the HRR scoring an amazing victory for QPF along with the ECMWF. The GFS/NAM did a very poor job for QPF, but as noted above QPF does not always tell the story. Sometimes it is important to note other layers of the atmosphere to help make a forecast. This point is why many forecasters missed the boat. Many forecasters (especially broadcoast meteorologists) are drawn to the easy to understand QPF/surface maps, but one has to look at all layers of the atmosphere to make a prediction.
Lower Susquehanna Valley Doppler...
(Courtesy of WGAL)
"Here northeast of Harrisburg 2010-2011 winter statistics"
Current Snow Cover- 0-3in
Monthly Total (November)- Trace
Monthly Total (December)- 0.6in
Monthly Total (January)- 18.90in
Monthly Total (February)- 1.35in
Seasonal Total- 20.85in
Winter Weather Advisories- 7
Winter Storm Warnings- 2
Ice Storm Warnings- 0
Blizzard Warnings- 0
Freezing Rain Advisories- 1
Winter Storm Watches- 2
Lowest High Temperature- 23.7F
Lowest Low Temperature- -1.7F
Wind Chill Advisories- 0
Wind Chill Warnings- 0
(Snow Storms Stats)
First accumulating snow - December 10 - 0.50in of snow
Clipper light snow - January 7-8 - 2.25in then another 1in of snow
Double Barrel Low - January 11 - 4.5in of snow
Coastal Low - January 17-17 - 1.8in of snow/sleet
Arctic Front - January 20-21 - 2.1in of snow
Upper level/coastal low - January 26 - 5.75in of snow
Two clippers - January 28-29 - 1.5in of snow