WEATHER BACKGROUND INFO

This page is to basically give you some background weather information regarding forecast models and techniques.

We will first look at some National Weather Service (NWS) factors in forecasting before looking at some good general rules of thumb for local use.

The NWS uses supercomputers to generate depictions of the atmosphere based on observations taken at regular intevals. Using a number of formulas (which take into account things such as thermodynamics, fluid dynamics, vector analysis, etc.- all in a three dimensional field), they are able to give projections of what the atmosphere will look like in the future. Such output is then "tweaked" for local variations such as mountains and water effects and then local forecasters use experience and some intuition to finalize a forecast.

A Quick summary of NWS-used models:

NGM (nested grid model) - Oldest still in use (and scheduled to be discontinued), but it still is helpful as a reference. Tends to be to "wet" and holds on the cold air masses too long. Can be slow in deepening surface cyclones across the Gulf Stream during winter. Too slow with Arctic airmasses plunging southward down the plains. Only goes out to 48 hours.

GFS (the old Aviation/Medium Range Forecast) - A spectral model (meaning global), was created to prepare aviation forecast for entire globe. Takes a broader look at patterns. Goes out to 16 days. Very good general model.

ETA (like Greek letter) - A newer model with higher resolution (more levels of the atmosphere, smaller grids etc.). This should give better detail and accuracy, but it also has it's error tendencies such as over-blowing coastal low development. But it does handle cold-air damming east of the mountain scenario better. Goes out to 60 hours. (There is an ETA extension to 84 hrs.)

MM5 - A new model by the Air Force. Goes out to 48 hours.

RUC (Rapid Update Cycle) - Outputted much more often for finer resolution on a time line. Only goes out to 12 hours. Good for close-in timing.

UKMET - The model used by Great Britain. Has problems with shallow cold air. The model tends to lower surface pressures too much and too far south and often implies synoptic-scale fronts too far south. Goes out to 72 hours.

NOGAPS - The model used by the Navy. Good for general features. Handles cold air well. Goes out to 144 hours.

ECMWF - The ECMWF does the best in predicting mid/upper tropospheric heights during the colder part of the year. Handles shallow cold air situations well. The European Model. Goes out 9 days, but is restricted to 7 days for the public.

RGEM/GGEM - Canadian Models - RGEM is higher resolution going out 48 hrs., GGEM is global and goes out 144 hrs.

ENSEMBLES - Variations of models such as GFS, ETA, European and  Canadian which uses slightly different variables in the equations which give slightly different results. Used for comparison.

As you can see there is quite an arsenal of models for comparison. Often forecasts are decided on "model clustering." That is, by concensus of the majority of the models. That along with rejecting certain model biases helps forecasters come to a result. By the way, NCEP stands for National Center for Environmental Prediction, the national forecasting and overview section.

Now some comments on local factors:

Many ask me "why isn't it snowing when it seems to be cold enough?" The simple answer is usually that it just isn't cold enough aloft for it to happen. That is, while it may be 31 at ground level, it may be 42 degrees up at 6,000 ft. Snow only happens when the precipitation FORMS at below freezing (in a cloud) and stays that way until ground level. If it falls as rain through sub-freezing air, then you will either get sleet (if it has enough time-a deep enough layer of cold) or freezing rain (not enough time-just stays liquid). It may begin as snow in cold enough air up higher in higher cloud, then fall through warm air to become raindrops, then through cold air to freeze as little ice pellets (another name for sleet).

That is why the NWS folks send up balloons to get a portait of the mid and higher parts of our atmosphere. That info is put into the models as well as surface data. The supercomputers can generally predict if we are going to have a very warm layer slide in over a cold one. That's how they can usually guess what kind of winter precip is possible. Rainfull isn't as hard, for variations are hardly noticeable - do you notice a big difference between .15 and .38 inches of rain? But in a liquid equivalent (usually 10 to 1), you WILL likely notice the difference between 1 and 4 inches for the same amount of liquid equivalent! All over hundreths and tenths of an inch in liquid equivalents! So give forecaster a break on the snowfall depth scenarios. Sometimes liquid equivalents from different models may give a range from .05 to 1.0" which means either a trace of snow or 10 inches (maybe 15 inches if a dry colder snow!).

I won't go into thickness level, vertical velocity, lifted index, constant height levels, and other deeper meteorological concepts, but suffice it to say that meteorologists don't just look at surface weather maps to guess what's coming. Models, satellite pics, radar, and more help put a forecast together. Sometimes I think that they give the models too much weight over the other tools. I have at times seen them blindly follow the models when radar trends and surface data clearly shows another direction. That with a lack of experience, particularly in a specific local area, can lead to some pretty bad forecasts.

One more point. Something I had to learn, but Roanokers know by experience. Snow hardly ever comes from the west, because of something called DOWNSLOPING. That is when snow and rain tries to cross over the mountains to the west. At first the precip UPSLOPES which causes condensation and precipitation (in WV). Then when it comes DOWN our way over the mountains, air tends to warm and dry as it gets lower, thus losing its moisture. Oh, we might get a light flurry, meanwhile they get inches of snow in the mountains to the west. Our snow comes when a low forms in south TX or northern Gulf and moves up south and east of us. That slings moisture in from the east or northeast which actually puts US in the UPSLOPE mode. Also, usually a cold high pressure center needs to be over the northeast US. As a long-time local tells me, "If they are calling for winter weather, and if it is colder in Roanoke than Bluefield, then we're gonna get it."

I may add to or adapt this document in the future.

If you have any questions, email me at mayo@weatheroanoke.com

WMayo

Weather Roanoke

www.weatherroanoke.com