Anyone who counts on computer model progs verifying beyond day 8 will spend a LOT of time being disappointed. Recognize that something COULD happen yes...but expecting it to...is only asking for anger or depression. Some people may not realize this, but the GFS truncates (loses resolution) past 192 hrs. That tends to provide less than optimal results at times because the model grid becomes coarser. The Swiss cheese holes get bigger and more numerous. One is that more stream phasing tends to occur in models than what happens in the real world atmosphere.

This is especially an issue on 06Z and 18Z runs...which are initialized on "Best Guess" fields rather than upper air observations. Any errors from reality in the initialized fields get compounded down the road, especially after truncation after day 8. This makes their output less reliable. Research has shown the 18Z run of the GFS is the most error prone.

All someone has to do is look at the ensemble that goes with a single model run to see how radical the differences are at 240 hours. You have P10 (perturbance 10) with the snow line over southern Wisconsin and southeastward to the North Carolina coast and a surface high covering much of the eastern U.S. Right beside it...P9 (perturbance #9) has the snow line in the middle of the Gulf of Mexico and extending northeast to Virginia...with low pressure off the Carolina coast: COMPLETELY different scenarios.

There are 22 possible solutions from the GFS ensemble (Control, Operational, and 20 Perturbations (10 positive and 10 negative)). So you as a user want to hang your hat on a solution that has less than a 5% chance of being the correct one verifying? I am not trying to be mean or snide, just instructional. The GFS alone is run 4X a day. There are dozens and dozens of different models run at least twice a day. No point watching just one nor just looking for one that shows what you want.

Those that put their faith in The Weather Channel, AccuWeather's (or anyone else's) 10-20 day forecasts need to keep that in mind. Those forecasts (point and click sites on the internet or your phone from cable and TV sites) that far out is totally based off of a specific model (probably the GFS operational, sometimes an in-house model). And the forecast will change with each model run. If you doubt that...try monitoring the extended forecasts (say for December 19th, for example) for a particular location a few times a day for several days. Check and see how much the forecast moves around.

What I hate is how everyone posts questions or their own "forecasts" based on these 10-20 day computer model projections that never verify. It's wishcasting and model worshiping time wasting. The models are so ever changing that you can't bank on storm systems that far out. The long-range models continue to be on and off the table for any kind of system. So I know the models are doing their normal cha cha cha of storms here today, gone tomorrow; we have to play the normal model games.

But weather forecasting is meteorology, not model-ology. As I've said in the past couple weeks, the general overall PATTERN is favorable during the next 1-3 weeks for at least a chance of snow and/or ice in the South. But lets take it one REAL storm at a time (if there are any) within 8 days, not the phantoms of la-la voo-doo land in the 7-15 days ahead range.

Here is a tip: if a model shows the same thing in the extended range, many days in a row--THEN and only THEN you might have something to get excited about. But I still said MIGHT. It's also not a good idea to just look at one model. It's an even better idea to let a meteorologist do their job and look at everything, and then tell you about it when professional judgment says its time, what a concept. Saves a lot of time and trouble for everybody. Happy holidays everyone.