The Weather Thread

  • Thread starter Thread starter nycfan
  • Start date Start date
  • Replies: 1K
  • Views: 41K
  • Off-Topic 
Looks like some models are trending north. Canadian model looks promising for a big snow in central NC. NAM aligning with that

Fingers crossed
 
They really flopped with this... Nothing expected in the western triangle now and the beach won't even get but a couple of inches. Dud
 
ChapelHillSooner said:
My problem is that they often present information in a way that implies some level of precision.

If you're going to say 1-3", that range itself should incorporate the variability - within one standard deviation or so. If that means your forecast looking a week out has to say you are expecting 0-20" then so be it. Don't pretend to have knowledge that you don't have.

I used to complain about this to my meteorologist brother. He explained that in the NWS (at least in their standard communication releases) they were pretty rigid and couldn't do statistical probability type forecasting.
Click to expand...
I'm a little late to reply to this as I haven't been on this thread, but here goes.

I'll amend my previous statement to be more accurate: "Forecasts a week plus out aren't useless, they just aren't highly reliable to a high degree of precision."

And the tradeoff we currently accept is we get a more precise prediction at that distance understanding that it carries a lower degree of probability.

But the opposite approach - higher degree of probability at the cost of a much wider range of precision - truly does make the forecast useless.

To use your example...if I'm told that the weather forecast for one week out is a chance of snow between 0-20", that means nothing to me in realistic terms. It means my day could be reasonably normal weather-wise or it could be the beginning of a week stuck at home with my kid waiting for the roads to be cleared. The range of potential outcomes is so large as to make the information non-operational in any pragmatic way. But saying that the most likely outcome is a chance of snow between 1-3" lets me know that there's a chance of a couple of days snow interruption with an implied understanding that the probability of that outcome is fairly low. However, it is useful because it provides me a heads-up that winter weather could occur and gives me the best understanding possible at that time of the amount of snow (which defines the impact of said winter weather).
 
Maybe we'll see some snow fall in central NC tomorrow night but it won't be much. I know it is really hard to forecast snow here and the weather models can't be trusted more than 48 to 72 hours out. I think WRAL purposely forecasts low qhen it comes to snow, though, and adjust if needed when the event starts. They would rather foeecast too little than to forecast a big storm and it end up being a bust.
 
CarolinaFever said:
Maybe we'll see some snow fall in central NC tomorrow night but it won't be much. I know it is really hard to forecast snow here and the weather models can't be trusted more than 48 to 72 hours out. I think WRAL purposely forecasts low qhen it comes to snow, though, and adjust if needed when the event starts. They would rather foeecast too little than to forecast a big storm and it end up being a bust.
Click to expand...
I once read a study about weather forecasts focused on atypical weather (all storms, not just snow) and how the location of the forecaster vis-a-vis the location forecasted impacted the "reliability" of the forecast (or how close the forecast came to being correct). If I remember correctly, this study looked at forecasts something ike 5 days out.

The best forecaster was the National Weather Service forecast for any particular community. Following not far behind were national forecasters like The Weather Channel - who largely rely on the NWS forecast - who do a bit of their own forecasts but are looking at forecasts across the country and not just in one location.

The worst forecasts came from local forecasters, which seems odd as you would imagine that folks could learn their local area better than the national folks and do better than them. What this study showed was that local forecasters were much more likely to predict atypical weather events - severe storms, snow fall, significant rain - for their area than the national forecasters. And the reason that this study posited these local forecasters did so was because they were incentivized to do so by viewers...in essence, viewers (a) paid more attention to the weather forecasts when something that would affect their day was predicted and (b) viewers were much more likely to get upset if the news didn't predict significant weather and then that weather happened than if the weather forecaster predicted a significant weather event that never panned out. In short, folks much prefer to be alerted when the weather is going to be potentially bad and then feel relief when it's a minor issue than to have a significant weather event that catches them off guard (i.e. they'd much rather prepare for snow that doesn't arrive or carry an umbrella they end up not needing than the inverse).

I've tried to find the study in the years since I read it, but I've been unable to locate it. I really wish I could get my hands on it again.
 
You must log in or register to reply here.
Back
Top