forecasts

Interesting article posted by Joel Gratz on OPensnow about how a meteorology class looked at how weather forecasts of various lengths of time in the future performed against reality. They just looked at temperature predictions. They found that forecasts of more than about 10- 11 days in the future are less accurate than just taking the historical average temperature for that day!

I wonder if anyone has done this with snow or rain forecasts. Tony?


http://www.washingtonpost.com/blogs/cap ... -the-test/

No great surprise with this study. I believe this has been the case with previous studies. Long-term weather forecasting is basically a crapshoot. Simply too many variables involved that can effect the weather. Even with super computers and much more sophisticated modeling, long-term forecasts are very unreliable.

Yep, true that! No huge surprise. Still, it's interesting to see it quantified, and see what is the "breakeven point", when the forecast becomes a WORSE predictor than the long-term mean. For temps at least, that seems to be around 10 days out.

It's not really fair to simply state that "long range forecasts are less accurate than the long term mean", as that has always been the case. IMHO, what is far more important is the improvement in the long range models between now and 5, 10, and 20 years ago. Now we're saying that the long range predictions (for temperatures, since that is our strawman argument) seems to go off the rails relative to the long-term mean at around 10 days. I recall a time when a mere 3 days was pretty much a crap shoot. There's also that other truism of weather prediction that is still in play: if you predict that tomorrow will be pretty much like today, you'll be correct something like 77% of the time (but probably with a standard deviation upwards of 2.5).

Did someone say "standard deviation"?

Sent from my Galaxy Note 3 using Tapatalk

I saw that link and was going to post it myself. We need to repost it every time we see those drivel forecasts in September about the upcoming winter.

MarcC":1kom8klt said:

It's not really fair to simply state that "long range forecasts are less accurate than the long term mean"

Click to expand...

Sure it is. If if someone wants to guess about temperature, snowfall or anything else 2+ weeks out, they should look at long term weather history rather than ANYONE's forecasts. And yes they should look at the standard deviations as well as the averages of that long term weather history and prepare accordingly (for outdoor sports like skiing).

MarcC":1kom8klt said:

I recall a time when a mere 3 days was pretty much a crap shoot.

Click to expand...

This is fair. It's certainly an improvement that one week forecasts are worth something now. Maybe there will be further improvement in the future, but for now the limit appears to be 10 days and it's nice to see a demonstration of what most of us thought anyway.

Tony Crocker":2veuql8v said:

If if someone wants to guess about temperature, snowfall or anything else 2+ weeks out, they should look at long term weather history rather than ANYONE's forecasts. And yes they should look at the standard deviations as well as the averages of that long term weather history and prepare accordingly (for outdoor sports like skiing).

Click to expand...

You mean like our predisposition for high pressure in mid-January?

Sent from my Galaxy Note 3 using Tapatalk

Admin":pm9hynl2 said:

Tony Crocker":pm9hynl2 said:

If if someone wants to guess about temperature, snowfall or anything else 2+ weeks out, they should look at long term weather history rather than ANYONE's forecasts. And yes they should look at the standard deviations as well as the averages of that long term weather history and prepare accordingly (for outdoor sports like skiing).

Click to expand...

You mean like our predisposition for high pressure in mid-January?

Sent from my Galaxy Note 3 using Tapatalk

Click to expand...

I'll bite, yes, except that the long term averages don't say you're more likely to have high pressure in mid January. I won't dig up Tony's data but the data says you're just as likely to have snow in mid January as your are in mid December and Mid February....



Sent from my SCH-I545 using Tapatalk

I'll bite, yes, except that the long term averages don't say you're more likely to have high pressure in mid January. I won't dig up Tony's data but the data says you're just as likely to have snow in mid January as your are in mid December and Mid February....

Click to expand...

And as likely to have the inversions in December as January, demonstrated as we speak and also a couple of weeks ago. And actually it was socal that dug up the inversion data. So I suppose one of those "standard deviations" to consider is not staying in the Salt Lake Valley in December/January if you're unusually sensitive to the smog. I'm guessing it's not as bad as growing up in Pasadena during the 60's though.

Tony Crocker":1bog362r said:

not staying in the Salt Lake Valley in December/January if you're unusually sensitive to the smog. I'm guessing it's not as bad as growing up in Pasadena during the 60's though.

Click to expand...

LOL, I don't know how anyone wouldn't be sensitive to the SLV smog that was there during my visit two weeks ago. How bad? From the end of Delta Concourse B, you couldn't see the planes on the runways and taxiways. Thank god I had the luck to be staying at Solitude.

same for predicting the coming hurricane season.
NOAA gave a 70% chance of 13 to 20 named storms, there were 2.
they predicted 3 to 6 major hurricanes(cat 3-5), there wasn't even a cat 2.
http://www.noaanews.noaa.gov/stories201 ... antic.html
http://en.wikipedia.org/wiki/2013_Atlan ... ane_season
these future predictions are always reported by the media as "news", when in fact they are sheer conjecture, they would do just as well by looking at the horoscope rather than using their fancy computer. oh well, everyone has to make a living somehow.

SnowbirdDevotee":ps5n7lbm said:

these future predictions are always reported by the media as "news", when in fact they are sheer conjecture, they would do just as well by looking at the horoscope rather than using their fancy computer. oh well, everyone has to make a living somehow.

Click to expand...

And somehow I get the feeling that if there were, say, 25 named storms instead of the predicted 13-20 and 10 of those were major (cat 5 say) instead, you'd be saying the same thing about horoscopes and fancy computers.