Author Topic: Washington Post spanks AccuWeather for it's long range forecasts  (Read 1013 times)

0 Members and 1 Guest are viewing this topic.

Offline DaleReid

  • Forecaster
  • *****
  • Posts: 2002
    • Weather at Eau Claire, WI
Washington Post spanks AccuWeather for it's long range forecasts
« on: November 07, 2019, 01:07:01 PM »
I couldn't figure out how to copy a link to the video, but today's Washington post has this video story, about 3.52 minutes long.....

Meteorologists say AccuWeather’s unscientific forecasts are hurting their credibility

If you go to the Washington Post and look to the right side, you'll see a bunch of video stories, and while I have a subscription, I think that you can look at a few articles without subscribing or paying.

Interesting to hear the description of what they are putting out, and 45 or more days in advance.  Of course, AW won't say how or why and the true accuracy is less than a coin flip.

One real meteorologist compares it to writing horoscopes.

I guess this is another publicity stunt, like naming winter storms, or non-tropical storms that another self-important national weather company tries to do.

ECWx.info
&
ECWx.info/t/index.php


Offline Jasiu

  • Forecaster
  • *****
  • Posts: 947
    • LexMAWeather
Re: Washington Post spanks AccuWeather for it's long range forecasts
« Reply #2 on: November 07, 2019, 01:46:50 PM »
Capital Weather Gang at WaPo generally produces good, and often entertaining, stuff.

  https://www.washingtonpost.com/news/capital-weather-gang/

https://lexmaweather.info
On Mastodon: @LexMAWeather@toot.community

Offline ValentineWeather

  • Forecaster
  • *****
  • Posts: 6364
    • Valentine Nebraska's Real-Time Weather
Re: Washington Post spanks AccuWeather for it's long range forecasts
« Reply #3 on: November 07, 2019, 06:07:07 PM »
I find it odd anyone cares, this has that stenchy smell of politics.  I don't think anyone takes a 45-day forecast seriously if they do they also believe in the Farmers Almanac forecast and get what they deserve. 
Who cares what they do its a private company?  Getting beyond 14 days is ridiculous, I've watched the models flip completely inside of 7 days.
Meteorologists should be concerned about their own forecasting failures and not worry about privately-owned companies.
Randy

Offline nincehelser

  • Forecaster
  • *****
  • Posts: 3337
Re: Washington Post spanks AccuWeather for it's long range forecasts
« Reply #4 on: November 07, 2019, 06:20:14 PM »
Has anyone come up with a website to judge the accuracy of forecasts?

For example, let's say the web site shows today's current conditions and then lists various past forecasts that were given for today.  You could then rate how accurate those forecasts were.

I suppose it would be best to have an impartial machine "grade" them, but it would also be interesting to see how humans perceive their accuracy.

Offline CW2274

  • Forecaster
  • *****
  • Posts: 6731
    • Conditions @ CW2274 West Tucson-Painted Hills Ranch
Re: Washington Post spanks AccuWeather for it's long range forecasts
« Reply #5 on: November 07, 2019, 06:21:40 PM »
this has that stenchy smell of politics. 
Well...it is the Post... :roll:

Offline DaleReid

  • Forecaster
  • *****
  • Posts: 2002
    • Weather at Eau Claire, WI
Re: Washington Post spanks AccuWeather for it's long range forecasts
« Reply #6 on: November 07, 2019, 08:00:11 PM »
As to the question of grading forecasts, I'm wondering the same thing:  Not of this clearly no-basis long range, but the various forecasts issued by the NWS.

I know there are at least four models running now, and there was a cut over to using a newer one, more like the European Model this spring.

My knee jerk reaction was, before I heard about the possible change, on just how good the forecasts were.  Rain occurring within an hour or so of when it was predicted in my area.  Sure there were occasional goofs, but nothing big.

Now, it seems they are predicting clouds and light rain when we are having a whole day of bright blue sky.

Clearly there is a scientific way to evaluate this,but I've not seen it and do not have the skill to design an analysis to test this.

So I'm left with the morning after reflection of what we were told may happen and what did happen.

Anyone else have a feeling about the quality of the new model?  I'd assume it is being analyzed six ways to heaven over time.
ECWx.info
&
ECWx.info/t/index.php

Offline galfert

  • Global Moderator
  • Forecaster
  • *****
  • Posts: 6822
Re: Washington Post spanks AccuWeather for it's long range forecasts
« Reply #7 on: November 07, 2019, 08:59:46 PM »
Has anyone come up with a website to judge the accuracy of forecasts?

For example, let's say the web site shows today's current conditions and then lists various past forecasts that were given for today.  You could then rate how accurate those forecasts were.

I suppose it would be best to have an impartial machine "grade" them, but it would also be interesting to see how humans perceive their accuracy.

Windy has this functionality. You can see your station data (or that of whatever station you are looking at on Windy) and then it shows in the past what the forecast was for a few metrics.

 [ You are not allowed to view attachments ]

The dotted lines are the forecast of the selected computer model which you can change on the bottom left to various ones (European model selected in this example). The pink line shaded area for example is the actual data from the station for temperature and dew point. The solid line for the barometer is actual and again the dotted line is was the forecast. The wind direction triangles where there is a bunch of them is the actual and the lesser line of wind direction triangles is the forecast. Wind speed is depicted on top of both arrow sets in color (blue in this case and other colors for higher speeds).

This is why I like sending my data to as many places as I can. You just get a little bit something different from each service.
« Last Edit: November 07, 2019, 09:04:09 PM by galfert »
Ecowitt GW1000 | Meteobridge on Raspberry Pi
WU: KFLWINTE111  |  PWSweather: KFLWINTE111
CWOP: FW3708  |  AWEKAS: 14814
Windy: pws-f075acbe
Weather Underground Issue Tracking
Tele-Pole

Offline SoMDWx

  • Forecaster
  • *****
  • Posts: 1014
    • Southern Maryland Weather
Re: Washington Post spanks AccuWeather for it's long range forecasts
« Reply #8 on: November 08, 2019, 01:09:04 PM »
I have been on a personal tour of the main facility @ Accuweather. Very impressive setup..

One long hallway there has posters/pictures and stats to all the forecast prediction they have done over many years and show their % of accuracy for them....

Really wouldn't call them un-professional....A large portion of their people come out of the Penn State Meteo program.....

Jim


Offline ValentineWeather

  • Forecaster
  • *****
  • Posts: 6364
    • Valentine Nebraska's Real-Time Weather
Re: Washington Post spanks AccuWeather for it's long range forecasts
« Reply #9 on: November 08, 2019, 01:33:00 PM »
Accuweather sent me a nice email last year asking if they could use my web cameras, of course, I said yes that's why I have them up.
Randy

Offline SnowHiker

  • Forecaster
  • *****
  • Posts: 352
Re: Washington Post spanks AccuWeather for it's long range forecasts
« Reply #10 on: November 08, 2019, 01:38:22 PM »
Has anyone come up with a website to judge the accuracy of forecasts?


https://www.forecastadvisor.com/

Put in your zip code and it supposedly compares the accuracy of various organizations for their forecasts.  But only for one to three day forecast, not long range as I understand.

« Last Edit: November 08, 2019, 01:42:47 PM by SnowHiker »

Offline ocala

  • Forecaster
  • *****
  • Posts: 4383
Re: Washington Post spanks AccuWeather for it's long range forecasts
« Reply #11 on: November 09, 2019, 09:45:16 AM »
Has anyone come up with a website to judge the accuracy of forecasts?


https://www.forecastadvisor.com/

Put in your zip code and it supposedly compares the accuracy of various organizations for their forecasts.  But only for one to three day forecast, not long range as I understand.
Interesting site. Thanks for posting.
The blues had a baby and they named it Rock & Roll

Offline DaleReid

  • Forecaster
  • *****
  • Posts: 2002
    • Weather at Eau Claire, WI
Re: Washington Post spanks AccuWeather for it's long range forecasts
« Reply #12 on: November 09, 2019, 02:49:22 PM »
I went to the forecastadvisor site and put in my zip code.  Nice little 5 day display. 

I looked down below at the listing, in accuracy order, of the various site and NWS was part way down.  I clicked on that and got the message that why bother since Forecastadvisor uses NWS for it's forecast.

I went back to the page for my zip code and I cannot see anywhere on there that the data are attributed to the NWS. 

I thought that while the data were free to use (and Lord knows almost every subscription service uses the NWS radar for example) there was, like NASA, a mandatory statement that these data were from NWS.

Maybe they relaxed that requirement, or there are folks who don't put that in a prominent place to be seen by those using the forecasts.  Our tax dollars at work.

Just curious.  Of course I could have missed something.

Dale
ECWx.info
&
ECWx.info/t/index.php

Offline K5GHS

  • Senior Member
  • **
  • Posts: 68
    • K5GHS's Website
Re: Washington Post spanks AccuWeather for it's long range forecasts
« Reply #13 on: January 14, 2020, 12:09:06 AM »
Its not just long range.  I get their forecasts morning and evening by email (I mainly subscribe to get alerts, just in case for some reason I'm at work and have to have my alert tones off on my phone).  Sometimes they predict thunderstorms and NWS says no and vice versa, and their highs and lows predicted are sometimes just way off.  I mean I know that sometimes stuff changes and suddenly it stays warmer or gets colder, but it seems like at times they are pre-prepared, supposed to be audited, but then aren't and they just go out.  I've occasionally seen temp predictions fluctuate wildly between the evening and morning email or vice versa too.

There is a lot of data out there, but sometimes I think the humans parsing it or the computers that are somehow picking it up are grabbing the wrong stuff and its making it out to the end user.

I do have to say though that alerts by county for weather related stuff, as well as any Silver or Amber alerts and similar are out within a minute or two of them getting transmitted out by weather radio or to the Cell phone system they use for emergency alerts though, which is why I still use them for that.  I get the forecasts to just see how wacky they are at times, though and there are times they are just....off....
Joshua
K5GHS
http://www.k5ghs.radio

Collin County, TX  Grid: EM13qd  CWOP: AV590
Skywarn Repeaters:  147.180 Primary 146.740 Secondary
100% Emergency Powered Ham Radio Station (540AH Batteries/Solar Rechargeable)
Ambient Weather WS-2000
Ecowitt GW1000

 

anything