New 25-Day Forecast on AccuWeather

[ChrisM]

Planning around a 12 day forecast is not to swift, how many times have you seen the Euro or GFS nail a 10 day forecast? Probably <20 times a year.

Exactly....but people do it. Thats my whole point

[hm8]

Terrible idea and yes the general public will look at this and take it for face value. When in reality anything beyond 7 days is rarely forecasted correctly.

Honestly I think even the general public realizes that predicting exact rainfall amount so far in advance is absurd.

In fact, according to this article and this one, many people already criticize AccuWx for their inaccurate long range forecasting and it ranks least popular among the major weather sites.

[turtlehurricane]

Ahh but they're all researchers and meteorologists... not our target audience. We don't mean to be disrespectful to that group, but what they're suggesting -- holding off until it's 100% accurate or giving a statistical analysis with the forecast -- is not what the public wants. It's a bittersweet realization for me.

Yeah, just ignore all the scientists who have been studying these sort of things their entire lives. You must realize the 10-25 day forecasts are going to lead to a "boy who cried wolf" situation as they will consistently fail. In the end it will just make accuweather look worse since people will think you guys are unreliable.

Maybe you should develop a different long range product that forecasts large-scale trends with the data you have, rather than being so specific.

[NJHurricane]

No offense to the Accu-WX employees who post here, but in my opinion this is 100% bunk. I have been in this game for over 30 years, and a forecast this far out is a total fantasy, plain and simple. It is just another way to "differentiate us us from the competition" or some other such corporate blather which is code for "ways we can rip folks off with a worthless product". I do a forecast to seven days for my job, and almost always days 6 and 7 tun out to be dead wrong. Specific temperatures and weather to 25 days? It is to laugh, but is also sad in a way.

I am rarely an accuweather defender but I have to say....the very attempt to forecast at different/extended ranges IMO is part of the process of identifying what features and data will need to be followed/developed to develop a degree of skill over time for this time range. I'm sure the idea of a seasonal forecast for hurricane season seemed like an absolute waste of time, same for a three month outlook. What I find interesting is to my knowledge no effort has been made to develop skill for individual days at this time range....less than 30 days but greater than two weeks. I think the use of the term "forecast" muddles things and does imply a degree of skill that will simply not exist early in this process. Not sure what you would call it....long lead medium range outlook?

[dtk]

I can see seasonal/monthly forecasts improving with things like the Euro/CFSv2 Long Range and Creatively tooled forecasts made by AER. I suppose one could apply probabilistic 25 week forecasts to these different tools along with a mix of climatology, it's still a trick though so damn far out. . Does actual verification exist in the 25 days range? It's hard to get any decent high model verification correlation coefficient especially +7 days. I guess I'll see it when I believe it, it sure isn't going to happen with the non-4d stuff the U.S. pumps out.

Irrelevant to the current topic, FYI. All you have to do is look at the day 10 verification for something "easy" to predict like 500 hPa heights (that's right, the GFS is better over the past 30 days for day-10 lead):

To extend actual forecasting capabilities to something like 3+ weeks, we still need further advancement in modeling, faster computers, coupling to land/ocean systems, reforecasting and other calibration techniques (bolded for emphasis). There are legitamite predictability boundaries to overcome here....

To your other point, there are certain things that have been shown to demonstrate some skill in the day 7-14 range for T2m (i.e. callibrated probabilistic forecasts using a training reforecast dataset).

[icebreaker5221]

Going out day by day in detail 25 days appears to me to be a farce of a forecast. I thought going out to day 16 in detail based on GFS runs was already a bit much due to the inherent inaccuracy. Now to day 25? OMG

What's next? A 365 day by day forecast?

You can already get that here:

http://www.weathertrends360.com/

[WxMidwest]

Irrelevant to the current topic, FYI. All you have to do is look at the day 10 verification for something "easy" to predict like 500 hPa heights (that's right, the GFS is better over the past 30 days for day-10 lead):

To extend actual forecasting capabilities to something like 3+ weeks, we still need further advancement in modeling, faster computers, coupling to land/ocean systems, reforecasting and other calibration techniques (bolded for emphasis). There are legitamite predictability boundaries to overcome here....

To your other point, there are certain things that have been shown to demonstrate some skill in the day 7-14 range for T2m (i.e. callibrated probabilistic forecasts using a training reforecast dataset).

I suppose so, but the 4d comment really wasn't my point more like a mention, all of it still basically sucks far out past 15 days..... But yeah, LENTZ's work @ MIT on probability of dynamic models tells us it's gonna be very hard to predict consistently in the planetary spatial/long range temporal time scale... of course, so does common sense...

[okie333]

Unfortunately, the big cheese says I can't tell (I asked). I'm sure someone here is familiar with what forecast data would be out there in that range. We take that data & historical data and run it through algorithms to improve the accuracy and resolution (in addition to any corrections that would already be in the database for our forecast points. Our meteorologists can edit the data,

It's a new frontier, for sure, but so was the 5-day when we introduced it in the 1960's. And what better place to get feedback on accuracy than the whole world?

Might that be the Euro ensembles by any chance?

[Juliancolton]

Props to the company for trying out something new. It made me look at my Accuweather forecast, which I haven't done in a long time, and I imagine the period extension had the same type of effect on many individuals and businesses. The people want to have something to look at – anything – while they're waiting for their wedding or group hike or firework shoot in three weeks, even though I think most people are well aware of the verificiation issues in weather forecasts. After all, the whole "weather liar" bit is a meme in most of urban America, but everyone still uses wx forecasts. If nothing else it provides a way to *track* the weather forecast for a specific day or week, even if the final idea isn't set in stone until within 5 days.

[Cheeznado]

Props to the company for trying out something new. It made me look at my Accuweather forecast, which I haven't done in a long time, and I imagine the period extension had the same type of effect on many individuals and businesses. The people want to have something to look at – anything – while they're waiting for their wedding or group hike or firework shoot in three weeks, even though I think most people are well aware of the verificiation issues in weather forecasts. After all, the whole "weather liar" bit is a meme in most of urban America, but everyone still uses wx forecasts. If nothing else it provides a way to *track* the weather forecast for a specific day or week, even if the final idea isn't set in stone until within 5 days.

I disagree, since I am pretty confident that a forecast at 25 days has an much chance of verifying as one in 60 days, or a year, or 25 years- so why do it at all? If folks want to look at something, look at climo, or look at a more reasonable outlook with probabilities of above or below average precip/temps as those will not change daily and will have a better chance of being closer to reality. You cannot just give in to every whim of the public. The fact they are so precise with rain to the hundredth of an inch etc makes it even more ludicrous.

[WEATHER53]

No hatin on AccuWeather is needed.

Their internet and basic stuff is OK. Some of their wintertime guys not soo good-call for snow too much.

[Stebo]

No hatin on AccuWeather is needed.

Their internet and basic stuff is OK. Some of their wintertime guys not soo good-call for snow too much.

Not hating just rationalization by several meteorologists here.

[WEATHER53]

Not hating just rationalization by several meteorologists here.

Can I be referee?

[binovc]

Just look at all the forecasts last fall for this past winter, how well did everyone do ?

[CoastalWx]

Just look at all the forecasts last fall for this past winter, how well did everyone do ?

You're missing the point. This isn't a probability forecast on a season saying it will be colder or warmer than normal and assigning probability to it, they are doing a deterministic forecast with variables like hi lo and precip for 25 days out! It's impossible to do with any meaningful accuracy and it truly scares me that this idea was brought forth and people are going to be paying money for a voodoo forecast. I don't always agree with her, but Wxtrix made a good point and so did OceanStWx and CTRain.

[doncat]

Would love to know what Elliot Abrams really thinks of this.

[MGorse]

You're missing the point. This isn't a probability forecast on a season saying it will be colder or warmer than normal and assigning probability to it, they are doing a deterministic forecast with variables like hi lo and precip for 25 days out! It's impossible to do with any meaningful accuracy and it truly scares me that this idea was brought forth and people are going to be paying money for a voodoo forecast. I don't always agree with her, but Wxtrix made a good point and so did OceanStWx and CTRain.

Their 25-day forecasts are on the free site, unless they end up changing that but as of now they are free.

[RodneyS]

You can already get that here:

http://www.weathertrends360.com/

I had never seen that site before, but decided I would get the advance scoop on the upcoming winter in DC. First, as of today (April 6, 2012), their daily forecasts run only through March 2, 2013 -- a mere 330 days. Second, the daily snow forecasts doom DC to another largely snowless meteorological winter. Specifically, the daily forecasts show no snow until February 2013, and then only on four February days -- the 2nd (wintry mix), the 14th (snow), the 17th (wintry mix), and the 18th (light snow). Then again, the site does not forecast the amount of snow on the 14th (Valentine's Day, no less), and so perhaps that will be the mother of all snowstorms.

[mappy]

I had never seen that site before, but decided I would get the advance scoop on the upcoming winter in DC. First, as of today (April 6, 2012), their daily forecasts run only through March 2, 2013 -- a mere 330 days. Second, the daily snow forecasts doom DC to another largely snowless meteorological winter. Specifically, the daily forecasts show no snow until February 2013, and then only on four February days -- the 2nd (wintry mix), the 14th (snow), the 17th (wintry mix), and the 18th (light snow). Then again, the site does not forecast the amount of snow on the 14th (Valentine's Day, no less), and so perhaps that will be the mother of all snowstorms.

Maybe it uses the BB algorithm

[CoastalWx]

Their 25-day forecasts are on the free site, unless they end up changing that but as of now they are free.

That's even more dangerous lol.

[Superstorm93]

I'd love to know why the AccuWx free site has this before the AccuPro site does...

[andyhb]

This is plain ridiculous, there is no way whatsoever that these forecasts can be very accurate at all, which has been highlighted by a plethora of posts here. For the people who do check out Accuweather, since the general public tends to not know better, they're will be thoughts like "Oh, looks sunny on X date", so they set up outdoor events for example, when in reality there could be a snowstorm or tornado outbreak on the same day realizing. This just screams unnecessary use of money.

[The_Global_Warmer]

We need a computer model that is more AI than straight number crunching. We need a smart model that is in flux, that has access to any data that can be put in a grid that the smart model can assimilate.

Then give the smart model future hindcasts. So it's creating forecasts at a new point every hour up to 5 days out, then every 3 hours if needed.

Then the Smart model will take all of those seperate full scale forecasts and start merging them to the final forecast that will always be subject to incoming data 24/7/365. every second of everyday the smart model will input new data and re-run it;s algorithms. As it gets better it changes it's own climo and bias's to adapt to the chaos of the Earth.

This would need full access to solar data, geology, anything and everything that could help.

I doubt we could build such a machine/program..but I think it's headed towards adapting programs.

[#NoPoles]

We need a computer model that is more AI than straight number crunching. We need a smart model that is in flux, that has access to any data that can be put in a grid that the smart model can assimilate.

Then give the smart model future hindcasts. So it's creating forecasts at a new point every hour up to 5 days out, then every 3 hours if needed.

Then the Smart model will take all of those seperate full scale forecasts and start merging them to the final forecast that will always be subject to incoming data 24/7/365. every second of everyday the smart model will input new data and re-run it;s algorithms. As it gets better it changes it's own climo and bias's to adapt to the chaos of the Earth.

This would need full access to solar data, geology, anything and everything that could help.

I doubt we could build such a machine/program..but I think it's headed towards adapting programs.

and then that computer realizes humans are a parasite on the earth, and launches "operation clean-sweep" against the human race...

actually i like that idea, but i didn't win megamillions...and while we're on the subject of huge innovations, how bout we try to solve the problem of our dependency on internal combustion engines?

[The_Global_Warmer]

and then that computer realizes humans are a parasite on the earth, and launches "operation clean-sweep" against the human race...

actually i like that idea, but i didn't win megamillions...and while we're on the subject of huge innovations, how bout we try to solve the problem of our dependency on internal combustion engines?

In an idealistic world that would be done in a month. But we are in a money driven world so.....yeah

[superjames1992]

And here I always thought it was ridiculous to make a 15-day forecast. Now we have a 25-day forecast...

[MGorse]

I recently noticed these forecasts are not just for the U.S., but worldwide. From their press release, "The AccuWeather 25-Day Forecast is available for 2.7 million locations worldwide, including every location where persons live and work".

http://www.accuweath...asp?entry=63630

[BillB]

I'd love to know why the AccuWx free site has this before the AccuPro site does...

Matching the forecast to its value?

[LocoAko]

[source: Harold Brooks]

[phlwx]

I noticed that Philly's original forecast had a high of 58 with 0.71" of rain for April 23rd...the forecast for the same date as of this morning has April 23rd with sunshine and a high of 79.

Consistency fail...