If you can read this, either the style sheet didn't load or you have an older browser that doesn't support style sheets. Try clearing your browser cache and refreshing the page.

(Computerworld)   The National Weather Service went live this morning with two new forecasting super computers capable of getting the weather wrong at even faster speeds and in greater resolution   (computerworld.com) divider line 31
    More: Cool, National Weather Service, National Oceanic and Atmospheric Administration, supercomputers, weather forecasts, decimals, parallel, weather, linux  
•       •       •

917 clicks; posted to Geek » on 26 Jul 2013 at 1:24 PM (1 year ago)   |  Favorite    |   share:  Share on Twitter share via Email Share on Facebook   more»



31 Comments   (+0 »)
   
View Voting Results: Smartest and Funniest
 
2013-07-26 01:35:16 PM
Do Wah Diddy
 
2013-07-26 01:43:37 PM
213 teraflops.  Substantial.
 
2013-07-26 02:00:40 PM
weather channels will still be all death and disaster all the time, in between shows with guys shooting ice bergs and people whacking rocks with hammers.
 
2013-07-26 02:02:53 PM

HotIgneous Intruder: weather channels will still be all death and disaster all the time, in between shows with guys shooting ice bergs and people whacking rocks with hammers.


The winter storm forecasting has become so apocalyptic in our area that I can't take it seriously anymore.  Multiple times over the past few years, they've predicted heavy snowfall (6-8 in.; heavy for the DC Metro area), and we haven't gotten so much as a farking flake.
 
2013-07-26 02:13:44 PM

born_yesterday: HotIgneous Intruder: weather channels will still be all death and disaster all the time, in between shows with guys shooting ice bergs and people whacking rocks with hammers.

The winter storm forecasting has become so apocalyptic in our area that I can't take it seriously anymore.  Multiple times over the past few years, they've predicted heavy snowfall (6-8 in.; heavy for the DC Metro area), and we haven't gotten so much as a farking flake.


It's the Congressional Thermal Updraft that keeps the snow away.

I've got a friend that worked on implementing the algorithms for this. She's really smart, but weather prediction is a good example of a system with non-linear dynamics.
 
2013-07-26 02:41:28 PM
Plus, there's a lot of real short memories out there.

It wasn't all that long ago that hurricane forecasting was little more than a dartboard-esque  guess. As someone living in Floriduh, I've really come to appreciate how much better they've gotten at this in the last decade or two.

Now, they're close enough that I know 4, or even 5 days, ahead of time if I need to get out the plywood and make new shutters. Makes life a hell of a lot easier. Back when I got here in the 90s you didn't know until about 48 hours out, and even THEN the area of possible landfall was so big that every Lowe's/Ho Depot/Ace Hardware would be a plywood fistfight pretty much instantly.
 
2013-07-26 02:50:36 PM

born_yesterday: HotIgneous Intruder: weather channels will still be all death and disaster all the time, in between shows with guys shooting ice bergs and people whacking rocks with hammers.

The winter storm forecasting has become so apocalyptic in our area that I can't take it seriously anymore.  Multiple times over the past few years, they've predicted heavy snowfall (6-8 in.; heavy for the DC Metro area), and we haven't gotten so much as a farking flake.


Twice this last winter I got 4"+ of snow while DCA and downtown got zilch. And I live inside the beltway on the west side. Temperature were just that borderline.

Forecasts are actually very good, but the high profile events often have sharp gradients that make the perception seem worse than reality.

/the media hype is awful
 
2013-07-26 02:54:14 PM

born_yesterday: HotIgneous Intruder: weather channels will still be all death and disaster all the time, in between shows with guys shooting ice bergs and people whacking rocks with hammers.

The winter storm forecasting has become so apocalyptic in our area that I can't take it seriously anymore.  Multiple times over the past few years, they've predicted heavy snowfall (6-8 in.; heavy for the DC Metro area), and we haven't gotten so much as a farking flake.


Snow forecasting in my neck of the woods (near Portland, Oregon) is also hit-or miss, and a few miles can mean the difference between a trace and several inches. If you read they NWS forecast discussions you'll find that the more experienced meteorologists often question the computer models if they think the models are off (usually too likely to predict a snowfall). On the other hand, the lead stories in the local media are typically OMG SNOWPOCALYPSE!!!
 
2013-07-26 02:54:31 PM
do they use COBOL?
 
2013-07-26 03:27:10 PM
During Sandy, the European Center had computers capable of producing resolution at 16 kilometers, with 130 vertical levels, according to NWS officials. In nine months, the NWS expects to be at 13 kilometers resolution.

SO YOU'RE STILL GOING TO BE BEHIND IN 9 MONTHS???? AND YOU'RE PROUD OF THIS???
 
2013-07-26 03:34:11 PM
Did a presentation last year discussing our failings versus the Euro models, and that we desperately needed computer upgrades... hooray for Christmas in July!

xaks: It wasn't all that long ago that hurricane forecasting was little more than a dartboard-esque  guess. As someone living in Floriduh, I've really come to appreciate how much better they've gotten at this in the last decade or two.


Heck, the NHC didn't even put stock into forecasting beyond 3 days before 2001.

www.nhc.noaa.gov
 
2013-07-26 04:04:51 PM

Mitt Romneys Tax Return: Snow forecasting in my neck of the woods (near Portland, Oregon) is also hit-or miss, and a few miles can mean the difference between a trace and several inches. If you read they NWS forecast discussions you'll find that the more experienced meteorologists often question the computer models if they think the models are off (usually too likely to predict a snowfall). On the other hand, the lead stories in the local media are typically OMG SNOWPOCALYPSE!!!


THERE"S SOMETHING ABOUT THE NWS FORECAST DISCUSSIONS THAT MAKES READING THEM SORT OF A PAIN IN THE ASS ALTHOUGH I AM NOT SURE I CAN PUT MY FINGER ON EXACTLY WHAT IT IS
 
2013-07-26 04:13:01 PM

you have pee hands: Mitt Romneys Tax Return: Snow forecasting in my neck of the woods (near Portland, Oregon) is also hit-or miss, and a few miles can mean the difference between a trace and several inches. If you read they NWS forecast discussions you'll find that the more experienced meteorologists often question the computer models if they think the models are off (usually too likely to predict a snowfall). On the other hand, the lead stories in the local media are typically OMG SNOWPOCALYPSE!!!

THERE"S SOMETHING ABOUT THE NWS FORECAST DISCUSSIONS THAT MAKES READING THEM SORT OF A PAIN IN THE ASS ALTHOUGH I AM NOT SURE I CAN PUT MY FINGER ON EXACTLY WHAT IT IS


Lol... yes. Our local office (NWS LCH - SWLA & SETX) has some late and weekend shift forecasters that try to mix it up a bit so it's not so boring, but they're pretty accurate and not all ZOMGWTFBBQ doom and gloom about weather events.
 
2013-07-26 04:20:45 PM

you have pee hands: THERE"S SOMETHING ABOUT THE NWS FORECAST DISCUSSIONS THAT MAKES READING THEM SORT OF A PAIN IN THE ASS ALTHOUGH I AM NOT SURE I CAN PUT MY FINGER ON EXACTLY WHAT IT IS


...RED FLAG WARNING REMAINS IN EFFECT FROM 4 PM MDT /3 PM PDT/THIS AFTERNOON TO MIDNIGHT MDT /11 PM PDT/ TONIGHT FOR LIGHTNINGFOR BURNS BLM AND VALE BLM...WHICH ARE FIRE WEATHER ZONES 636 AND637... Any better in lowercase?

...red flag warning remains in effect from 4 pm mdt /3 pm pdt/
this afternoon to midnight mdt /11 pm pdt/ tonight for lightning
for burns blm and vale blm...which are fire weather zones 636 and
637...
 
2013-07-26 04:27:17 PM

you have pee hands: THERE"S SOMETHING ABOUT THE NWS FORECAST DISCUSSIONS THAT MAKES READING THEM SORT OF A PAIN IN THE ASS ALTHOUGH I AM NOT SURE I CAN PUT MY FINGER ON EXACTLY WHAT IT IS


OUR ADVANCED TECHNOLOGY SNEERS AT YOUR LOWER CASE LETTERS

www.columbia.edu
 
2013-07-26 04:39:12 PM

Towermonkey: Our local office (NWS LCH - SWLA & SETX) has some late and weekend shift forecasters that try to mix it up a bit so it's not so boring, but they're pretty accurate and not all ZOMGWTFBBQ doom and gloom about weather events.


Relevant.
 
2013-07-26 04:43:32 PM

E_Henry_Thripshaws_Disease: do they use COBOL?


HWRF is mostly written in FORTAN, as is GFS.
 
2013-07-26 04:49:55 PM
The tornado forecasting actually seems to be getting pretty damn old as of late. The day before the big OKC outbreak this year, the weather experts on here were warning that some bad shiat was headed for that general area the next day. Maybe they didn't get the exact city nailed, but the message was loud and clear to watch your ass if you lived there.
 
2013-07-26 05:10:04 PM

Walker: During Sandy, the European Center had computers capable of producing resolution at 16 kilometers, with 130 vertical levels, according to NWS officials. In nine months, the NWS expects to be at 13 kilometers resolution.

SO YOU'RE STILL GOING TO BE BEHIND IN 9 MONTHS???? AND YOU'RE PROUD OF THIS???


I think you're confused about something. 13 is better than 16. Or am I missing something?
 
2013-07-26 05:11:31 PM
I had a short, but interesting discussion with the folks at the NHC about this, asking them about the assorted differences between the systems.   It seems that even with identical datasets and identical software, the two systems output slightly different results, owing to subtle differences in the hardware architechure (previous system was based on IBM POWER6 chips, new system is based on x86-64) and different compiler optimizations.

It is a pretty damn good example of the butterfly effect.
 
2013-07-26 05:12:09 PM

Walker: During Sandy, the European Center had computers capable of producing resolution at 16 kilometers, with 130 vertical levels, according to NWS officials. In nine months, the NWS expects to be at 13 kilometers resolution.

SO YOU'RE STILL GOING TO BE BEHIND IN 9 MONTHS???? AND YOU'RE PROUD OF THIS???


notsureifserious.jpg
 
2013-07-26 05:25:16 PM
Is there east coast satellite still down
 
2013-07-26 05:33:36 PM

Rossi_84: Is there east coast satellite still down


Back online as of June 10th.
 
2013-07-26 05:38:42 PM

xaks: Back when I got here in the 90s you didn't know until about 48 hours out, and even THEN the area of possible landfall was so big that every Lowe's/Ho Depot/Ace Hardware would be a plywood fistfight pretty much instantly.


Plywood Fistfight is the new name of my Argent cover band.
 
2013-07-26 05:53:31 PM

xaks: Plus, there's a lot of real short memories out there.

It wasn't all that long ago that hurricane forecasting was little more than a dartboard-esque  guess. As someone living in Floriduh, I've really come to appreciate how much better they've gotten at this in the last decade or two.

Now, they're close enough that I know 4, or even 5 days, ahead of time if I need to get out the plywood and make new shutters. Makes life a hell of a lot easier. Back when I got here in the 90s you didn't know until about 48 hours out, and even THEN the area of possible landfall was so big that every Lowe's/Ho Depot/Ace Hardware would be a plywood fistfight pretty much instantly.


I have vague memories of a central Gulf of Mexico storm in the late 80s (I think) that was forecast to hit somewhere between Corpus and the TX/LA border, but at the last minute decided to hang a left, accelerate, and plow into Northern Mexico.  Of course, my memories could be completely fabricated by my own psyche.
 
2013-07-26 06:47:23 PM
i2.listal.com
/heartbroken
 
2013-07-26 08:02:47 PM

ds_4815: Did a presentation last year discussing our failings versus the Euro models, and that we desperately needed computer upgrades... hooray for Christmas in July!

xaks: It wasn't all that long ago that hurricane forecasting was little more than a dartboard-esque  guess. As someone living in Floriduh, I've really come to appreciate how much better they've gotten at this in the last decade or two.

Heck, the NHC didn't even put stock into forecasting beyond 3 days before 2001.

[www.nhc.noaa.gov image 850x637]


Your profile says you're Canadian. So by "our" failings, do you mean the CMC or the GFS? Both still shiat the bed compared to the ECMWF (Euro), and they will continue to do so until their data assimilation algorithms are brought more into the 21st century.

P.S. I miss the days of the CMC being trigger-happy on cyclogenesis along the east coast. Hurricanes all summer, nor'easters all winter, and both (!) in the fall
 
2013-07-26 08:18:28 PM

hershy799: Your profile says you're Canadian. So by "our" failings, do you mean the CMC or the GFS? Both still shiat the bed compared to the ECMWF (Euro), and they will continue to do so until their data assimilation algorithms are brought more into the 21st century.

P.S. I miss the days of the CMC being trigger-happy on cyclogenesis along the east coast. Hurricanes all summer, nor'easters all winter, and both (!) in the fall


My bad... got my degree in Florida, so I'm referring to the GFS, though the CMC always appeared to run fairly equal for results.

(Top: comparison of accuracy of global models, 5-day point forecast @ 500mb. Bottom: Accuracy departure from GFS. CMC is in blue.)

i.imgur.com

/CMC's still pretty trigger-happy for tropical genesis as well
 
2013-07-26 10:21:20 PM
Still can't change the weather.
I haven't had a decent weekend to swim in the pool yet this year...
Weekdays are shot, with work and all.


Yeah, yeah, yeah...I know, we need the rain...STFU, I need my rays.
 
2013-07-26 10:59:04 PM
Can anyone explain why the US is leasing its systems while the Europeans are buying theirs?

/besides, durrr...money...what is the economics of this like?
 
2013-07-26 11:00:47 PM

ds_4815: Did a presentation last year discussing our failings versus the Euro models, and that we desperately needed computer upgrades... hooray for Christmas in July!

xaks: It wasn't all that long ago that hurricane forecasting was little more than a dartboard-esque  guess. As someone living in Floriduh, I've really come to appreciate how much better they've gotten at this in the last decade or two.

Heck, the NHC didn't even put stock into forecasting beyond 3 days before 2001.

[www.nhc.noaa.gov image 850x637]


9/11!!!!
 
Displayed 31 of 31 comments

View Voting Results: Smartest and Funniest


This thread is closed to new comments.

Continue Farking
Submit a Link »





Report