NOAA Goes 'Live' with New Weather Supercomputers

[CommanderFrank]

NOAA has beefed up its ability to forecast the weather, especially hurricane forecasting during this part of the year. The new super-computers will be 15% more accurate in weather forecasting than the less powerful systems they will be replacing. At a cost of $20 million a year to lease and operate the system, the burning question here is will the new computers be able to run Crysis?

The supercomputers are each 213 teraflop systems, running a Linux operating system on Intel processors.

 

[tubular]

Rofl, nice commentary Domo

 

[Krenum]

How long until NOAA becomes self aware?

 

[The Recessionist]

Rofl, nice commentary Domo

Agreed!

*waits for the inevitable debate to begin.*

 

[CreepyUncleGoogle]

Sooo...NOAA is 15% more accurate? I'm thinking that accuracy increase is about as good of a guess as most of their weather forecasts. Like how they've been predicting a 70% chance of afternoon thunderstorms for where I live for the past six days and it's hasn't even rained yet.

 

[kbrickley]

Sooo...NOAA is 15% more accurate? I'm thinking that accuracy increase is about as good of a guess as most of their weather forecasts. Like how they've been predicting a 70% chance of afternoon thunderstorms for where I live for the past six days and it's hasn't even rained yet.

The problem with weather prediction is they have to use the Navier-Stokes partial differential equations to try and predict the changes of the atmosphere that affect weather ... since partial differential equations are generally unsolvable you must try and use accurate approximations (and the more accurate the approximation, the more computing power that is needed to calculate it) ... the trickiest part of partial differential equations is they are input dependent (your input parameters affect the output calculation) and the atmosphere is constantly changing the variables you need to input into the equation (especially when you try and predict more than 12 hours in the future)

On a side note, there is a million dollar prize for discovering an analytic solution to the Navier-Stokes equations so closet mathematicians channel your inner Sheldon and have at it

 

[Chunder]

It's nice to see our weather prediction services being updated. However the biggest thing we can do to improve our weather forecasting is to actually update our satellites that are long past retirement. European weather forecasters are able to more accurately predict where hurricanes are going to hit the US almost an entire week before we get our more inaccurate spaghetti models up.