Here's some from an interesting piece by William Pentland (who looks a lot like Thomas Dolby) on Forbes.com's Clean Beta energy blog.
Basically, using utility giant Con Edison as his example, the piece calls the effectiveness of IBM's super-duper weather forecasting program called Deep Thunder. (Whoever names stuff at IBM must have a fun job. Heh)
I point this out mainly because a 2010 case study by IBM says at that time, Deep Thunder was running on an IBM System P server at the IBM Data Center in Rochester.
I wonder how KTTC's Randy Brock would fare in a head-to-head weather forecast cage match? I bet he'd do better than those guys did on Jeopardy! against IBM's Watson.
Anyway, here some from Pentland's piece:
Deep Thunder couples high-resolution weather forecasts with business processes to enhance operational planning and facilitate efficient and effective decision making processes. Or, as IBM puts it: “It is not about weather but integrating forecasts into decision making to optimize business processes.”
In 2001, a prototype was pioneered to provide 24-hour forecasts for the New York City area at 1 kilometer resolution twice daily. More recently, IBM has extended the Deep Thunder platform for forecasting local weather conditions in Atlanta, Chicago, Kansas City, Baltimore and Washington D.C. and other metropolitan areas.
-------------------
The big question for IBM is whether Deep Thunder delivers on its promise. At least one high-profile customer, Consolidated Edison of New York, has concluded that it does not.
Here's what Con Ed said in 2008:
[Con Edison] has initiated a project with IBM to apply IBM’s Deep Thunder technology to the problem of forecasting weather-caused damage at a micro-geographic level. The goal . . . is to develop and integrate more precise weather forecasting capability into our emergency response management utilizing finer resolution forecasting models. The training of key operations and planning personnel commenced on June 1, 2008 and continues as the application is modified. The Deep Thunder forecasting model is being included as appropriate in operational decisions to continue to test against real time weather data for accuracy and refinement. We are now developing the specifications and implementation for specific customizations of the current Deep Thunder capability to enable the Company to evaluate “predicted” overhead system damage based on Deep Thunder enhanced weather forecasts.
Then this is a follow up report more than two years later:
“[i]nvestigation into integrating the Deep Thunder micro-weather modeling system with the STAR system determined that it is not feasible with the current software . . The [Deep Thunder] predictive model has proved to be no more accurate than the existing static matrix currently in use. The model’s usefulness is limited and no further modifications are anticipated.”
Recent Comments