Sound impossible?
Back in 2001, I spotted a weak signal that pointed to the fact that this forecasting capability could soon be a reality.
Problem:
Weather is usually thought of as a chaotic system, beyond long-term prediction.
Smart Solution:
Researchers at Northwest Research Associates in Bellevue, Washington, have found strong correlations between bad weather in the northern hemisphere, and changes in air circulation in the stratosphere. The stratosphere is between 12 and 50 kilometres above the troposphere, the lowest layer of the atmosphere.
The team says that:
"cold and wintry weather at the Earth's surface tends to follow a weakening of the polar vortex, the stratospheric circulation of winds clockwise around the North Pole. It is easy for weather balloons and satellites to monitor winds in the stratosphere."
Scientists say that for two months after the polar vortex weakens, bad weather is twice as likely as normal across the UK, Central Europe, North America and North Asia.
Benefits
This discovery could provide many weeks or months of warning to retailers, farmers, airline, insurance or oil companies whose business depends on the weather.
ETA: 2015
I contacted the research group to find out their progress since 2001.
Here's what they had to say:
Q: Has your research advanced to the application stage?
A: The answer is that the major forecast centers have not yet implemented such a system. However, I am told that since they are aware of stratospheric effects, they do take stratospheric effects into account when they issue monthly forecasts."
"However, I have a post-doc, who is working on such a system. As I see it, an impediment is that the major forecast centers only issue occasional long-range forecasts. The system we have in mind would be run daily. Ultimately, as computers become faster, it may be possible to run forecast models out to 60-90 days every day. These models would incorporate the stratosphere, so there would be no need for statistical incorporation of stratospheric information. But this is a long way off."
Q: How far off?
A: "I am guessing. Over the next 10+ years models will be used more and more for long-range forecasting. The models exist today; it is really just a matter of computing power. Forecast models are run in "ensemble mode" in which the model is run ~10 times with very slightly different initial conditions. The forecast is the average of the 10 runs. It would be a lot to run the model 10 times out to 60 or 90 days."
© 2005
Walter Derzko
Expert, Consultant and Guest Speaker on emerging Smart Technologies, Strategic Planning, Business Development, Lateral Creative Thinking and author of an upcoming book on the Smart Economy "
To explore the opportunities and threats of any new smart technology in your industry - Contact Me
.....Strategy without action is a day-dream; action without strategy is a nightmare"
- old Japanese proverb
P. S. if this is your first visit to my blog, please go to our Welcome page
Drawing from Ray Kurzweil's assertion that computer performance per $1,000 is doubling annually, 6 or 7 years would represent a 64 or 128-fold increase in computing power. So, even if they can only do their ten current runs for a one-day prediction, they would have the raw processing power at hand. That ignores any parallel improvement in the models themselves.
Posted by: David Walter | September 13, 2006 at 04:02 PM