Researchers Create Extreme Weather Early Warning System from ‘Obsolete’ Tool

By | February 6, 2020

  • February 7, 2020 at 10:05 am
    PolarBeaRepeal says:
    Like or Dislike:
    Thumb up 1
    Thumb down 2

    With a quick reading, I found one way to improve the accuracy of this forecasting approach above the 85% claimed. Key meteorological variables measured at set geo and temporal points could be iteratively measured at varied geo and temporal points to optimize the accuracy of the machine learning process above 85% accuracy. The authors of the study may have done exactly that, so I used the words ‘could be’ rather than ‘should be’.

    To clarify the comment above, the height of the air pressure measurements was given to be ‘5 km’, and the temporal gap was ‘several days apart’. Those two parameters could be varied iteratively in model runs to find the optimal model. A question that arises with regard to the temperature measures at ground level is the altitude (above sea level) of the ground. I believe the altitude may also affect the accuracy of the model that uses many points on a map… which have different altitudes. Topography is important as it affects winds and cloud movements.

    The article mentions in paragraphs 5 and 20 that it is in its infancy; i.e. “With further development,…” and “Though much more work is needed before Rice’s system can be incorporated into operational forecasting,…”, respectively.

    I await another article on this subject in the future.

    (I anticipate down votes on this post by people who didn’t read it or the subject article.)

    • February 7, 2020 at 1:42 pm
      Craig Cornell says:
      Like or Dislike:
      Thumb up 0
      Thumb down 0

      Good comment, Polar. And your prediction on Down Votes nailed it.

      (Great book that defines most people who post on IJ: The Smallest Minority. Explains perfectly why so many trolls and haters have infected social media, and why the new media now reflects the same thing.)

Add a Comment

Your email address will not be published. Required fields are marked *