Slightly offtopic, but the windy.com service referenced in the article is really a fantastic resource if you learn to use it properly. I rely on it for paragliding a lot, the animated winds are much easier to understand for the brain than static map. Comparing different weather models helps to assess the accuracy of the prediction.
(This is a bit off-topic from the subject of the article about particular methodology of rendering data as opposed to usage of these kind of wind maps)<p>I worked on X’s project loon in operations for a spell. We were interacting with balloons in flight regularly. The referenced nullschool wind map was unendingly useful.<p>Something I always wanted for using nullschool or other similar publicly available “tools” was more granularity between wind layers, or derived estimations of data between wind layers.<p>When putting any flight systems in the atmosphere, having visualizations (even estimates) of wind direction and speed estimations at more altitude levels is more valuable than visualizing more particles more efficiently, IMO.<p>I wish similar thought and processing power was put towards smoothing out guesses at wind speeds at different altitudes.<p>Tl;dr: I wish this demo map had an altitude slider, even if it was smoothed out guestimates between available data layers.