Hacker News Reader

I personally feel like the "spaghetti bundle" representation of the data (where a random sample of potential paths are drawn) was the most straightforward image of the uncertainty embedded in the forecast. Has it been considered as a means of public communication?

View on HN

I've just implemented this in Flowx[1], the Android weather app I develop.

I have the "spaghetti bundle" but it was a major improvement when I added dots indicating the location of the hurricane center with time, as shown in this YouTube video[2]

[1] https://play.google.com/store/apps/details?id=com.enzuredigi...

[2] https://www.youtube.com/watch?v=I46IJcnf_bY

BTW, these are ensemble models for the NOAA GFS (FV3) and CMC GDPS models. Ensemble models are where the starting conditions of weather simulation is perturbed slightly and the simulations re-solved. They do this 20-odd times and you end up with 20-odd tracks. So for both GFS and GDPS there will give 40-odd paths.

Sometimes, it can be interesting how the cluster of GFS tracks may differ from the GDPS tracks.

These look very cool!!

Reply on news.ycombinator.com