I mentioned a few days ago that I'm training my very own custom tornado model that blends traditional weather modeling algorithms with quantum entanglement to more accurately predict the timing, location and intensity of tornadoes. The objective is to create a model that can predict individual tornado tracks, timing and intensity within a 5 mile margin of error and with a <15 minute margin, hours before the event itself.
I'm happy to report the results so far have been quite promising, although more refinement is needed. Here's the projected risk map from an individual event I backtested (December 10th, 2023, the Mayfield Tornado)
The image below shows the model's prediction of the event roughly 12 hours before the main tornado forms (about the same time as an SPC 06Z update, or the first D1 forecast).
Obviously there's some noise/data leakage that needs to be addressed (the big 2nd risk are on the left is a clear false positive), but the model produced a highly accurate forecast for the tornadogensis coordinates of the Mayfield tornado.
I'll continue refining the model and will start running it on upcoming events!