Here's the latest reason to not take long-range model output verbatim – these can show signals, especially when coupled with a dispersive ensemble, but systematic biases can still be at play.
Seriously though @splillo's dprog/dt of Caribbean MSLP plot should look great here. pic.twitter.com/qYgKQvrBPX
— Tomer Burg (@burgwx) October 20, 2020
