Spacecoast wrote:TimSmith wrote:Very good point, and also to add to that is the fact that these models generally work as solutions to mathematical differential equations which are very susceptible to small changes in their initial conditions. i.e., a small change to the data fed in regarding the current structure of the storm and/or if it develops a lot more or less as it tracks west over the next few days can have huge impacts on the longer range. ...
Exactly why ensembles were developed...Small changes in the initial conditions (perturbations) are introduced to make up for the fact that the current structure can never be measured exactly. This is due to measurement errors, and equipment tolerances.
Yes, those help and I expect as computing power continues to grow we will gain more information, but we will always have a limitation. IMHO 3-5 days forecasting is doable but with quite a bit of error, beyond 5 is very high uncertainty