First a quick review – Douglas skimmed just north of the Hawai’ian islands, causing much less damage than anticipated as only the weak side of the storm swiped them. In Texas, as anticipated Hanna caused power outages, scattered damage, and some flooding. As for the “investigation area” in the Atlantic, 92L, it continues to move across the Atlantic as an elongated tropical wave. Here is what it looks like this morning, with the Tropical Analysis and Forecast Branch surface analysis as an overlay. As usual, click to expand and see the details …
NHC still gives the system a 90% chance of becoming a storm in the next 5 days, with a completely unpronounceable name (Isaias) unless you are a fluent Spanish speaker. So of course the usual suspects are sifting the entrails and trying to see where the computer model tracks are going. Here is the track model map as of 6am this morning:
When looking at these maps it is vital to realize that they are lumping together models that probably shouldn’t be put together (like purely statistical models, as well as individual ensemble members). You will hear terms like “ensemble runs” and “ensemble members”. Just what does that mean? Here are two examples. The problem with a storm is that even in a stronger storm the exact position, and the surrounding environment, isn’t perfectly known (much less an invest area – look at the satellite map above and pin the fix on the storm!). What an ensemble run does is start the storm in slightly different positions and intensities (perturbed initial positions is the technical term), and re-runs the model to see what happens. Here is one example, the US Global Forecast System (GFS) run:
Notice the initial position for each track line is different. The blue line is the “main” GFS forecast run, while the brown line is the average of the ensembles. Notice how much variation there is? (Actually, this isn’t bad for a weak system!). Now, let’s look at another example, the Canadian model:
Again, for AL92, these are really rather consistent, which indicates a more stable environment (or, which is possible, they are all wrong in the same way!). The European model looks similar (but because of licensing restrictions I can’t show you that map) as do the Navy’s model and others. The key point is to again reinforce the fact you can’t just pluck one model track – even a good model – out and get terribly excited about it (much less bet your life on it one way or the other). This data has to be properly interpreted, and by far your best bet with respect to publicly available data is the official NHC forecast.
Administrative note: As previously noted, I don’t have funding to do these posts and commentaries. The last couple of months I’ve been trying to figure out how to do them that covers the costs – and even expand the features. I don’t want to run advertising, so on August 1st I’m going to start a Patreon page where those of you who read all the way to the end can contribute to keeping these things running and help support the research work we do here, and get extra information including site specific forecasts from our commercial system. I think we can do some pretty cool stuff, especially trying to get “plain English” no-hype forecasts for hurricanes and eventually other severe weather. I hope you’ll think about contributing.