This is another long read, with some SCIENCE!, but the details of this matter – I hope you will be patient and read on. While I often touch on hurricane track forecast models, I normally stick with the official NHC forecast track when discussing the impacts of storms. There are a lot of good reasons for doing this. First, track forecasting is a very specialized profession, with a lot of complex factors involved. NHC has specialists to do that, and while I sometimes disagree with details, they are very good at it. Interpreting the multitude of track models is not an easy task. Second, given it is their responsibility to do official watches and warning, diluting their message to the general public is not a good idea. Some may say this is a bit hypocritical on my part, given my criticisms of Emergency Managers and evacuations, but there is a key difference – NHC are professionals who consistently do a good job. Governors issuing evacuations orders have a much spottier track record, in my clearly not so humble opinion 🙂 , so I feel free to trash them.
Unfortunately, there is a trend in the media to talk about specific track models at the expense of the official forecast, and to publicly second guess or anticipate the next NHC forecast. Part of this is because beautiful, dynamic animations are available for individual track models. These graphics are great for TV, and the temptation to show a graphic that passes over or near your viewing areas is overwhelming. But it can also be very misleading. So let’s take a closer look at track model outputs for Florence, and what they mean. Here’s the track model map available at about 4am this morning:
The Global models (exemplified by the dark blue GFS line show the storm pulling up short of the NC shoreline before diving south. The dedicated hurricane models such as HMON (the replacement for the GFDL, shown in light green), the HWRF (shown in light blue) and the official track (red lines) don’t have the sharp right turn. There are a lot of reasons for this. For the objective models, it is important to realize that while they are often good at forecasting hurricanes, and are used to provide “boundary” conditions for specialized models, they are not specialized, high resolution track forecast models. They often lack the detailed coupling to ocean models, and have less sophisticated model physics than the dedicated models. This gap has closed in recent years as global models have gotten better, but the gap remains. In their forecast package development, NHC takes in to account factors like this using their involvement in, and detailed knowledge of, these models. They also realize that models can shift radically from one iteration to the next. Every 6 hours, data from weather stations and satellites is collected globally and used to create a picture of what the atmosphere looks like at that time. These are the initial conditions. Those initial conditions are then used to spin up the global models, as well as the dedicated tropical cyclone models. That’s the simple version, but the details are very complex. Not all initial conditions are created equal. One small example: the 00Z (8pm ET) and 12Z (8am ET) initial conditions are often “better” than the 06Z (2am) and 18z (2pm) data sets. This is because most upper air observation sited only launch two balloons per day, at 00 and 12z. Sometimes you get a satellite pass over a storm at the right time, sometimes its 3 or 4 hours earlier. Sometimes you have airplane data, sometimes not. So when I fuss at meteorologists, weathercasters, and storm watchers for latching on to a model shift, that’s one reason why. It’s also why the NHC sometimes seems slow to shift tracks to follow what seems to be shifting model data – they realize that it can just as easily shift back 6 or 12 hours later. There is much more to it than this, but I hope that gives you at least a glimpse into the fact this isn’t simple, or for amateurs.
Another factor is that the storm isn’t just some line on the map: it’s a broad, complex system. Let’s look at the impact from two very different model simulations. Here’s the HMON run, and for comparison, the CTCX run, which shows the storm hitting Savannah, GA …
Note especially the impacts onshore. If you compare the areas of severe impacts, despite the fact that the track line is scary for people in Savannah and southern SC, the dangerous, life threatening impacts are all contained within the warning areas NHC has set. Now here is the impact estimate based on the official track:
What about economic impacts? On the official track we are still looking at about $12 Billion in direct impacts. The HMON track would product “only” $5 or $6 Billion, while the CTCX track would likely cause $3 or $4Billion, because the stronger winds would stay offshore!
And what about the infamous “cone of uncertainty?” I see people making much of this or that city being “touched by the cone.” Well, to be honest, who cares. I hate that graphic. What you should care about is how strong the storm will be if and when it gets to you. As NHC notes, bad impacts can exist outside the cone. But the opposite is also true. If the death cone touches you on day 5, when the storm is inland and barely even a tropical depression, and you probably wouldn’t even know it was a storm if somebody didn’t tell you, so what?
So despite all the rending of garments and gnashing of teeth, because of the intensity of the storms on the scary scenarios showing a turn along the coast, the storm would weaken fairly radically, and the impacts would be less. So the conservative path for NHC is to continue to slowly shift the track south (assuming the forecasts continue that trend), adjust the intensity accordingly. And, as you can see, the present watch and warning areas are in fact perfectly fine for this storm.
Bottom line: yes, the forecast tracks are shifting. Yes, forecasting is not an exact science. But it’s not really as big a deal as some would like to make it out to be. Stay cool …