Very good post from a local met on forecast error, including an apology. This has caused quite an uproar in our community. Some people lost everything they had and they're lashing out. I feel this is a good explanation. https://m.facebook.com/story.php?story_ ... 2365187416
"Where the forecast failed…
Some of you are angry, upset, or disappointed. I’ve seen the messages and comments, and I totally understand. And I am angry, upset, and disappointed that we didn’t do a better job regarding Matthew’s impacts here in Hampton Roads.
Forecasting tropical cyclones is not easy. A small shift, wobble, or change in a storm’s path, even by 50 miles, can mean the difference between 40 mph winds and 80 mph winds. And Matthew was one of the more difficult storms to forecast. From the beginning, Matthew never did what it was supposed to do. Remember that sudden strengthening in the Caribbean, when it went from a Category 1 to a Category 5 hurricane in 24 hours? Or the shift in the track from Jamaica to Haiti? Or that Matthew was supposed to miss, then hit the Florida coast? It’s not an exact science, and it won’t be for a long time to come.
That brings us to uncertainty. Uncertainty in the track and uncertainty in the effects. We, as meteorologists, need to do a better job of expressing uncertainty to you, the public. Often times, we are met with criticism for being wishy-washy or vague when we try to express the parts of the forecast we are uncertain about. In the past few years, the meteorological team at 13News Now has moved away from showing the “center line” of a tropical forecast track within the “cone of uncertainty” for that very reason: it’s not certain. The storm could go anywhere in the cone, and even, on some occasions, outside the cone. In fact, the current day-five margin of error in a tropical cyclone’s position is still over 230 miles. And intensity forecasts are often off by quite a bit as well. Why? Because the atmosphere is too complex, too dynamic to be pinned down by a bunch of mathematical equations. We are getting better, our computer models are getting better, but we are not perfect.
So what happened? The hook. There was growing confidence by mid-week that the storm would move up the coast, and an approaching cold front/trough would cause the storm to turn east. There was uncertainty as to when exactly that would happen, but nearly all the models called for it well south of Cape Hatteras. And, that’s pretty much what happened. Except for one thing…instead of simply deflecting Matthew away from the area, Matthew began to be influenced by the front in ways nearly all the models failed to see until it was almost happening. Matthew was transitioning from a tropical cyclone into a nor’easter – and those are two completely different storms.
Tropical cyclones tend to be compact storms in many ways, with their strongest winds near the center, the eye. When Matthew was moving up the east coast of Florida, the eye remained a few dozen miles off the shoreline. This is why many areas of Florida saw winds and wind gusts less than ours. Once Matthew made it to the Georgia and South Carolina coasts, we saw the greater impact because the storm came ashore. Nor’easters, however, tend to have large, expansive wind fields. As Matthew began that transition from a tropical cyclone to a nor’easter, its wind field began to expand, spreading north and west. The approaching cold front was then no longer working to push a tropical cyclone out to sea, but was working to enhance a transitioning nor’easter, focusing the band of heavy rain over Southeast Virginia into central North Carolina, and pulling the storm northeast instead of deflecting it into the forecasted hook.
Yes, this was something we should have seen as a possible scenario earlier. There were small signs Friday night that this scenario might happen. In response to those signs, we started increasing some of the rainfall and wind forecasts. There was quite a bit of discussion behind-the-scenes, as the new models came in, that the storm might be starting to change. Was this an aberration in the models? Or were they picking up on something more? And that was the dilemma – when confronted with conflicting information – is the new data in error, or is it showing us something new, something unexpected?
At one point Friday we showed a bar graph that had various rainfall forecasts from four different computer models. Some called for 8, 9, or 10 inches of rain, while others called for 2, 3 or 4 inches. As had been the case with Matthew throughout the storm’s lifespan, we were left with two very different solutions. The time was drawing close and we needed to make the call. The call was made to increase our rainfall and wind forecasts from where they had been on Thursday and Friday morning. Still, it was not enough. To use a sports analogy, we punted. And we didn’t do the best job expressing the uncertainty I mentioned early on. We had caveats to our forecasts, stated verbally, saying if the storm got closer than expected, we would see worse conditions. Unfortunately, those sorts of caveats can’t always be covered in 140 characters on social media.
The one concern we did address, quite extensively in fact, was the saturated ground, and how that would impact the trees and power outages. After coming off of the second wettest September on record, we were worried how the tress, with their canopies full of leaves and their roots in saturated soil, would react to the gusty winds. It was a fine line. Had this storm and these winds come after a dry spell, we wouldn’t see the issues we are seeing now. We learned from Isabel, back in 2003, what a saturated soil can do to trees, particularly pines, when their roots can’t keep hold. While it doesn’t appear to be as widespread as Isabel, we are seeing power outages due to wind-felled trees. And those effects will be felt for several more days.
After each storm, we take a look back at the data and the forecasts and see what we did right, what we did wrong, and where we can improve. We will do that in the coming days and weeks. We will learn from our mistakes.
You trust us and welcome us into your homes each day, expecting us to give you the best and most accurate forecast. Most of the time, we deliver on those expectations. I feel like we failed to deliver on those expectations, and that’s why I am angry, upset, and disappointed today."