Editorial Comment on the Flood Forecasts for Helene
I've had a half dozen articles forwarded to me starting this morning that contain various complaints pertaining to the forecasts for Helene. I'm not sure what to make of them: with one exception (which I noted at the time), I thought the forecasts were quite good.
That exception was that the NWS was far too slow to pick up on the threat of catastrophic flooding in the Appalachians. But, that was noted on this blog at the time. Published here on the 24th:
Meteorologist Dr. Marshall Shepherd wrote in Forbes today:Addition Sunday Evening: From Virginia Postrel this evening.
As noted above and below, that may have been true pertaining to the National Weather Service but it certainly was not true here nor was it on at least one other site that I viewed last week. I wrote on the 25th,Please look at what I published farther down in the same piece. The red X is in the original.No areas of catastrophic flooding [pink] were on that map nor were there forecasts of "major" or record flooding on the major rivers [dots] in the Carolinas or Georgia. Atlanta endured a rare "flash flood emergency" from Helene's rains but Atlantans' would never know it was coming from by looking at that forecast map!
On May 1, the NWS fixed something that wasn't broken: the method of generating flood warnings for our nation. That is when the "National Water Center" in Tuscaloosa got put into the mix. To try to be polite, I've not seen anything positive from the Water Center in terms of forecasts and warnings. They are nearly always behind the meteorological curve and the their maps are often a confusing mess.
I understand the system for forecasting floods in the NWS currently works something like this:
- Forecasts for local rivers are made by the 134 NWS field forecast offices. They work with the GFS which is the worst of the national models. While model improvements are planned, they cannot come soon enough!
- The 13 River Forecast Centers (map below) still exist but I'm not sure their role is because.....
- ....the National Water Model is run in Tuscaloosa at the National Water Center. But, neither Tuscaloosa nor the River Forecast Centers are 24/7 operations! To an Atlanta resident interested in, say, Peachtree Creek, you don't go to the Atlanta NWS office any more. You have to go to the National Water Center's site and the new format of the flood warnings is confusing (at least to me).
- The rainfall amount forecasts come from the Weather Prediction Center in Washington (actually, College Park, MD).
- The National Hurricane Center in Miami adds its comments (that is where the "isolated flash flood..." comment above came from) to the hurricane advisories.
This is a Rube Goldberg flood forecasting system if there ever was one! This vital responsibility is spread out over too many organizations in too many locations, some of which aren't even staffed 24/7!
The above stated, there were flash flood watches and accurate forecasts from the NWS 24 hours in advance, this blog was calling for major flooding three days in advance and "catastrophic" flooding two days in advance, and Jim Cantore's Twitter account was also calling for severe flooding. I'm sure that at least some of the local broadcast meteorologists had to be forecasting the same.
One of the items I received by email today explicitly said that the logistics network, due to the severing of interstate highways and major U.S. highways, was going to take "weeks" to unscramble because there was no advance warning to logistics organizations. If you go to LinkedIn which is a B2B site, you can see "logistics" highlighted.
Note that this was published four days ago.If you think you are detecting a bit of frustration in this -- you are. It is unclear to me what more we can do to get the word out.
So where it the perception the flood aspect of the storm was inadequately forecast coming from?
I don't know but I will offer a few guesses:
- "Failure of imagination" -- even though we mentioned closed roads, toppled trees and power out for "days or even weeks," it is possible that people simply couldn't personalize that information and so didn't proactively evacuate (even though that was suggested multiple times here).
- "Experience worked against them." This is a real problem identified by sociologists. A hypothetical person is 45 years old and lived in their area all of their life. "I've been through floods before and come out okay, those meteorologists always exaggerate the threat." Can't tell you the number of times I heard that during my 50+ year career. This is not blaming the victim, it is explaining something social scientists have found in a number of investigations of situations like these -- which have often involved hurricanes.
- Failure to pay attention. I have been shocked the number of times people have been preoccupied with their lives and simply haven't tuned in weather information before a critical event occurs. With WEA alerts (the ones that automatically sound obnoxious tones on your radio/TV whether you want them or not), tone-alert weather radios, sirens, etc., we think we are getting to everyone but maybe we aren't.
Again, these are just guesses. I'm hoping social scientists will look at this event. With all of these warnings a day or more in advance, it is hard to understand how 54 hospital employees and patients ended up on the roof of the hospital surrounded by floodwaters.
Hope this post stimulates something and assists in finding better ways to communicate critical information about pending catastrophic storms. I invite comments related to this topic.
NWS meteorologist here. Your description of the NWS rainfall and flood forecasting system is incorrect. Forecasts for river levels are made by the River Forecast Centers, in coordination with the local office hydrologists. The river forecasts are then passed on to the local offices, which use those RFC forecasts to issue any warnings. The RFCs are staffed 24 hours when there is active flooding in their area. Forecast rainfall amounts are generated first at the Weather Prediction Center (as you state), and may be adjusted by the local forecast offices at their discretion to improve for local topographic effects. We use a blend of models, not just the GFS, as you say. The National Water Center issues Flood Outlooks, National Hydrologic Discussions, and Flood Inundation Mapping Services. I hope you will correct your statements.
ReplyDeleteMike Smith here. I appreciate the NWS's meteorologist's comments.
DeleteHowever, the NWS River Forecast Centers are NOT routinely staffed 24/7 and I have personally been through major nighttime flash flood situations as they evolved into river flooding and the local NWS meteorologist told me no one was at the RFC.
That said, I firmly stand by conclusion that the NWS's flood forecasting system has deteriorated in quality the past year.