Blizzard 2015: Blame, Modelology, and The Art of Snow Forecasting

Note…my comments here are my own, based on what I witnessed in the forecasts from the Northeast, what I did myself, what I told friends/family that lived in the Northeast, and what I observed on TV and in social media and what I have learned in my 10+ years in the field. I’m not the authority on this. No one is. But I saw enough to feel I can make an informed judgment.

Let’s start with a few things.

There’s no reason to “blame” anyone today. Blame is a pretty harsh critique and implies that one individual or group caused an action. No one should blame the meteorologists, the media, the politicians, or the weather models. No one deserves blame for today because today’s snowtastrophe in New York City/NJ was a combination of a lot of issues we all have to contend with, regardless if we’re in the media, in the private sector, a viewer, a consumer, a decision maker, anyone.

First, let’s get this out of the way. This was a glowing success for forecasters in Eastern New England. The models did well there for the most part, and we saw a ferocious monster of a storm, as was predicted. But we need to stop saying that this was a “good” forecast that only missed a city by 40 miles. Once the weather model data locked in on this storm, the fate of Eastern New England really wasn’t in doubt. But the fact is, the forecast failed over a city of 8.5 million people, and the nation’s largest metropolitan area, the nation’s most densely populated state, and pretty much the epicenter of the country. We shut it down for a glorified nuisance storm. The negative reaction is warranted because for the largest urban center in America, the forecast failed.

Let’s not blame politicians. I am almost always on the side of my fellow meteorologists when the old battle of a politician blaming the meteorological community comes up. It happens far too much because of selfishness, blame deflection, and poor judgment of politicians. But they do not deserve blame in NY or NJ today. The official forecasts that were posted by the National Weather Service called for over 18-24″ for a large chunk of the NYC metro area, if not all of it, and they stuck to their guns til the end. My question to people blaming politicians is “what would you do if you saw that forecast?” We can argue all day about probabilities and risks to the forecast being mentioned to officials, but the simple fact of the matter is…if the NWS or broadcast TV meteorologists are issuing a forecast for 24-30″+ of snow, the presumption is going to be that the folks issuing that forecast are confident in it. Any forecast I issue, I assume is the most likely outcome in a particular situation. That will fail at times for sure. But when I issue it, I have to be confident that I’m putting forth my best effort. If I am a politician or decision maker, I’m going to make decisions based on the best available information, which is presumably a forecast and conversation with NWS*, who will have the best available information at the time. So we can argue if we overreact to snow in general, but in today’s world, the decisions made by the politicians were probably the right ones at the time based on the forecast data as it was presented to them.

*Note: I have an assumption (based on instinct alone) that there may have been a lot of pressure on some folks to pull the trigger and go with something to allow officials to make a decision on what to do. I strongly disagree with that, and I think there needs to be push back on theses elected officials to be realistic. But, we know where our bread is buttered, so I guess there’s not much that can be done there. But if I were NWS in NYC, I’d force a pow wow with government officials and explain to them realities of weather forecasting. This isn’t going to get any better.

Let’s not blame the media. Most of the same points above can be made. We know how the cycle works today. It will be hyped, it will be blowout coverage, it will be ridiculous. But that’s only if the forecast being fed to the “suits” justifies that. And in this case, a forecast of 2-3 feet of snow and potential risk (and explicitly mentioned one) for the largest snowstorm in the history of the largest city in America justified media going all out. Did you expect anything different? We shouldn’t have. Again, the decision the media made was probably the right one at the time, based on the forecast data as it was presented to them.

Let’s not blame the National Weather Service. The people who work at NWS (including good personal friends of mine) are good people. They try hard to do the right thing, and like any meteorologist, they’re going to have a miss sometimes too. So they do not deserve all out blame here. There were a few contrarian voices during this storm, but for the most part, we all went down together on this one.
Don’t blame the models either. I could go back through all the runs of the GFS, the NAM, the Euro, the SREF, etc. during the duration of the storm’s modeled life cycle and probably find at least one run of each that had the right idea. This isn’t a Euro vs. GFS thing. We joke so often that the GFS sucks. Sure, it does a lot. But guess what, so does the Euro. The Euro has been flawed frequently for a long time. It’s considered the gold standard of weather modeling, but maybe it should be considered the bronze standard instead. There is no gold standard. None performed particularly well, but models are models. It’s our job to scrutinize if they may be right or wrong, which brings me to my next point….

Modelology must die. Repeat. Modelology must die. This point is intertwined with another which I’ll make. I’m very guilty of this sometimes too. But we have to stop leaning on models like the Euro or the NAM or the GFS. Models aren’t meant to be gospel. They’re meant to guide a human into using their experience to create what is ideally an objective forecast based on a large amount of evidence to support it, coupled with that personal experience. Throughout this event, we saw Eastern New England getting clobbered run after run. But NY/NJ were getting clobbered, then nothing, clobbered, then nothing. In my experience, that means there’s a problem and you need to dig beyond the models and start blending ideas and throwing in a more subjective “gut” factor (that is real in meteorology and frankly, that’s why some of us get paychecks whether anyone wants to admit to it or not). For my friends and family I started sounding the alarm Sunday night that this might not pan out quite as expected. I got suckered back in Monday morning after looking at the Euro and saying, “Well, this sure has been mostly consistent.” I think most of us felt uneasy about this on Sunday. We need to really try hard to let go of the models, even if they have a good/bad reputation and blend some, fudge some, and use our experience. The message was great for Philly and Southern NJ that this was going to be a close call. I think people there understood it. But in Northern NJ/Central NJ and NYC a lot of people just saw bulled up NWS forecast and the support it received elsewhere and thought this was going to happen, and it was going to be big. Which brings me to…

Snow forecasting is not a pure science. It’s truly an art. And it should be treated as such. When I worked in lake effect country, we dealt with sharp gradients constantly, periodically in the larger cities in that area. I recall one instance that I went all in and went with a 12-18″ forecast for our main city. The band stopped just north and never got its act together. It was 6″ if we were lucky. I learned then never to go all-in on a snow forecast. Someone might argue with me that “well all the best data said feet of snow.” That’s fine. But snow forecasting is as much about managing and communicating a message as it is about getting it right. We’re never going to get it perfectly right. Snow is incredibly difficult to forecast, especially when it falls on a gradient like the one east of NYC. We can’t jump from “maybe at most” 6-10″ to 20-30″ in one day. You have to be gradual and emphasize those risks, both upside and downside. This prevents a “whipsaw” effect like going 8″, 30″, 28″, 24″, 12″. And it gives you so much flexibility as a forecaster.

So this is a lot of words. What is this one person’s opinion on what we should take from this?

  • We live in a new social media reality. We can’t complain about these “weenie” sites making bogus forecasts all the time. Hopefully the fall out from this helps people realize the next time we have a borderline situation and someone shows a BIG SNOW map that they’re just showing models and there’s a good chance it doesn’t evolve that way. In a sense, this storm may help that cause a little. But we need to learn to live with this, and rather than complaining about it, work through it and deal with it.
  • We need to stop talking about research on communication methods and how to do it better and start actually doing it. I think this conversation has been ongoing since Katrina, maybe longer, maybe shorter. I’ve seen a number of mets show simple visual graphics that explain impact risks to different areas and from different variables. Those are tremendously effective in my opinion, and there needs to be more of that.
  • I think the NWS needs to break this regimented product setup. I think at times their hands get tied. They can’t be nimble. And they need to be. Products should target impacts and be able to highlight those impacts and risks above anything else. There needs to be less meeting of a checklist or formatting something under specific guidelines.
  • The NWS and others need to also focus a little less on Twitter. As of last September, 23% of the American adult population used Twitter. Is Twitter great? Yes. But less than 1/4 of the US population uses it! Having awesome graphics on Twitter doesn’t substitute for having a good forecast and successfully communicating the risks to the forecast. Lots of people (over 70%) use Facebook, but the reach is so wonky because of their news feed algorithm. I like a lot of the stuff the NWS produces on Twitter and Facebook, but I’m questioning how much value it ultimately adds.
  • We all need to learn to stop speaking meteorology and speak English to people and be simple. Throwing in cool terms like “bombogenesis” is great, but if I’m the average viewer or consumer, I just want to know what’s going to happen, when. Maybe tell me about how it might change, but I mostly want to know what, where, and when. That’s it. Science lessons are wonderful, but I’m not sure they’re particularly valuable. I used to be vehemently anti “dumbing it down.” As I’ve worked more, I’ve learned that when you “dumb it down,” you’re not actually dumbing it down. You’re providing simple, useful information for a person. And that’s a wonderful thing.
  • I wish respected meteorologists would stop showing model maps at all. I think it plants seeds that can’t be dug back up and creates a preconceived notion no matter how often you caveat it.

There are a number of lessons to be taken from this storm I think. I just hope we’re all humble enough to accept that, take them, and apply them.