House Calls

On the hazards of posting model imagery to social media

This month, we’re going to look at something that happens every winter. And I’m not talking about snow, shoveling, or snowball fights. While those are fun, I’m talking about…posting GFS, NAM and other model imagery of forecast snowfall on social media.


Now, you’re thinking: I’m violating copyright laws by doing that, aren’t I? Actually, if you post still shots or even animations of an event online, it can be very instructive. Posting our satellite, radar and upper air images and animations of an event, along with a good explanation, is something I like to see! Go ahead and do that, with credit to AllisonHouse for the imagery.

But… problems occur when model data is posted without any interpretation. It becomes a real problem… especially in life-threatening, or life-altering situations, when the model (NOT data!) is only going to get it right out of, more than anything, luck. And more often than not, beyond 48 hours, it frequently blows the forecast.

Now: I’m not talking about posting an upper level map of a hurricane’s 850 MB winds posted 12 hours before landfall, with an explanation of what the map is about, and that those winds are aloft. No, you know what I’m talking about:

I’m talking about posting GFS and European model output of snowfall totals of a potential storm system 3 days in advance or more. Without context, without explanation, or even worse: “look what this storm is going to do to (insert area in question)!”.

But why? Isn’t the European model the greatest thing since sliced bread? And since the GFS got upgraded early in 2015, didn’t it nail the big ones so far this year? Isn’t it now doing the strut down the road to the Model Hall of Fame?

Not really. You see: 3, and especially 4 days out and more…those systems, more than likely, are somewhere out in the Pacific Ocean. And all that is there are usually two things for the models to use, to try to figure out what is happening: satellite data, which is of coarse or low resolution, and…ship reports.

That’s it.

That’s like blindfolding someone and walking them to a zoo, and then having them feel an animal, and have them try to guess what it is, and how old it is, with one touch in one area of the animal!

Which is why it is remarkable that the models do as well as they do. Sure: I’ve seen the success stories where a model, 7 days in advance, predicted that storm X was going to produce a swath of heavy snow across part of the country. And it did just that…but was many miles off where the model said it would be. And even if you are only 20 miles off, it can mean the difference between 4″ of snow, and 12″. That’s the difference between everything shutting down, and a somewhat slower commute into work. What that does is when you post that model image without understanding its limitations, is this:  it quickly ruins your credibility, it scares people unnecessarily, and more than likely, the amounts and the path of the heavy snow will be wrong…many times, horribly wrong. And that causes a lot of unnecessary and sometimes very serious problems.

Don’t get me wrong: this extends into the spring, summer, and fall as well. The HRRR model, which you can see via AllisonHouse Maps and GRearth, are of such high resolution that they will resolve a supercell, or rotating thunderstorms, right down to the “hook echo”, like it’s on radar. If you post one of those maps with a supercell over town X, without an explanation, non-weather weenies may think that will hit THEM, when the model is merely suggesting a *general area* for the development of storms, and the model “thinks” that supercells are *possible*.

It doesn’t stop there. Model precipitation totals in the spring, summer and fall are often wildly exaggerated due to a phenomenon called “convective feedback”, where the model mishandles the amount of warm, moist air once thunderstorms develop. Usually seen on days 2 and beyond, but it can be seen even in the first 12 hours: when this happens, the model causes predicted rainfall areas and rainfall rates to go bonkers…without a corresponding and appropriate loss of overall instability, reducing the amount of time a storm spends in one area, or the intensity of the precipitation. The result is wildly over inflated precipitation “bullseyes”, where rainfall amounts are way higher than what one would expect, given the amount of instability and moisture available. As a result, people thinking this would bust a drought they are in would be disappointed, and those in areas where everything is saturated or where flooding is already occurring could be freaked out. And believe me, I have seen that happen many times.

So what, then? Should I never post any model graphics? Of course not. But, if you do, you need to do it wisely. Here’s some guidelines I have learned over the years to help you to do just that:

1. Any graphic you post on Facebook could be shared by others, and frequently, the explanation gets erased as the graphic is shared. That actually happened to me, and I got rightfully yelled at for posting a model snowfall graphic on there. Although Facebook has gotten better about it, they now only show the top 160 characters or so of what you share, and then the rest you have to click through to see the description. Some people won’t do that, or don’t know how, and that causes problems. And yes, it IS your problem when that happens. Scaring people with dangerous weather possibilities is NEVER a good thing to do.

2. European model forecast snowfall graphics are not legal to post publicly on social media! I don’t care if you get it from Accu-Weather, WSI, WeatherBell…or if you got it from someone else. It’s not legal…period! And the ECMWF folks are now taking a dim view of Americans sharing it online, and are starting to come after some people. This is why that data is not found in a lot of places: it’s very expensive, and it’s for personal use only.

3. All other models: don’t post them on social media unless you are 48 hours out or less, and you have reasons why the model is right, or wrong, using sound scientific reasoning. DO NOT say, to the effect, “you’re all in big trouble in this area” or even “well, this model is out to lunch, LOL!” UNLESS you have valid scientific reasoning to support such a claim.

4. IF YOU MUST post about model data showing 12″ of snow for you 7 days out, post it on a private weather forum, along with the questions and concerns that you have. Post the questions (but not the graphic itself) to the National Weather Service social media sites. They’ll get back to you…though it may take a bit, as they have many duties that they must perform. Hang in there, and be patient!

But…I want to warn my neighbors! That’s great! Share current NWS graphics/website links instead, or current data/radar images from AllisonHouse (but tell people what TIME this is occurring; weather information gets dated very quickly!). One of the things we’re learning about in the meteorology world is that one of the best ways to share forecast uncertainty to the public is via the web. You know what: I bet your neighbor or friends would like to know that the NWS is forecasting 8″-12″ of snow, starting tomorrow evening, but the forecast confidence isn’t good. Or maybe it is good. Either way, share their graphics and link to their website. That builds YOUR credibility, and that of the NWS as well.

AllisonHouse is always to be used in conjunction with information, alerts and discussions supplied by the National Weather Service. If you don’t know of an atmospheric phenomena you see on radar, satellite, or a METAR/mesonet site, and it’s happening right now, do ask on the NWS or your favorite social media pages. But, again, putting our forecast snowfall forecasts from the GFS model and saying “SNOWMG!!!” on your Facebook page? You’ll just cause trouble, rather than inform and help your readers.

What we at AllisonHouse want to do is take the weather data and information we receive, put it in your hands in an organized and very helpful way, help you interpret the information as easily as possible, and then make sound decisions from it. If posting one of our graphics (or someone else’s) doesn’t fall into that genre, you may cause unnecessary angst or concern, and bad decision making by people you care for, or even don’t know! Remember: information is not power; the wisdom to use it well is. And we hope by this post, you can be responsible and help others understand what’s heading their way…without causing undue concern and alarm.

Why is my data so bad?

One of the more common questions I’ve received over the past several years working at AllisonHouse is: why is the weather station, at the airport, on an interstate highway, or at someone’s personal weather station always wrong? It’s not -10 degrees, it’s in the middle of summer! Or the winds are always calm, when it’s windy at all the other surrounding stations? What’s going on?

A lot of things can be happening to a station that’s reporting erroneous or missing data. One is a lack of maintenance and calibration. Every year, or more as needed, the FAA and the National Weather Service require that their instruments be calibrated to “reference” equipment: the one by all others are judged to be correct, or not. Some higher end weather stations, such as official weather stations known as Automated Surface Observing Systems (ASOS), are more frequent. At smaller airports, Automated Weather Observing Systems (AWOS), calibration may only occur once per year. Due to age or faulty equipment, the measuring components (thermometer, hygrometer, etc) can break down, causing one or more of the data fields to be missing. Some, due to frayed wires or other issues, can report erroneous conditions. Sometimes spiders will build a web in front of the ASOS or AWOS visibility sensor, causing the station to report haze or fog. This occurs even though a typical AWOS costs around $100,000, and a typical ASOS, $250,000. That does not include maintenance costs, or costs to get the observation via dialup or other secure connection to the FAA. Mesonet stations may also not be calibrated as much, and may have a notable instrumentation failure rate.

Department of Transportation and Army Corps of Engineer stations can be less reliable, because they typically are not maintained as well as the stations as airports. Especially in these days of shrinking budgets, it may take a very long time before equipment or measuring devices are fixed. Furthermore, they are frequently not sited as well, placed on the sides of bridges or concrete abutments, causing one or more of the reporting data to be incorrect. This is a no-no with the National Weather Service and the FAA, since the stations they use must use strict guidelines as to where a station can be placed: away from all buildings and other obstructions, and concrete/asphalt, to avoid a heat and wind bias as much as possible.

Home weather stations are even more problematic. They are almost never sited to National Weather Service standards; placing them on roofs of houses and garages subjects them to a warm bias. They might report close to reality, and may be “good enough” for you and to be useful, but the only maintenance they may ever see is the annual replacement of an internal battery.
If the unit goes down, in many cases, the entire unit is replaced, rather than the failed instrument(s). More often than not, adjacent houses, trees or other entities block the wind from reaching the anemometer properly, and the warm roof sends the temperature upward and significantly above the actual air temperature. While a lot of the technology to build home weather stations has substantially reduced this bias…it is still there.

When any station goes down or has problems, whether it be from an ASOS at a major airport, to a home weather station, there is absolutely nothing we can do to correct the situation. Your best bet is to contact the airport, or owner of the weather station, and tell them how much you appreciate their data. Then explain to them that something is wrong, and if they were aware and willing to correct the issue.

The job of AllisonHouse is to give you the data. It’s up to you to look at it, and decide for yourself if what you are seeing is valid. We would run out of time and not be able to develop products if we were watching for bad stations to “block” from our systems. We let the erroneous data flow to you, and then you can decide if you want to contact the owner to complain. If it is an airport station, it’s usually sending out alerts that it needs maintenance, so I wouldn’t call them…except to see when the troublesome equipment will be repaired, and only if it’s been down for several days. As for the others, just be patient or ignore the data…and hope that it eventually comes back.

Helping those who help us

People are outside on this warm and humid day. Some, on this Saturday, are in town celebrating a wedding. But it’s also time for graduations, and the city is packed with people celebrating Judy’s graduation from the University. Bill and Janice just heard the first notes from Wagner’s “Bridal Chorus”, as the church organist has done for many in the congregation for years. Unfortunately, on the far west side of town, a house is on fire. The firefighters are struggling with containing it because of the sustained southeast winds at 20 MPH, gusting over 30 MPH. They don’t know if the winds will change, and if they shift a little more east, it could impact a storage tank a quarter mile down the road. That makes the fire chief nervous, as the flaming embers are going quite a distance away, and setting spot fires a considerable distance downstream in a cornfield. Meanwhile, the truck stop down by the Interstate is packed with truckers downing an early dinner, trying to figure out how they’re doing time-wise, and when they need to pull off and rest for the night.

Amidst all of that, the low, continuous roar of thunder is getting louder. Over the last 20 minutes, the police become frustrated as dozens of cars have started parking on the side of the road. Some have “light bars”, some have “storm chase team” logos and companies you’ve never heard of. The sky is turning black. The National Weather Service weather radio is beeping about a tornado warning one county to your west. Where is it going? The smart phone app doesn’t show warnings, or where this supposed tornado is. All I see is a lot of red, and suddenly you have a lot of questions.

Emergency management in the United States has had successes, and has had failures over the years. One of the areas where it has really struggled is how to prepare for weather disasters, and deal with them as they are about to happen. When you think of those words, you may think of “Sandy”, or “Joplin”. The fact is, billions of dollars, yes, literally MANY billions of dollars are lost in our economy due to weather disasters, with single incidents within those disasters. Truck drivers who don’t know about the Chinook winds 100 miles ahead that will flip their truck on their side, and land them in a hospital, and destroying their cargo. The roads that cars and trucks will drive over that are washed away by flood waters at night. The commuter system that stops due to severe weather, as the dispatchers are unsure of the scope of how bad things are. The blizzard on the cold side of the storm system bringing everything to a stop. Trucks, trains, cars, people. All of us are affected in a significant way to bad weather.

This month, as you prepare for Christmas, New Years, and other holidays, consider this. Have you prepared for your community, or your business, or yourself, when the weather gets ugly? And most importantly, HOW are you going to know that something bad is going to happen beforehand? Sure, the weather is cold and dry now. It won’t stay that way. You know that.

While there is no solution that fits everyone, the good news is that there ARE solutions! Here at AllisonHouse, we’re here to help you arrive at the solutions that work for YOU, your business, or your community. Because as a truck driver, you don’t want to drive into a microburst with 80 MPH winds, or a tornado, or a blizzard and get stuck for days in the middle of nowhere. And if you are dealing with a house fire, a truck spilling toxic chemicals, or a train spilling ammonia, you need to know where the wind is blowing to, and what it will be doing in the hours to come. And seeing storm chasers on the side of your roads is NOT the way you want to find out that a half-mile wide tornado is about to slam into your community. You’ve trained for these moments. Now, it’s time to learn how you can be ahead of the curve, even when the weather throws a curve at you. We know that a lot of the weather-related issues can be prevented with just some knowledge of what is about to happen in the days and hours to come. Drop us a line and let us know how we can help you get there…because we are all in this together.