FLAKE NEWS: Beware of weather app snowfall amounts
WASHINGTON - When you look at it, smartphones are really one of the most incredible inventions of the modern era. Thanks to modern technology, a nearly infinite amount of information is now available right at your fingertips.
This includes weather forecast information, with a number of applications available promising to give you the most up to date and accurate forecast. However, it is the very nature of attempting to bring users the latest data that can sometimes be misleading to actual expectations.
On Wednesday, I received the above tweet from a concerned viewer about her weather app on her cellphone. The model was showing 8 to 12 inches of snow here in Washington on Saturday afternoon. Obviously, such a tweet is alarming. That is a lot of snow after all! But is it correct?
The answer is almost certainly not. The problem is how these weather applications push you data. These apps promise you an up-to-date forecast for your exact location.
According to the 2015 National Census, there are a total of 19,505 cities and towns across the United States. A human forecaster cannot possibly hit each of those locations every day and write a forecast for each. Many of the applications instead rely on raw computer-generated weather data to complete this extraordinary task. In a general sense, this gets the job done just fine. However, it is when these applications try to get into the details of the forecast that things can become a little misleading.
First, you need to understand a little bit about these weather models themselves. They are not all-knowing crystal balls that tell us exactly what to say or do. They are tools, and like any toolbox, there are many of them. The above is just one example of a number of weather models that forecasters look at every single day before we make the forecast here at FOX 5.
There is the American (GFS) model, the European model, the North American Model (NAM), the Canadian Model, the High Resolution Rapid Refresh Model (HRRR), the German Model (ICON), the Rapid Precision Model (RPM) … the list can go on. The point being -- there are a number of different weather models out there. Each has its own strengths and weaknesses, and it is the job of the human forecaster to determine which weather model is performing best in a given situation.
The above image is raw data from Wednesday morning's run of the European weather model, typically considered the most accurate by weather forecasters. As you can see from the scale, the shade of purple indicated where the model is showing potentially more than 12 inches of snow across the region.
So why the skepticism on that number? Well, I will tell you a little secret about weather models -- they do not do snow. Meaning the weather model itself does not come to us and say, "Hey, this precipitation is snow, not rain." Instead, weather modeling sites use various methods involving temperatures across multiple layers of the atmosphere to calculate how much snow will fall based on the raw data that it is given. The result is images like the one you see above.
Your weather app will take these raw numbers and throw them at you, often updating with each and every run of the particular weather model or blend of models that it chooses to utilize. This is why they change so often! In an ideal winter setting, these maps are fine! However, snow in April is not an ideal setting, and this is where some of the shortcuts and calculations used to get snowfall forecasts can come back to haunt you if numbers are put out before a human forecaster is able to look at them and correct them for expected errors.
Problem No. 1 has to do with the snow itself. Many weather modeling websites calculate surface snowfall at a very common back-of-the-book rate that meteorologist use known as a 10:1 ratio, meaning an inch of liquid rain that the model sees as snow will accumulate 10 inches of snow on the ground. So a half-inch of liquid rain would equal 5 inches of snow. A tenth an inch of rain would be an inch of snow, etc. This works well when snow is falling under ideal conditions where the atmosphere is below freezing at all layers from the surface all the way up to the cloud layer.
The problem with snow in the spring is that it often is not the case, meaning the snow will fall more wet as opposed to the dry, flaky snow that we can get in the middle of winter. This means that the liquid-to-snow ratio will actually be lower than 10:1, sometimes 7:1 or 6:1 or lower.
Another thing many weather models do not see well is sleet. Many snow calculations see "sleet" as "snow" and accumulate it as such even though it does not under any circumstances accumulate like snow. This was seen during our March snow event, where models at one time were showing over a foot and a half of snow in downtown D.C. due to the models confusing sleet with snow.
Finally, the weather maps like the ones above assume that every flake that falls will stick. They do not account for things like ground temperatures and snow melting on contact, which is something that is very common with snowfall in spring. There are some snow calculation methods that do take this into account, which can be more accurate, but the vast majority of maps I see tweeted around social media do not take these factors into account.
The upcoming storm is a prime example of why you need to stick with and listen to the FOX 5 Weather Team in the days ahead instead of relying solely on a weather application. It is our job as meteorologists to be able to sort through the mess of weather models and bring you a more accurate forecast with a human touch. We do things like look at our weather history, knowing that Washington D.C. has never in the city's recorded history had more than 5.5 inches of snow in the month of April, and has not seen more than an inch of snow in any single event since 1924.
We take into account the fact that as cool as March has been, downtown D.C. has not fallen below freezing once since March 21, meaning that the ground temperatures around the region have risen above freezing over the past couple of weeks. We look at the fact that we have added two hours of additional daylight in the past two months. We look at the time of day that weather models are showing the heaviest snows here in the District, most of which are currently during the middle afternoon when it is more difficult to accumulate snow than the overnight hours.
Most importantly, we do not just look at the model itself, but also interpret the model to try and determine how it could change between now and Saturday.
Compare the above forecast issued by the weather forecasters at the National Weather Service to the raw model data from the European model seen in the earlier image. Quite the difference, right? These are the differences you will see when the forecast is given a human touch because we do more than just throw the raw data at you as fact. Weather models are tools, but they are only useful tools if you know how to interpret and utilize them.
The beauty of it is that you do not have to. Instead, you can just sit back, relax and let the FOX 5 Weather Team do the work for you. We are here working hard to bring you the most accurate forecast we possibly can without the hype of the big numbers the models are putting out. The storm this weekend is a very tricky one, but count on the Fox 5 Weather Team to get you through this one. After all, we have the human touch!