No model got this right today. And I look at all of them. The bow-echo that formed had been predicted to occur up near Allentown.
Radar 12:06 PM—
So how bad were the models?
Here’s the 10 AM HRRR model forecast for 12:00 PM, (two hours before the storms hit) —
Here’s the 11 AM HRRR model forecast for 12:00PM—
So even one hour before the storms hit, the high resolution model didn’t forecast this bow echo line of storms. Wow!
Over many years of being a weather hobbyist and prior to having complete weather data available on the Internet, it was never clear to me what happens when a predicted cold front and its associated thunderstorms “falls apart” before reaching us.
So what happened to today’s cold front and the expected thunderstorms?
Basically, the front lost its “upper air support”. But what does that mean?
Usually when a strong cold front approaches, the upper air “heights” typically represented by the 500 mb lines (blue lines on the map below) move southward, meaning a reduction in height and a resulting vertical upward motion in the atmosphere. This supports rain and thunderstorms.
(500 mb heights are lines of equal height above ground where the pressure is 500 mb. This is about 18,000 feet or about the “middle of the atmosphere”. More about this here.)
That was originally expected to occur.
However, the most recent models today show these 500 mb height lines not moving south but instead, halting and then moving northward after midnight.
This causes sinking motion and downward vertical movement. Precipitation and thunderstorms can’t form.
That’s what’s happening tonight.
Why did the models get this wrong on previous days? Nobody knows but there is a strong tropical system off the southeastern coast and it’s been my observation over many years that when tropical systems are present, the models just don’t do that well.
With the change to Eastern Daylight Saving Time, the TV weather forecasts you will see on the 10 and 11 PM news won’t/can’t be based on the latest models. Why? Read my post from March 2018.
I tried to further explain this snag in another post last year.
Over the past year and a half, I’ve been directly downloading weather model “grib” (gridded binaries) data from NOAA for these forecasts instead of relying on the adequate, but limited data freely available from university sites on the Internet.
Over the past day I’ve been redoing my scripts/programs to have them run properly starting at 2 AM Saturday night.
To give you an idea of how late some of these weather models become available in Eastern Daylight Saving Time, here’s a few of the changes I need to make:
NAM model 9:38 PM EST —> 10:38 PM EDT
GFS model 10:38 PM EST —> 11:38 PM EDT (first 24 hour forecast data)
HIRESW Models 10:10 PM EST —> 11:10 EDT
HIREF ENSEMBLE: 11:19 EST. —> 12: 19 AM EDT
CMC GDPS 11: 45 PM EST —> 12: 45 AM EDT
ICON Model 11:21 PM EST —> 12: 21 AM EDT
To those times, add 5-10 minutes to download the data, time to review the data and you’re well into the wee hours of the morning.
Even the hourly HRRR, RAP and NBM models require several hours to incorporate the changes in the upper air measurements (also done an hour later.)
So when you hear “check back at 11” for the weather on the TV during Eastern Daylight Saving time, you’re not going to get an updated forecast on new data. It’s true for the next day forecast and even more so for the five day forecasts.