Dog days of summer and storms in Hotlanta 

Posted: 5:34 pm Wednesday, July 12th, 2017

By Kirk Mellish

There has been a noticeable lack of heat this spring and summer, and a concurrent and naturally related conspicuous presence of greater than normal wetness.

This has not been “the norm” in recent decades though with extreme heat and drought common for months on end. So this spring and early summer has been a break. My A/C and water bills are appreciative.

I’ve explained in many previous blogs what it takes to get high heat with little rain and about “positive feedback loops” that perpetuate one type of weather pattern or the other and how and why they tend to persist for long stretches.

The models have repeatedly since May, tried to bring the hot and rain-free upper level ridge to the eastern U.S. for long periods, but then they back off and it never materializes except for a day or two. They are doing it again. Eventually they may be right as sooner or later the pendulum usually swings, but it doesn’t have to do so!

*More on that in the next blog.*

Thunderstorm forecasts unlike other types, must be refreshed every two-three hours.

Because what drives them most of the time are very small scale and subtle features in the atmosphere (microscale and mesoscale), many of which can not be discerned long in advance and many of which “fall between the cracks” of the model grids or net. They also move and change constantly like an ameba in unexpected ways.

Therefore, in advance we can only give a best estimate of the most likely areas and time-frame for storms to pop up… but then it has to be revised frequently as actual measured and observed data reveal variations from expectations and forecast equations (numerical weather prediction output NWP).

Take a look below at how much rain from storms can vary over vary short distances on any given day:



(note the difference at any given moment is even greater)

You can see why sometimes it really IS 50-50. It’s not a guess, it’s a fact of life.

Many of us in Atlanta have had the experience of driving in AND out of storms more than once on the same trip. Many of us have experienced a heavy thunderstorm at the store blocks from home while home was bone dry or vice versa! Or actually seen it rain on one side of the street or house and not the other.

This is why it is NOT scientifically possible to give a yes or no answer to will it rain where I am today? Or do you think we’ll get a thunderstorm this afternoon?

So instead we give you our best estimate at the odds or chance at any given spot in the metro during the forecast period. X percent chance (related to isolated, scattered, likely, or numerous). That way, depending on your sensitivity or the cost factor of your outdoor plans you can adjust accordingly.

Sometimes of course our initial estimate (like estimates in any arena of life) will be too high and other times too low. HENCE, that need to refresh a thunderstorm forecast every few hours.

In mid-summer the typical chance of a thunderstorm where you are is 30-40%. (diurnal air mass convection, you can google) So the lower or higher from that the odds that day are then the safer or riskier outdoor activities may be.

I personally do not even consider canceling outdoor plans until the threat is at least 70%, but that’s me, and even then it depends.

Here’s a recent example of a HiRes thunderstorm forecast from a model 12 hours in advance for 5pm Wednesday July 13, 2017, vs. what the real world radar ended up looking like at 5pm. Big difference, not good. BUT the day before, its projection did very well!!

“Predicted radar” MODEL:

Screen Shot 2017-07-12 at 5.16.44 PM



Screen Shot 2017-07-12 at 5.03.11 PM

Pretty and fancy colorful “forecasts in motion” you may see somewhere can be entertaining, but don’t put too much stock in it. Entertainment is entertainment.

For a more accurate forecast a meteorologist must look at all available information and not just a cool looking colorful prediction map.

It may be the reason SOME forecasters predicted an above-normal storm chance today Wednesday July 12th with likely coverage, yet I did not. It’s tricky stuff, sometimes you’re right sometimes you get fooled.

The big problem is, on the surface the weather charts for the “active” day didn’t look any different from the “not active” day!  Welcome to our world.

That goes to the point I made earlier about thunderstorm forecasts versus other forecasts.

This is why a meteorologists training and experience matter greatly.

I always try to remind people not to be model huggers, it’s NOT model-ology, it’s meteorology. Choose your forecast source wisely. APPs on your phone or desk top are clueless, and so are some forecasters.

FOLLOW ME on Twitter @MellishMeterWSB