View Full Version : Chicago's snow was not

11-22-2007, 10:41 AM
Hi John,

Where I live, northwest suburbs of Chicago, the local weather stations where showing that we would receive 2 to 3 inches of snow. This morning I noticed a trace. It appears to me that the storms would either go north or south once it gets to the city. Do the keepers of the models every calibrate the models? After this many years, how can they be this far off?

11-22-2007, 12:05 PM
I noticed that it took much longer than anticipated for the temps to drop last night.

Yesterday, we were only supposed to get to 38* as a high, but I reached 42* at my home in bloomingdale.

Thus, it rained until 8/9pm vs. the predicted 4/5pm change to snow.

11-22-2007, 12:56 PM
How does the radar know to change color. Does it measure the temp of the clouds? Last night in northern Illinois, I noticed it was green (rain) with pink and white making it so far west and then changing back to green.

11-22-2007, 07:35 PM

The models are under constant review and occasional upgrades. I suppose it is a matter of opinion on how far they miss by when they are wrong, but in this case they missed the critical temps for snow by just a few degrees. Had it been just a few degrees colder once the sun set, then the change over to snow would have been quicker and more snow would have fallen.

The models are attempting to do an extremely difficult job in forecasting the weather. They are getting better, but obviously still have a ways to go.


I am not completely sure of all of the parameters used to apply the color code for the radars some companies use. I can say the number one parameter is to read what the surface stations are reporting as far as precip type. Then they use things like temp and dewpoint to try and fill in the blanks.


11-22-2007, 10:47 PM
OK, but would they take the results of this storm and adjust the model to see if they can forecast the actual results. For example, when I create a XP-SWMM model for rainfall and storm sewer modeling, I am given data from rainfall gages in the area, like Wheaton, for a particular storm event. I then adjust perimeters in my model to reproduce the same results. Now my model is calibrated. Now I take that model and apply different rainfalls, durations, and rainfall curves to model existing conditions and proposed conditions.

I understand weather models are more complicated then hydrology and hydraulic models, but it doesn't seem like they are getting any better. Is there something else I'm not understanding?

(This is why I don't trust Global Warming models.)

11-23-2007, 11:38 AM

The models that try to emulate the atmosphere are quite a bit different than what it sounds like you are running. Your model sounds like it takes one variable rainfall (along with all the other fixed items) and arrives at a result. Pretty straight forward and since there is only one variable (how much rain in a given time), you can expect pretty accurate results once the model is dialed in.

The atmospheric models have litteraly 100's of 1000's of variables they have to deal with (perhaps even millions) with a synoptic scale weather feature such as we had on Wednesday in the Midwest. No two of these systems are ever the same either. Plus this is all being done in 3-D.

I think you would be amazed at how many computations are going on in a single model run. I can only venture to guess it is in the trillions.

The models are improving, but they can only improve as our understanding of the atmosphere and computing power improves. After all, the computers are just number crunchers, it is the humans that need to tell them what to do.


11-23-2007, 02:23 PM

Does there have to be a low pressure system in place in order for there to be lake effect snow?

Thank You,


11-24-2007, 06:21 AM

No. What you need is the air passing over the lake to be cold enough to support the instability to bring about the clouds and LES. There are actually a number of other factors, but low pressure is not one of them.