A cool low pressure center will move through the Pacific Northwest over the weekend, dropping our high temperatures into the 80s Saturday and then into the 70s on Sunday. The cold front, which will be dry and mostly clear of clouds, will make for a very windy Saturday, and keeps us in a Red Flag Warning until late Saturday night. After Sunday, the ridge of high pressure builds back in and we will warm back into the 80s on Monday and then the 90s for the rest of the week.

Here’s more on why weather prediction models have limitations. Molecules in the atmosphere don’t live in a vacuum… either metaphorically or physically. Even if you know the initial direction and speed of one at one point in time, it is bound to run into one of its neighbors soon, changing its speed and direction. Because they are always interacting with each other, trying to predict their behavior, either individually or corporately, becomes a real challenge.

Think of it in terms of a billiard table. With the balls scattered randomly over the table, toss a cue ball onto the table, and try to predict exactly where each ball will be, and what speed they will be going in… say… four seconds time. A challenging, but probably not impossible task if you know the exact speed and direction of the cue ball at time zero. If you didn’t know the exact speed and direction of the cue ball, but could only give an estimate, you might get a reasonably accurate picture four seconds later, but it could also be hugely different if just one ball just kisses another when you thought they’d miss.

But there you are only dealing with 16 balls, in two dimensions. Now blow that up to an unimaginable number (probably on the order of 10 to the 100^th power?) of air molecules, and put it in three dimensions, with all of them moving and without the ability to know exactly where they are all moving, and you can see your problems in getting an accurate picture days later.

In essence, that’s the problem we face in trying to rely on computers to get it right exactly every time… it’s impossible to know the exact initial state with the information from a couple of hundred weather balloons. So we have to simplify the atmosphere, “modeling” it to make approximations of the movement of larger chunks of it.

It works pretty well. Until the *^&%$#&# butterflies get involved.