Ensemble agrees: It'll be cloudy this week
Posted September 28, 2015
Many of you have heard us talk about ensembles from time to time. The ensemble idea is based on the undeniable truth that no single model will ever be exactly right. The only question is how wrong will it be.
Why is this? There are two answers.
First, the atmosphere is way too complex to be able to perfectly simulate every process involved. Second, we can't observe the atmosphere perfectly to start with.
Small errors in the beginning can grow into large errors down the road. Imagine aiming a squirt gun at a spot on the nearest wall. If your aim is off by the smallest of amounts when you squirt, that amount will be magnified when the water hits the wall. And the farther away the wall is, the more you will miss the target.
The ensemble approach allows us to see how much of a potential effect these initial errors can have over two weeks of weather forecasting. Sometimes the atmosphere is relatively stable, and these small initial errors don't have much of an effect, but sometimes they can lead to huge variances in the solutions.
This chart shows an ensemble forecast from Sunday night for cloud cover through the end of the week. This chart is called a "Box and Whisker plot." In this case there are 42 ensemble members, each with slightly different initial conditions. The charts split the data up into four sections, and these sections are called quartiles. The top and bottom quartiles represent the 25 percent of the solutions that have the lowest values and the 25 percent of the solutions with the highest values. The middle two quartiles represent 50 percent of the solutions with the values closest to the median.
By the way, the median is the point where half of all the solutions are higher and half of all the solutions are lower.
If the yellow box is small, there is general agreement among the ensemble members and forecast confidence is high, but if the yellow box is large, don't call your bookie in Vegas, because all bets are off!
There is remarkable agreement among the ensemble members pointing to every day this week being overcast, and the first hope for sunshine doesn't arrive until Saturday. It will be interesting to see how accurate the ensemble members are regarding their certainty of the low degree of uncertainty!