Ensembles are produced by running the same weather models many different times with slightly varying initial conditions. Because we can’t observe every tiny bit of air in our atmosphere, our picture of the weather currently is incomplete. Any small error in the weather model initially due to this gap in observation is compounded exponentially out through time due to chaos. Ensembles attempt to fix this problem by starting the model with a bunch of ideas of what the atmosphere could be doing right now. Each one of these ideas will create its own outcome, known as an ensemble member. Ensembles are a great tool for gauging uncertainty in a forecast. If all, or almost all, the ensemble members agree on a particular outcome, you can have high confidence that that outcome will occur. If the ensembles disagree, it’s wise not to put too much confidence in one outcome or another. Different ensemble systems have different numbers of ensemble members and the more ensemble members there are, the better the forecast will be as it will take into account a wider range of possibilities. Because there are more potential forecast outcomes as you head farther out into the future, ensembles become especially useful after Day 4 or 5.
Hovering the time series with the cursor will show the minimum, maximum and mean vaules of the ensemble and the result of the main run.
