What of the real world?

• Should we expect climate change in the real world to lie inside the range of predictions?

• It depends on the quantity of interest:

  • Has the ensemble converged, or does perturbing more parameters change the estimated range?
  • Is the ensemble consistent with observations in this quantity?
  • Do we expect this variable to be well-simulated?

• probabilistic forecast cannot be tested with a single event -- the world can always surprise us.