Robustness of pattern scaled climate change scenarios for adaptation decision support
Pattern scaling offers the promise of exploring spatial details of the climate system response to anthropogenic climate forcings without their full simulation by state-of-the-art Global Climate Models. The circumstances in which pattern scaling methods are capable of delivering on this promise are explored by quantifying its performance in an idealized setting. Given a large ensemble that is assumed to sample the full range of variability and provide quantitative decision-relevant information, the soundness of applying the pattern scaling methodology to generate decision relevant climate scenarios is explored. Pattern scaling is not expected to reproduce its target exactly, of course, and its generic limitations have been well documented since it was first proposed. In this work, using as a particular example the quantification of the risk of heat waves in Southern Europe, it is shown that the magnitude of the error in the pattern scaled estimates can be significant enough to disqualify the use of this approach in quantitative decision-support. This suggests that future application of pattern scaling in climate science should provide decision makers not just a restatement of the assumptions made, but also evidence that the methodology is adequate for purpose in practice for the case under consideration.