Ensemble Climate Prediction: Key Questions and Controversies

Ohio University, Athens OH

Investigators

Abstract

Introduction: The PI proposes to spend a year of full-time research examining methodological and philosophical questions having to do with a particular type of climate modeling, one that is at the forefront of prediction efforts. She will investigate the issue of model inclusion in ensembles, which ties in with general issues of model evaluation. The PI already has a manuscript on this topic, and wants to develop a more detailed analysis. At issue are general questions of how to interpret computational models in science, their use in predictions, and their evaluation. She will investigate the presentation of ensemble predictions as probability distributions, and in doing so will address the following questions. How do you judge model quality (meaning which models should be included in the ensemble)? How should the results of ensemble modeling be used? How should the models and the inferred probabilities be interpreted? What alternatives are there? Intellectual Merit: The disagreement among climate scientists about the practices of ensemble prediction provides a prime opportunity for philosophical work. The proposed work will strengthen philosophy of science by bringing climate science into the realm of sciences attended to by the field, complementing work on biology, physics, and psychology. The idea behind this project is outstanding noting that the methodological challenges in the interpretation of the results of predictive modeling of this sort could not be more stark, nor the philosophical dimension of the issues more compelling. Broader Impact: The philosophical issues raised in the proposal are of wide interest, spanning all of the modeling sciences, including cognitive science, neuroscience, geography, etc. The researcher should be able to maximize this potential impact by examining the similarities and dissimilarities with ensemble modeling in biology and/or physics and to describe the results in a way that is accessible to a broad audience.