A Review on the Optimal Fingerprinting Approach in Climate Change Studies

TL;DR

This review analyzes the robustness of the optimal fingerprinting method in climate change detection, emphasizing conditions for its validity and its relation to statistical estimation techniques.

Contribution

It clarifies the conditions under which the optimal fingerprinting approach remains valid and connects it to Feasible Generalized Least Squares, addressing recent criticisms.

Findings

Optimal fingerprinting survives certain criticisms if assumptions hold.

Residual covariance estimation is crucial for the method's validity.

The residual consistency test effectively checks model-data agreement.

Abstract

We provide a review on the "optimal fingerprinting" approach as summarized in Allen and Tett (1999) from a point view of statistical inference in light of the recent criticism of McKitrick (2021). Our review finds that the "optimal fingerprinting" approach would survive much of McKitrick (2021)'s criticism under two conditions: (i) the null simulation of the climate model is independent of the physical observations and (ii) the null simulation provides consistent estimation of the residual covariance matrix of the physical observations, both depend on the conduction and the quality of the climate models. If the latter condition fails, the estimator would be still unbiased and consistent under routine conditions, but losing the "optimal" aspect of the approach. The residual consistency test suggested by Allen and Tett (1999) is valid for checking the agreement between the residual covariances of the null simulation and the physical observations. We further outline the connection between the "optimal fingerprinting" approach and the Feasible Generalized Least Square.