Bootstrap-Based Robustness Analysis of Parameter Optimization in Climate Models Using QuadTune

Luis Hasenauer
Graduate Student
University of Wisconsin-Milwaukee

Tuning the parameters of climate models is essential for improving their performance, but this process is often complicated by structural limitations, overfitting, and trade-offs between different regions or variables. In my thesis, I combined the QuadTune optimization framework with nonparametric bootstrap resampling to analyze parameter uncertainty and identify tuning conflicts.

Advisor:
Vincent Larson

Committee Members:
David Spade, Daniel Gervini