Paper 07: Robust uncertainty assessment of the spatio-temporal transferability of glacier mass and energy balance models¶
Citation
Zolles, T., Maussion, F., Galos, S. P., Gurgiser, W. and Nicholson, L.: Robust uncertainty assessment of the spatio-temporal transferability of glacier mass and energy balance models, Cryosph., 13(2), 469–489, doi:10.5194/tc-13-469-2019, 2019.
This study is the outcome of a master thesis that Tobias Zolles wrote under my supervision. The original idea came during a “paper club” were we discussed an interesting paper by Rye and co-authors [46]. In their study, the authors argued for the use of “multiobjective optimization” in glacier mass balance models to reduce model uncertainty. While we agreed with the main premise of the paper, we also noted that the main argument (that a model calibrated on more than one observable variable was more accurate) was based on a validation procedure realized on the same dataset as the one used for calibration. We therefore decided to apply multiobjective calibration in a new setting: with a full SEB/SMB model, and using cross-validation to assess the value of the calibration with data unseen by the model. In addition, we conducted an extensive sensitivity analysis to reduce the parameter space before calibration.
One of the main findings of the paper was that the uncertainty estimates for models with many free parameters are likely to be largely underestimated when not quantified with robust uncertainty assessments. Even in this European Alps context with a comparatively high data availability, the model parameters could not be constrained well enough to allow for a robust transferability of the model in space an time. We then discussed the consequences for the application of such a model at large-scales.
I contributed to the design of this study and supervised Tobias during his thesis and the writing of the paper. It is the first master thesis that I supervised that led to a publication: Tobias was a brilliant and ambitious student, now working on his PhD at the University of Bergen.
Link to the paper (open-access)