Emulator-based Bayesian calibration of a subglacial drainage model

Subglacial drainage model sensitively depend on numerous uncertain parameters, for example as shown by our previous sensitivity analysis. However, formal calibration has rarely been attempted for these models.

For this projct, we explore using borehole water pressure observations to constrain the value of GlaDS mogel parameters. To make the inference computationally tractable, we adapt the GlaDS-GP emulator to accelerate the MCMC sampling.

Preprint

Coming soon