An iterative process for efficient optimisation of parameters in geoscientific models: a demonstration using the Parallel Ice Sheet Model (PISM) version 0.7.3

Abstract

Abstract. Physical processes within geoscientific models are sometimes described by simplified schemes known as parameterisations. The values of the parameters within these schemes can be poorly constrained by theory or observation. Uncertainty in the parameter values translates into uncertainty in the outputs of the models. Proper quantification of the uncertainty in model predictions therefore requires a systematic approach for sampling parameter space. In this study, we develop a simple and efficient approach to identify regions of multi-dimensional parameter space that are consistent with observations. Using the Parallel Ice Sheet Model to simulate the present-day state of the Antarctic Ice Sheet, we find that co-dependencies between parameters preclude any simple identification of a single optimal set of parameter values. Approaches such as large ensemble modelling are therefore required in order to generate model predictions that incorporate proper quantification of the uncertainty arising from the parameterisation of physical processes.

Systematic Calibration of a Convection‐Resolving Model: Application Over Tropical Atlantic

“…The other, instead of trying to find the optimum parameter setting, involves using Bayesian approaches to provide the uncertainty for the parameters (Bony & Dufresne, 2005; Couvreux et al., 2021; Rougier, 2007; Salter et al., 2019; Sanderson, 2011; Sexton et al., 2012). Except for some studies that use particle‐based approaches (Lee et al., 2020) or adaptive sampling algorithms (Phipps et al., 2021), most of the research uses emulators, mapping model inputs with outputs to reduce computational resources. In terms of the emulators, the calibration methods can also be divided into those that use statistical models (Voudouri et al., 2021) and machine learning methods (Li et al., 2019).…”

Understanding surface melt in Antarctica and implications for future ice sheet evolution

“…In this thesis, all of these PDD models with optimal PDD parameters, determined by the minimal RMSE method, contain only one combination of the T 0 and DDF that best represents the melt days and melt intensity with respect to the training data. However, geoscientific models that use parameterizations to approximate physical processes can rarely be tuned perfectly by a sole configuration of parameters (Phipps et al, 2021). The PDD scheme, as an approach to approximate the calculation of surface melt, most probably does not have a unique configuration of parameters.…”