23 January 2017

An Ice Sheet Model Validation Framework for the Greenland Ice Sheet

Greenland ice sheet model validation


We present a new, online software framework for use in validating model simulations of the Greenland ice sheet, which makes use of satellite-based surface elevation and mass change observations over the past two decades. We use high-resolution ice sheet model simulations to demonstrate the validation framework and the qualitative and quantitative metrics it provides. Simulations of varying degrees of complexity are used to show that these same metrics can be used to provide a measure of model skill with respect to observations of ice sheet change.

  • simulations demonstrate that quantitative metrics provided by the validation software are able to distinguish simulations of differing skill

  • modelers avoid need for expertise in processing of large, complex data sets

  • datasets can be altered or augmented remotely by relevant experts

  • validation software is extensible; new types of data, new metrics, and new geographic regions (e.g. AntarcRca) can be easily added


We propose a ice sheet model validation framework that takes advantage of observations of Greenland ice sheet change collected over the past several decades. We use realistic, high-resolution model simulations to demonstrate the framework and its use and propose and demonstrate qualitative and quantitative metrics for use in evaluating simulations against observations. The new framework demonstrates that our proposed metrics can distinguish relatively better from relatively worse simulations and that dynamic ice sheet models, when appropriately initialized and forced with the right boundary conditions, demonstrate predictive skill with respect to observed dynamic changes occurring on Greenland over the past few decades. An extensible design will allow for continued use of software as observational data become available for use in ice sheet model validation.

Stephen F. Price
Los Alamos National Laboratory