Constraining uncertainty in modelling of the North American ice sheet

An ice sheet covered much of North America during the last ice age. Computer models have failed to capture the extent of this ice sheet due to uncertainties in the modelling and demanding computational calculations. In this study, a fast-running computer model was used to undertake hundreds of ice sheet simulations on the ARC4 HPC system. The results show that ice sheet reflectivity impacts most on the predictions of ice sheet extent. The study found that the model parameters that produce the best simulation of the present-day Greenland ice sheet produce a poor simulation of the past North American ice sheet.

(a) North American and Greenland Ice volume evolution for each ensemble member. (b) The final ice volume and extent error (compared to geological constraints) for each ensemble member. (c) The % of ensemble simulations with ice cover compared to constrained margin shown in red. (d) The % of Not Ruled Out simulations with ice cover compared to the constrained margin shown in red.

Being able to run simulations at speed enabled researchers to experiment with many different parameters in order to improve the accuracy of the model in different scenarios.

- Dr Niall Gandy, Researcher

Reference

Gandy, N., Astfalck, L. C., Gregoire, L. J., Ivanovic, R. F., Patterson, V. L., Sherriff-Tadano, S., et al. (2023). De-tuning albedo parameters in a coupled Climate Ice Sheet model to simulate the North American Ice Sheet at the Last Glacial Maximum. Journal of Geophysical Research: Earth Surface, 128, e2023JF007250. https://doi.org/10.1029/2023JF007250