New methods for evaluating sea ice in climate models based on energy budgets

• Main Author: West, A
• Other Authors: Collins, Mat, Blockley, Ed, Beare, Robert
• Format: Doctoral or Postdoctoral Thesis
• Language: unknown
• Published: University of Exeter 2021
• Subjects: Sea; Ice; Arctic; climate; model; evaluation; energy; surface; buoy; albedo; Sea ice
• Online Access: http://hdl.handle.net/10871/127762

Arctic sea ice plays a vital role in the Earth’s climate system, through its reflection of solar energy and insulation of ocean heat, and has changed rapidly in the past 20 years. Model simulations of Arctic sea ice display a wide spread both in the present-day and in the future. Due to lack of observations however, evaluation of sea ice simulation has historically been limited in scope mainly to ice extent and (sometimes) volume, with little attempt to evaluate at large scale simulation of the fundamental thermodynamic processes governing sea ice growth and melt. In this thesis two new, contrasting methods are presented for evaluating Arctic climate simulation that address this: firstly, the induced surface flux (ISF) framework attributes model biases (differences) to specific proximate drivers using existing reference datasets. Secondly, the Arctic ice mass balance buoy (IMB) network is used to build a dataset with which to evaluate many sea ice thermodynamic processes directly. We use three UK CMIP models for analysis: HadGEM2-ES, HadGEM3-GC3.1 and UKESM1.0. These models display very different Arctic sea ice simulations, with ice in HadGEM2-ES thinnest and ice in UKESM1.0 thickest. Using the ISF framework and IMB evaluation, it is shown that modelled sea ice volume is tightly coupled to modelled ice growth and melt, and that most of the model biases and differences are caused by differences in albedo and atmospheric forcing arising in the late spring and early summer. Despite this, a downwelling longwave radiation bias present in all models during winter ‘predisposes’ them towards a thicker ice cover. The methods can also be used to evaluate the proximate impact of specific model improvements in the latter two models on sea ice growth and melt, which is seen to be small but non-negligible. The results also show that more accurate observations of Arctic radiative fluxes, and of snow area and thickness, would be particularly useful in improving model evaluation, and that ice mass balance buoy measurements would ...