Sensitivity of the Greenland surface mass and energy balance to uncertainties in key model parameters

Tobias Zolles,Andreas Born

doi:10.5194/tc-15-2917-2021

Tobias Zolles, Andreas Born

Open Access

https://doi.org/10.5194/tc-15-2917-2021

Copy DOI

Abstract

Abstract. We investigate the sensitivity of a distributed glacier surface mass and energy balance model using a variance-based analysis, for two distinct periods of the last glacial cycle: the present day (PD) and the Last Glacial Maximum (LGM). The results can be summarized in three major findings: the sensitivity towards individual model parameters and parameterizations is as variable in space as it is in time. The model is most sensitive to uncertainty related to atmospheric emissivity and the down-welling longwave radiation. While the turbulent latent heat flux has a sizable contribution to the surface mass balance uncertainty in central Greenland today, it dominates over the entire ice sheet during the cold climate of the LGM, in spite of its low impact on the overall surface mass balance of the Greenland ice sheet in the modern climate. We conclude that quantifying the model sensitivity is very helpful for tuning free model parameters because it clarifies the relative importance of individual parameters and highlights interactions between them that need to be considered.

Highlights

The main focus of the results is on the surface mass balance, and discussion is limited to the total sensitivity index (ST i), due to limited information that can be extracted from SXi in complex models
The surface mass and energy balance model BErgen Snow SImulator (BESSI) has been improved by accounting for turbulent latent heat flux and snow aging
The sensitivity of the model to the new implementations and uncertain model parameters was assessed with a variance-based sensitivity method based on two ensembles with a total of 16 500 simulations

Summary

Introduction

Models to calculate SMB cover a whole range of complexities from empirical index models that only account for air temperature (Ohmura, 2001; Zemp et al, 2019) or temperature and solar radiation (Bintanja et al, 2002; Van Den Berg et al, 2008; Robinson et al, 2011) to coupled atmosphere– snow models that simulate the snowpack in multiple layers and give a full representation of the atmospheric circulation, based on physical first principles (Lehning et al, 2002; Fettweis, 2007; Noël et al, 2018). This situation motivated the development of models that balance the defensible representation of the relevant physical processes with computational efficiency (Krapp et al, 2017; Krebs-Kanzow et al, 2018; Born et al, 2019)

Results

Discussion

Conclusion