Investigations of physical processes in polar firn through modeling and field measurements

Abstract

Evolution of firn on the polar ice sheets is important for several applications in glaciology, including interpreting climate records from ice cores and correcting ice-sheet-elevation change measurements from altimetry for changes in the amount of air in firn. I have developed the Community Firn Model (CFM), a modular model framework that can simulate numerous physical processes in firn, including densification, air transport, heat diffusion, and grain growth. In this dissertation, I use the CFM and measurements from Greenland to investigate firn densification, gas transport in firn, and uncertainty within firn models. First, I have coupled a firn-air model to a firn-densification model. I use this new model to investigate air transport in the lock-in zone and show that the lock-in zone could be created by impermeable layers, or it could be caused by differences between the advective and diffusive timescales of air transport. I also show that δ^15 N and δ^40 Ar data from the GISP2 ice core can be better explained than previously by including transient firn thickening during rapid climate-change events. Additionally, I explore the effects that impermeable layers in the firn have on gas records. I conclude that including transient firn evolution is essential to modeling gas transport correctly during climate changes, but limited knowledge of the microstructure of firn limits our ability to model firn-air transport accurately in the past. Next, I intercompare 11 firn-densification models and test their sensitivities to initial and boundary conditions at Summit, Greenland. I show that there is no ‘best’ firn model, and that there are numerous sources of uncertainty in firn models. The uncertainty in predicting the depth-integrated porosity of the firn is ~10%, and the uncertainty in predicting the depth and age of the firn-ice transition is greater than 10%. Uncertainties in surface boundary conditions, including the climate forcing and surface density, as well as model parameterizations contribute to model uncertainties. Finally, I present firn-compaction data from Greenland and compare firn-densification-model predictions to those data. I use the firn-compaction data to derive a new firn-densification equation for Summit. I show that performance of firn densification models varies by site and by depth in the firn. At Summit, several models can predict the compaction rate to within 10%, but at EastGRIP, all models have an RMS error greater than 20%. My results indicate that the models do not yet represent physical processes correctly; I identify future research that is needed to help improve the models.