Modeling of soot concentration in firn in the Greenland ice sheet

Abstract

The albedo of the firn layer on the Greenland ice sheet (GrIS) is affected by the impurity soot concentration in firn. This thesis describes the evaluation of a stand-alone firn layer model with a dynamically modelled soot distribution. The model is evaluated for different locations on the GrIS. Analysis of model cross sections of the firn makes clear that it remains challenging to make cross sections which are in line with observational cross sections. Analyzing the mean surface soot concentration during the melt season shows that the effect of the dynamic soot distribution is almost independent of the initial soot concentration but dependent of the location for which the simulation is performed. Comparing model data with observational data makes clear that it is possible to reduce the model bias in the upward shortwave radiation and albedo by choosing the right values for the model parameters. This reduction is optimal for Automatic Weather Station (AWS) S9, but still suboptimal for AWS sites S6 and S10. The biases can be reduced using both a static and dynamic soot distribution. Overall, a static soot concentration between 0.000 and 0.002 ppb seems to give the best results. Analysis of the sensitivity of the model shows that the albedo is more sensitive to changing soot concentrations than to changing soot removal rates. The obtained optimal parameter values have a different SMB at S9, but the suboptimal parameter values for S6 and S10 give more similar effects for the separate locations. The SMB is more sensitive to changing soot concentrations compared to changing soot removal rates. Along the K-transect the order size of the SMB sensitivity is 0.5 m w.e.yr^-1.