Comparison of Ice Nucleation Parameterizations for Dust Minerals in Climatological Simulations with a Global Model

The effect of aerosol particles on ice nucleation and, in turn, the formation of ice and mixed phase clouds is recognized as one of the largest sources of uncertainty in climate prediction. We utilize an improved sectional dust module in NASA GISS Earth System ModelE2.1, which distinguishes eight different mineral species and accretions between iron oxides and the other minerals. Simulations over a period of 20 years have been carried out with this model, and the mineral fields and other model variables (temperature, relative humidity) are used to calculate the ice nucleating particle (INP) number concentration, applying time-independent and time-dependent INP parameterizations, such as active site parameterization and water activity based immersion freezing model (ABIFM). We study how the dependence of the parameterizations on different model variables affects the mean INP number concentration. The sensitivity of the INP number concentration to fundamental dust properties such as emitted mineral size distributions and mixing state between minerals is also investigated. Results show that the sensitivity of the total INP number concentration to the emitted dust size distribution is rather small, but the sensitivity over the whole size range obscures offsetting differences in the magnitude and the sign of the sensitivity between smaller and larger particles. The sensitivity also displays geographical variations, depending on the mineral composition of the parent soils in the dust source regions. Results also show that the sensitivity to dust properties needs to be analyzed in the context of the calibration of the simulated dust cycle in the model, which is required to avoid a large observational mismatch with respect to other simulated variables such as aerosol optical depths.