The effect of Saharan dust and the seeder-feeder mechanism on the relationship between temperature and ice formation in clouds observed with lidar

We present a lidar-based study of the relationship between ice formation and atmospheric temperature and the potential role of Saharan dust particles as ice nuclei. The underlying lidar dataset is based on 1300 measurements carried out between 1997 and 2008 at Leibniz Institute for Tropospheric Research (IfT), Leipzig, Germany. More than 2300 distinct cloud layers were identified from the measurements. The lidar depolarization ratio was used to separate the observed cloud layers into either water clouds or ice-containing clouds. Cloud top temperatures were derived from model reanalyses or radiosonde data. The statistical analysis revealed that the fraction of ice--containing clouds increases strongly with decreasing temperature. Between -10 and -15 °C already 40% of all observed clouds contained detectable amounts of ice particles. At temperatures below -25 °C almost 100% of the clouds contained ice. This finding is in agreement with existing mid-latitudinal studies based on airborne in-situ measurements. The investigation of the role of the seeder-feeder mechanism indicated that this process was of minor relevance for the presented study. In the next step, the cloud data set was separated into dust-laden and dust-free fractions by using vertical profiles of mineral dust concentration provided by the Dust Regional Atmospheric Modeling System (DREAM). This approach was possible because of the unique location of Leipzig in central Europe with its frequent changes in the large-scale circulation pattern varying from aerosol-background conditions in air masses coming from the North Atlantic to dust-laden conditions in westerly and southwesterly flows. Air masses that contained mineral dust were found to show 10%-30% more ice-containing clouds between -5 and -25 °C compared to dust-free air masses. Consequently, Saharan dust appears to have a detectable effect on heterogeneous ice formation over central Europe. In a last step we compared the relationship between ice-containing clouds and temperature of the Leipzig dataset with a similar data set from Cape Verde (15 °N, 23.5 °W). Even though dust is omnipresent in the troposphere at this location, the study that is based on depolarization lidar observations of more than 200 spatially well-defined altocumulus clouds did not show a significant number of ice clouds at temperatures above -20 °C after correction for the seeder-feeder mechanism. Possible explanations for the observed differences between the two data sets are contrasts in atmospheric dynamics over Central Europe and western Africa or the presence of different aerosol types over Europe that may provide additional effective ice nuclei besides Saharan dust.