1Institute for Atmospheric and Climate Science, ETH Zurich, 8092 Zurich, Switzerland
2NASA Langley Research Center, Hampton, Virginia 23681, USA
3Science Systems and Applications, Incorporated, Hampton, Virginia 23666, USA
4Laboratory of Atmospheric Chemistry, Paul Scherrer Institut, 5232 Villigen, Switzerland
5Institut für Energie- und Klimaforschung – Stratosphäre (IEK-7), Forschungszentrum Jülich, 52425 Jülich, Germany
6Institut für Physik der Atmosphäre, DLR Oberpfaffenhofen, 82230 Oberpfaffenhofen, Germany
Abstract. This paper provides unprecedented evidence for the importance of heterogeneous nucleation, likely on solid particles of meteoritic origin, and of small-scale temperature fluctuations, for the formation of ice particles in the Arctic stratosphere. During January 2010, ice PSCs (Polar Stratospheric Clouds) were shown by CALIPSO (Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations) to have occurred on a synoptic scale (~ 1000 km dimension). CALIPSO observations also showed widespread PSCs containing nitric acid trihydrate (NAT) particles in December 2009, prior to the occurrence of synoptic-scale regions of ice PSCs during mid-January 2010. We demonstrate by means of detailed microphysical modeling along air parcel trajectories that the formation of these PSCs is not readily reconciled with expectations from the conventional understanding of PSC nucleation mechanisms. The measurements are at odds with the previous laboratory-based understanding of PSC formation, which deemed direct heterogeneous nucleation of NAT and ice on preexisting solid particles unlikely. While a companion paper (Part 1) addresses the heterogeneous nucleation of NAT during December 2009, before the existence of ice PSCs, this paper shows that also the large-scale occurrence of stratospheric ice in January 2010 cannot be explained merely by homogeneous ice nucleation but requires the heterogeneous nucleation of ice, e.g. on meteoritic dust or preexisting NAT particles. The required efficiency of the ice nuclei is surprisingly high, namely comparable to that of known tropospheric ice nuclei such as mineral dust particles. To gain model agreement with the ice number densities inferred from observations, the presence of small-scale temperature fluctuations, with wavelengths unresolved by the numerical weather prediction models, is required. With the derived rate parameterization for heterogeneous ice nucleation we are able to explain and reproduce CALIPSO observations throughout the entire Arctic winter 2009/2010.