Atmos. Chem. Phys., 3, 823–838, 2003 www.atmos-chem-phys.org/acp/3/823/ Atmospheric Chemistry and Physics The roles of dynamical variability and aerosols in cirrus cloud formation B. Karcher¨ 1 and J. Strom¨ 2 1Deutsches Zentrum fur¨ Luft- und Raumfahrt (DLR), Institut fur¨ Physik der Atmosphare,¨ Oberpfaffenhofen, Germany 2Stockholm University, Institute of Applied Environmental Research, Stockholm, Sweden Received: 5 February 2003 – Published in Atmos. Chem. Phys. Discuss.: 14 March 2003 Revised: 17 June 2003 – Accepted: 17 June 2003 – Published: 23 June 2003 Abstract. The probability of occurrence of ice crystal num- role of vertical velocities and mesoscale variability in ver- ber densities in young cirrus clouds is examined based on tical velocities in controlling cirrus properties. The results airborne measurements. The observations have been carried suggest that, in any effort to ascribe cause to trends of cirrus out at midlatitudes in both hemispheres at equivalent lati- cloud properties, a careful evaluation of dynamical changes tudes (52 − 55◦ N/S) during the same season (local autumn in cloud formation should be done before conclusions regard- in 2000). The in situ measurements considered in the present ing the role of other anthropogenic factors, such as changes study include temperatures, vertical velocities, and total ice in aerosol composition, are made. crystal concentrations, the latter determined with high preci- sion and accuracy using a counterflow virtual impactor. Most young cirrus clouds typically contain high number densities 1 Introduction (1−10 cm−3) of small (diameter < 20 µm) ice crystals. This mode dominates the probability distributions and is shown It has long been recognized that dynamical processes influ- to be caused by rapid cooling rates associated with updraft ence the characteristic, heterogeneous macrostructure of cir- −1 speeds in the range 10 − 100 cm s . A second mode con- rus clouds. Relevant dynamical processes in the upper tro- −3 taining larger crystals extends from ∼ 1 cm to low concen- posphere and tropopause region include mesoscale gravity −4 −3 trations close to the detection threshold (∼ 3 × 10 cm ) waves (e.g. arising from convection or from synoptic-scale and could be associated with lower updraft speeds. Results of weather systems operating over a wide range of scales be- a statistical analysis provide compelling evidence that the dy- tween a few and many tens of kilometers); lee waves gen- namical variability of vertical air motions on the mesoscale erated downwind of mountain ridges; background air turbu- is the key factor determining the observed probability distri- lence as a result of shear generation (e.g. near the jet stream); butions of pristine ice crystal concentrations in cirrus. Other and intrinsic turbulence generated by latent heat and radia- factors considered are changes of temperature as well as size, tive processes within cloud. Turbulence and wave activity number, and ice nucleation thresholds of the freezing aerosol often appear as coupled phenomena and cannot easily be particles. The variability in vertical velocities is caused by separated. Model studies highlighted the complex suite of atmospheric gravity waves leading to small-scale tempera- interactions between radiative, microphysical, and dynami- ture fluctuations. Inasmuch as gravity waves are widespread, cal processes in cirrus clouds (Starr and Cox, 1985; Zhang mesoscale variability in vertical velocities can be viewed as et al., 1992). Knowledge of dynamical factors influencing a universal feature of young cirrus clouds. Large-scale mod- both, cirrus cloud macrophysical properties and microphysi- els that do not account for this subgrid-scale variability yield cal structure is required to assess the radiative effects of cir- erroneous predictions of the variability of basic cirrus cloud rus and hence their role in climate (Stephens et al., 1990; properties. Climate change may bring about changes in the Quante and Starr, 2002). global distribution of updraft speeds, mean air temperatures, Comparatively little emphasis has been put on the role dy- and aerosol properties. As shown in this work, these changes namical processes play during the formation of cirrus clouds, could significantly modify the probability distribution of cir- although the processes affecting cloud formation may in- rus ice crystal concentrations. This study emphasizes the key fluence the entire life cycle of the cloud. An early field study revealed that ice crystal properties in cirrus clouds are Correspondence to: B. Karcher¨ ([email protected]) strong functions of the vertical velocity w and temperature c European Geosciences Union 2003 824 B. Karcher¨ and J. Strom:¨ Dynamical variability, aerosols, and the formation of cirrus clouds T (Heymsfield, 1977). The potential impact of atmospheric 2 INCA Experiment waves on the physical properties of young cirrus clouds, in particular on the number concentration ni of small (diame- During the INCA experiment, the DLR (Deutsches Zentrum ter below 20 µm) ice crystals, was demonstrated in airborne fur¨ Luft- und Raumfahrt) research aircraft Falcon was op- measurements (Strom¨ et al., 1997). Orographic effects on erated out of Punta Arenas, Chile, in the southern hemi- cirrus were reported in field studies (Heymsfield and Milo- sphere (SH) campaign and out of Prestwick, Scotland, in the shevich, 1993). The importance of mountain waves for the nothern hemisphere (NH) campaign. The campaigns com- generation of polar stratospheric (ice) clouds in the Arctic is prised 30 − 40 flight hours each in autumn and employed an well recognized (Carslaw et al., 1998). identical set of instruments to measure meteorological vari- Most parcel models used a simplified dynamical setup ables and radiative, chemical, and microphysical properties with constant cooling rates to study the initiation of the of aerosol and cirrus cloud particles. Throughout this work, ice phase in detail (Lin et al., 2002). One model study we use the abbreviations NH and SH to distinguish between pointed out the essential role of scale-dependent vertical the two data sets. wind fields on ice crystal concentration and other cloud pa- We investigate measurements taken at air temperatures be- rameters (Jensen et al., 1994). Another study investigated the low 235 K to focus on pure ice clouds. Most flight patterns influence of waves on cirrus formed by homogeneous freez- were designed to probe young cirrus clouds and their for- ing (Lin et al., 1998). A parameterization scheme describing mation regions in upper cloud layers. Typically, the Falcon the formation of cirrus clouds was derived from basic physi- climbed to high altitudes above the top of any cirrus present. cal principles (Karcher¨ and Lohmann, 2002, 2003), connect- Very often, being clearly above the main cloud layer, the air- craft flew in a veil of patchy cirrus. The aircraft then de- ing ni with the variables w and T describing the dynamical cloud forcing, and with parameters of the freezing aerosol scended into the main cloud layer. Several flight legs at dif- − particles establishing the microphysical link. Simulations of ferent levels of 10 15 min duration were flown within cloud. cirrus clouds in a general circulation model (based on the pa- The flights ended by climbing to the stratosphere or cloud top rameterization) called for the need to better represent subgrid altitude and then starting a stacked downward flight with lev- − scale variability of vertical motions to predict cloud proper- els of only 3 5 min duration. An overview of the INCA ties (Lohmann and Karcher,¨ 2002), a crucial aspect we elu- experiment and the instruments deployed is available else- cidate further in the present study. where (Gayet et al., 2002; Seifert et al., 2002). The role aerosol particles from natural and anthropogenic 2.1 Temperatures and vertical velocities sources play in cirrus cloud formation is not yet adequately understood. The possible aerosol impact on cloud properties The Falcon temperature measurements were made with a is difficult to quantify. Concerns were raised about poten- Rosemount total temperature probe located underneath the tial modifications of cirrus clouds by aircraft particle emis- nose of the aircraft. Airflow approaching the housing and the sions and contrails (Boucher, 1999; IPCC, 1999). One model sensor in flight generates a dynamical heating due to flow study demonstrated that carbonaceous aerosol particles ex- stagnation. This heating must be accounted for using the hausted by aircraft jet engines, when assumed to act as effi- pressure measurements of the airflow system to obtain the de- cient ice nuclei in weak updrafts, could expand cirrus cover sired static air temperature T , constituting the major source and change physical and optical cloud properties (Jensen and of error in the temperature measurements. The uncertainty Toon, 1997). However, systematic studies aimed at assessing of T is better than ±0.5 K. the role of aerosols in cirrus modification that cover a wider The vertical wind component was deduced by substract- range of dynamical forcings – a point we are addressing in ing the vertical component of aircraft motion relative to the more detail in this work – are not available. ground and the vertical component of aircraft motion rela- In the present study, we gain insight into basic properties tive to the surrounding air. The latter is measured with a 5- and formation mechanisms of upper tropospheric cirrus from hole-probe system with a relative