Recorded presentation3.6
Using experiments in the Manchester Ice Cloud Chamber to improve modelling of ice-ice aggregation
Paul J. Connolly, Univ. of Manchester, Manchester, United Kingdom; and C. Emersic
Ice-ice aggregation in clouds is important for climate and weather forecasts. Over the UK much of the precipitation that falls to the ground has been ice at some point during its descent, whether for warm or cold fronts. Also in decadal climate integrations, where a doubling of CO2 is assumed, large sensitivities to the assumed ice crystal fall-speed have been found by several authors (Mitchell et al 2008, Sanderson et al 2008). There are many processes that could affect the fall-speed of ice crystals, including the number of crystals nucleated for instance, but ice-ice aggregation, leading to the formation of larger particles, would seem to be a dominant factor. As cloud microphysical parametrisation schemes increase in complexity it becomes more likely that the relative influences of different microphysical processes can start to be seen in the larger scale models and we can therefore move towards decreasing the uncertainty in climate sensitivity integrations. In their early experiments Hosler and Halgren (1960) investigated the efficiency of ice-ice aggregation by drawing small (10 μm) ice crystals from a cloud chamber past a large (3 mm) ice target. By measuring the increase in mass of the large target and by knowing the concentration of ice crystals he was able to derive aggregation efficiencies of the order of 0.1. These data are probably the most comprehensive set of laboratory data to date. In this contribution we will show results from laboratory experiments using the new Manchester Ice Cloud Chamber (MICC), which is a 10m high, 1m diameter stainless steel tube contained within a cold room, whose temperature can be controlled down to -50 degC. Clouds are simulated using two main methods: (1) by quasi-adiabatic expansion of the chamber volume by evacuating the fall tube volume with a pumping system and (2) by the introduction of steam. In these experiments we produced the clouds using method (2), and the result was a supercooled cloud of drops at the temperature of interest. Ice crystals were nucleated via a rapid expansion of a very small volume of air near the top of the chamber and once formed the crystals started to grow rapidly, firstly by vapour diffusion and then by aggregation as they descended under free-fall. The Particle Size Distribution (PSD) was measured at two locations using two separate Cloud Particle Imager (CPI) probes (manufactured by SPEC Inc). The first location was about 5 m below the top of the chamber, where the ice was nucleated, and the second location was at the base of the chamber, 10 m below the point of nucleation. The images of the resulting crystals and snowflakes show aggregates of pristine crystals that depend on the temperature and supersaturation (as expected) and having maximum dimensions of up to 1.5 mm. We found that we could fit exponential functions to the measured PSDs, which characterise the PSD with just two parameters: the `intercept' and the `slope' of the PSD. It was found that these parameters did not change appreciably with time during the course of the experiment and so a steady state assumption could be made. We analysed the results using a theory described by Mitchell (1988) for exponential PSDs, which allowed us to estimate the aggregation efficiency, based on the geometrical sweep out assumption. Our initial results suggested that the aggregation efficiency was close to unity for two experiments. We have since performed a large array of experiments to test this further. At the conference we will provide estimates of the aggregation efficiency for these experiments. Interestingly our initial results corroborate a recent paper by Field et al (2006) who presented aircraft measurements during a LaGrangian spiral descent taken from the crystal-face project. This data suggested the aggregation of similar sized ice particles was higher than previously thought. In order to analyse this further we have applied a more complex model of the chamber to see if the conclusions are the same. The model is 1-D with height as the dimension and calculates the evolution of the laboratory clouds in time from an initial condition. The importance of specific processes can be isolated by switching them on or off in the model scheme. In this model both the bin and bulk microphysical approaches can be used. The talk will discuss some necessary changes to the microphysical schemes that are necessary to yield agreement with the laboratory data and also discuss some possible implications. The results of this modelling exercise will be presented at the conference and hypotheses for the high aggregation efficiencies will be discussed.
References Mitchell, D., P. Rasch, D. Ivanova, G. M. McFarquhar, and T. Nousiainen, 2008: Impact of small ice crystal assumptions on ice sedimentation rates in cirrus clouds and gcm simulations. Geophys. Res. Lett., 35(doi:10.1029/2008GL033552), L09,806. Mitchell, D., 1988: Evolution of snow size spectra in cyclonic storms. Part I: Snow growth by vapour deposition and aggregation. Journal of the Atmospheric Sciences, 45, 3431-3451 Sanderson, B. M., C. Piani, W. J. Ingram, D. A. Stone, and M. R. Allen, 2008: Towards constraining climate sensitivity by linear analysis of feedback patterns in thousands of perturbed-physics gcm simulations. Clim. Dyn., 30, 175–190. Hosler C. L. and R. E. Halgren, 1960: The aggregation of small ice crystals. Discussions from the faraday society, 30. 200-206 Field, P. R, A. J. Heymsfield, A. Bansemer, 2006: A test of ice self collection kernels using aircraft data, journal of the atmospheric sciences, 63, 651-666
Session 3, Laboratory Studies
Monday, 28 June 2010, 1:30 PM-3:00 PM, Cascade Ballroom
Previous paper