Comparison of particle size distributions of subvisual cirrus clouds measured by optical particle counters and a lidar with those calculated by a parcel model

1. Introduction Subvisual cirrus clouds (SVCs) generally exist at a height of around 17 km in the tropical tropopause layer (TTL), hence they would lead to the dehydration of air entering the stratosphere. In order to research SVC, in situ measurements are effective. However, since all in situ measurements were airborne measurements, they are expensive to conduct and are not suitable for measuring the vertical profiles of the particle size distribution. Hence, we launched 11 balloon-borne optical particle counters (OPCs) from April to June 2003 in Thailand (17.9 oN, 99.5 oE). We also carried out a lidar measurement at the same place.

2. Optical particle counter Our OPC has 8 channels, of which radii are from 0.15 to 3.5 um for spherical particles, to measure the accumulated number concentrations. Because ice particles are not spherical, the measurement error is estimated by the finite-difference time domain method (FDTD). The minimum detectable number concentration and the vertical resolution are approximately 1.5e4 number/m3 and 50 m in the TTL.

3. Observation Of 11 OPC launches, cirrus cloud was detected at 10-15 km high on 7 occasions, cirrus was detected in the TTL in 6 cases, and simultaneous OPC and lidar measurements were made on two occasions. Comparison of lidar and OPC measurements reveal that the cloud height of cirrus in the TTL varies by several hundred meters over distances of tens kilometers; hence the height is not horizontally uniform. The mode radii of particles constituting the clouds are estimated by lidar and OPC measurements to be less than approximately 10 um. The regression lines of the particle size distribution with and without cirrus cloud exhibit similar features at equivalent radii of < 0.7 um. In addition, a peak between 0.5 um and 1.7 um in the ratio of the standard deviation of count values to that of the Poisson distribution of the averaged count values, where the ratio shows the vertical homogeneity of the particle number concentration.

4. Model result We then calculate cirrus cloud nucleation processes by using a parcel model; our model is the same as Spice et al. (1999, QJRM) but we applied the freezing rate (Tabazadeh et al. 2000, GRL) and neglect the heterogeneous nucleation. The results whose initial relative humidity with respect to ice and ambient temperature are 120 % and -80 oC are similar to the abovementioned observational results when the vertical wind velocity was 20 cm/s * cos(2pi t/7min), the Brunt-Vaisala frequency. On the other hand, it is not easy to reappear in other conditions. For example, if the vertical velocity is greater than 1 cm/s, a number concentration of particles with sub-micro meter in radius becomes much greater than that of observational results. Hence, knowledge of a number concentration of submicron particles is important to know the ice nucleation mechanism.