Cirrus Occurrence and Properties Determined From Ground-Based Remote Sensing

Abstract

The ultimate application of this work is constraining the optical properties of cirrus particles, which are poorly understood, by providing an automatic method, using all-sky cameras and an infrared radiometer, to identify the occurrence of the 22° halo formed by cirrus. This is done by interpreting all sky images in terms of a scattering phase function (SPF), from which the halo ratio (HR) is calculated, and by implementing a cirrus detection algorithm to associate HR measures to ice cloud occurrences. Cirrus reflectivity at solar wavelengths is inversely related to the HR which, being an indirect measure of the regularity of the shape of the ice crystals forming the cloud, relates in turn inversely to the asymmetry parameter g. Therefore, the method proposed here to derive statistics of HRs is expected to reduce the uncertainty over the optical and microphysical properties of cirrus. The light intensity measured by the all sky camera is transformed into a scattering phase function, from which the halo formation is identified. This is done by developing image transformations and corrections needed to interpret all sky images quantitatively in terms of scattering phase function, specifically by transforming the original image from the zenith-centred to the light-source-centred system of coordinates and correcting for the air mass and for vignetting. The SPF is then determined by averaging the image brightness over the azimuth angle and the HR by calculating the ratio of brightness at two scattering angles in the vicinity of the 22⁰ halo peak. The instrument transformation and corrections are performed using a series of Matlab scripts. Given that the HR is an ice cloud characteristic and since the method needs additional temperature information if the halo observation is to be associated with cirrus, a cirrus detection algorithm is necessary to screen out non-ice clouds before deriving reliable HR statistics. Cloud detection is determined by quantifying the temporal fluctuations of sky radiance, expressed as brightness temperature (BT), through De-trended Fluctuation Analysis and setting a clear sky fluctuation threshold. Cloud phase discrimination instead is achieved through first constructing an analytic radiative transfer model to obtain an estimate for average molecular absorption cross-section of water vapour within the spectral window of the radiometer. This is done to model the down-welling clear sky radiance, which is in turn used to correct cirrus emissivity and ultimately determine a dynamic BT threshold for the transition from ice to liquid-containing clouds. In addition to the molecular cross section the screen level air temperature and integrated water vapour are used as input parameters to the model. The utilisation of the all sky camera for such quantitative measurement was the particularly novel aspect of this work; this has not been done previously to the best of my knowledge. The cirrus detection method proposed is also innovative in that with respect to previous works it does not rely on the use of additional techniques such as LIDAR or microwave radiometry for discriminating cloud phase. Furthermore, the cirrus threshold proposed is not fixed but accounts for the attenuating properties of the atmosphere below the cloud. Once the cirrus detection algorithm is validated and cirrus occurrences determinable, the HR could be extended to estimating the asymmetry parameter and crystal roughness. These are retrievable, for instance, from in-situ observations of single ice crystal 2D scattering patterns from cloud probes of the SID (Small Ice Detector) type. This would be significant for the constraining of the optical and microphysical properties of cirrus.