Optical and geometrical characteristics of high-level clouds from the 2009-2023 data on laser polarization sensing in Tomsk
I.D. Bryukhanov1,2, O.I. Kuchinskaia1, E.V. Ni1, M.S. Penzin1, I.V. Zhivotenyuk1, A.A. Doroshkevich1, N.S. Kirillov1, A.P. Stykon1, V.V. Bryukhanova1, I.V. Samokhvalov1
1National Research Tomsk State University, Tomsk, Russia
2V.E. Zuev Institute of Atmospheric Optics of Siberian Branch of the Russian Academy of Science, Tomsk, Russia
Keywords: high-level clouds, oriented ice crystal, polarization lidar, backscattering phase matrix, radiosonde observations, ERA5 reanalysis, database, artificial neural network, simple multi-layer perceptron, random forest method, primary component analysis, autoencoder, cross-validation
Abstract
To improve the accuracy of weather and climate forecasts, a deeper understanding of atmospheric processes and phenomena, which are determined, among other things, by high-level clouds (HLCs), is required. The experimental results on polarization laser sensing of high-level clouds are presented. The data of systematic (from December 2009 to present) lidar measurements performed with the high-altitude matrix polarization lidar developed at the Tomsk State University are combined. Optical (backscattering phase matrix, optical depth, and scattering ratio) and geometric (lower and upper boundary altitudes and vertical thickness) characteristics of clouds are determined from the lidar measurements. The dataset is supplemented with corresponding vertical profiles of meteorological quantities (temperature, relative and specific humidity, and wind direction and speed) obtained from radiosonde observations and ERA5 reanalysis. The frequency of lidar detection of HLCs and those of them which are characterized by the preferred horizontal orientation of non-spherical ice particles is estimated. The results were combined into a database and used to create a software product based on neural networks to retrieve the dependencies between the atmospheric meteorological parameters and HLC optical characteristics. The database can be used for various training options in solving problems of atmospheric optics including independent ones.
|
