Interpreting the results of a high-level clouds (HLCs) lidar study requires a comparison with the vertical profiles of meteorological quantities. There are no regular radiosonde measurements of vertical profiles of meteorological quantities in Tomsk. The nearest aerological stations are several hundred kilometers away from the lidar and perform radiosonde measurements only a few times a day, whereas lidar experiments are performed continuously throughout the day. To estimate meteorological conditions at the HLC altitudes, we propose to use the ERA5 reanalysis. Its reliability was tested by comparing with the data from five aerological stations within a radius of 500 km around Tomsk. A labeled database of the lidar, radiosonde, and ERA5 data (2016–2020) for isobaric levels 1000–50 hPa was created. The temperature reconstruction error over the entire altitude range was characterized by an RMSE of 0.8–2.8 °C, bias of 0–0.9, and Corr ~1. The accuracy of the relative vertical profiles (RMSE 25–40%, Bias 10–22%, and Corr <0.7) and specific humidity (RMSE 0.2–1.2 g/kg, Bias ~0 g/kg, and Corr ~0) at the HLC altitudes were unsatisfying. The ERA5 data on wind direction and speed for the HLC altitudes were promising.

The special features of the applicability of artificial neural networks to the task of identifying relationships between meteorological parameters of the atmosphere and optical and geometric characteristics of high-level clouds (HLCs) containing ice crystals are investigated. The existing models describing such relationships do not take into account a number of atmospheric effects, in particular, the orientation of crystalline ice particles due to the simplified physical description of the medium, or within the framework of these models, accounting for such dependencies becomes a highly nontrivial task. Neural networks are able to take into account the complex interaction of meteorological parameters with each other, as well as reconstruct almost any dependence of the HLC characteristics on these parameters. In the process of prototyping the software product, the greatest difficulty was in determining the network architecture, the loss function, and the method of supplying the input parameters (attributes). Each of these problems affected the most important issue of neural networks—the overtraining problem, which occurs when the neural network stops summarizing data and starts to tune to them. Dependence on meteorological parameters was revealed for the following quantities: the altitude of the cloud center; elements m₂₂ and m₄₄ of the backscattering phase matrix (BSPM); and the m₃₃ element of BSPM requires further investigation and expansion of the analyzed dataset. Significantly, the result is not affected by the compression method chosen to reduce the data dimensionality. In almost all cases, the random forest method gave a better result than a simple multilayer perceptron.

The task of laser sensing of droplet clouds by coaxial lidar is considered. Lidar return due to single scattering is formed in the volume bounded by the radiation pattern of the transmitter, while the double-scattering is determined by a receiving system field of view. The volume of the scattering medium exceeding a receiving system field of view forms the signal higher scattering orders ( < 2). The results of the numerical modeling of the distribution (in the recording plane) polarization characteristics of lidar signal from droplet clouds in the double scattering approximation in comparison with the results of the physical model experiment simulating sounding of a droplet cloud are discussed in this paper.