Research On Cloud Parameter And Cloud Phase Inversion Method Based On Micro-pulse Lida

Clouds cover more than half of the Earth’s sky and are the most important regulator of the radiation budget in the Earth-atmosphere system.Strengthening the research on the macro and micro physical properties of clouds is of great significance in analyzing the characteristics of global climate change,ensuring the safety of aircraft navigation and improving the accuracy of weather forecast.In recent years,ground-based atmospheric lidar,as an active optical remote sensing device,has been widely used because of its long detection distance and high accuracy.Under the influence of background noise and empirical threshold selection,the traditional cloud detection algorithm has a high misjudgment rate and missing rate,and the cloud thickness is smaller than the actual situation when retrieving cloud parameters.Traditional cloud phase classification algorithm is difficult to distinguish effectively only by depolarization ratio characteristic and fixed threshold.Aiming at the above problems,the research content of this paper is as follows:1.The cloud measurement principle of micro-pulse lidar and the related pre-processing method for calibrating the backscattered signal are introduced.According to the characteristics of backscattered signals in ideal clean atmosphere and actual conditions,a cloud detection algorithm of two-way reconstructed backscattered signals is proposed.Combining empirical threshold and adaptive threshold,clouds and aerosols are distinguished.Cloud parameters are retrieved through reconstructed simulated cloud-free signals.The correlation coefficients of cloud base and cloud top height were increased to 0.9836 and 0.9334,respectively,and the root mean square errors were reduced to 43.8 m and 280.2 m,respectively,which effectively improved the accuracy of cloud parameter inversion.2.Traditional algorithms for cloud phase inversion using micro-pulse lidar need to delineate the range of the declination ratio to achieve the classification of cloud phase,but the setting of fixed threshold is subjective,which affects the reliability of inversion results.In this paper,based on the cloud parameters of inversion,a cloud phase state inversion algorithm combining the random forest algorithm and the African vulture optimization algorithm is proposed.The African vulture optimization algorithm is used to optimize the tree of the decision tree and the number of randomly selected attributes in the random forest,so that the trained model can obtain better cloud phase state classification performance.The experimental results show that the machine learning method can be applied to the cloud phase inversion,which effectively solves the problem of poor inversion results caused by depolarization ratio.3、According to the cloud parameters and cloud phase state inversion algorithm proposed in this paper,the effective observation data of micro-pulse lidar provided by the Southern Great Plains station of the United States atmospheric radiation survey Program are selected for experiment and analysis.The results show that the cloud spatial and temporal distribution and phase state information of the station are in good agreement with the reference standard.The monthly frequency and phase distribution of cloud cover also accord with the characteristics of this region.