| Millimeter Wave Cloud Radar(MMCR)can not only detect the features of micro particles in cloud,but also describe the internal microphysical structure of the cloud;therefore,MMCR has been applied widely in the cloud observation.However,due to the influence of meteorological factors produced by uneven atmospheric refractive index and turbulence,it is difficult to accurately detect cloud information in boundary layer,especially in the complex condition of cloud,precipitation and clear-air echo mixed.It is of great significance and practical value to eliminate the interference of clear-air echo in the complex weather background,and to extract useful meteorological information from radar echoes,such as the cloud and precipitation information.This paper has realized the automatic classification of clear-air echo and meteorological echo in the cloud radar echoes by using machine learning algorithm,and the distribution characteristics of clear-air echoes over time are also been found.The main contents of this paper are summarized as follows:The main object of this paper is clear-air echo caused by turbulent scattering,so it is necessary to analyze the scattering mechanism of it theoretically,and reveals its physical nature.Turbulence in the atmosphere causes turbulent scattering for electromagnetic waves emitted by radar,and the relationship between the radar wavelength and the detected turbulent block follows Bragg’s law.This project uses laser ceilometer,MMCR,L-band sounding system,and all-sky camera to judge cloud information comprehensively,thus clear-air echo and cloud echo can be extracted.First,their principle,data and geographical location distribution are introduced,and then the data acquisition capability and the ability to detect cloud are compared for MMCR,laser ceilometer,and L-band sounding system by using the two months of data on December 1,2015 to January 31,2016.The results show that all three instruments can detect the cloud bottom accurately,the ability of laser ceilometer to detect cloud is the most powerful,and the cloud base height detected by MMCR is slightly lower than those of laser ceilometer and L-band sounding device.This paper presents a new algorithm for the automatic identification of clear-air echo of MMCR,and it mainly consists of three steps: feature extraction,feature selection,and the recognition of clear-air echo.Firstly,the basic products and power spectrum data of MMCR are both used to construct a sample collection contained 28 features;then,the feature is filtered by Relief algorithm,and the optimal feature subset can be obtained;finally,the neutral network algorithm is used to realize the automatic identification of clear-air echo.Moreover,the algorithm model was tested by using data from March,2016 to August,2016,and results show that the recognition rate of clear-air echo can reach 95.86% for simple weather conditions;for complex weather conditions,the recognition rate is 88.38%.In order to study the distribution characteristics of clear-air echo of MMCR,this paper firstly extracts all the clear-air echoes for the MMCR data from September,2015 to August,2016;then,the monthly clear-air echo reflectance graphs are calculated;finally,the coverage rates of clear-air echo in every month are also counted,and the distribution characteristics with season of clear-air echo are summarized.The results show that,clear-air echo occurs most frequently in summer,and there is almost no clear-air echo in winter;in the process of spring to summer,the coverage rate of clear-air echo gradually increases,and the coverage rate decreases gradually in autumn to winter. |
