Based on the theory of spectral filtering and the classical 7-layer model of atmospheric turbulence, the performance of the adaptive optics system on ground layer under different number of guide starsand different guide star arrangements is explored by using the theoretical filter model of the Ground Layer Adaptive Optics (GLAO) system and the error transfer function (ETF) under the condition of finite guide stars. The results show that when 1 guide star is in the center and the other 5 guide stars are evenly distributed around the edge of the field of view, the calibration performance of the system is the best. The research results have important guiding role and reference value for the multi-conjugate adaptive optics (MCAO) system (including GLAO system) of the one-meter new vacuum solar telescope being developed.

Using the WINDCUBE 100S Doppler wind lidar (DWL) and L-band radiosonde (LRS) installed at Beijing National Basic Meteorological Observing Station, the wind measurement synchronous comparative observation test was carried out from January 1 to May 31 in 2020．Based on the time and space matching of the observation data, the data quality of wind lidar was analyzed taking L-band radiosonde data as the reference standard from three aspects, i.e., the height of the wind profile, the wind speed and the wind direction. The results show that the maximum detection height of wind lidar was not less than 2000 m in 56.5% of the observation time and less than 1000 m in 2.9% of the observation time. The two set ofwind measurements from lidar and L-band radiosonde have good consistency with each other. The total correlation coefficient of 8491 groups of matched observation data for wind direction and wind speed are 0.965 and 0.986, respectively. The mean error (ME) and root mean square error (RMSE) of overall wind direction (WD) and wind speed (WS) are -1.3° and 16.1°, 0.21 m·s-1 and 1.06 m·s-1, respectively. Due to the weak signal-to-noise ratio of the observation data ofwind lidar, the amount of reliable matching data is reduced above 2000 m altitude, which results in a negative impact on the reliability of the consistency comparison. These results can provide references for the reliability test method of wind lidar andits application in meteorological.

In order to explore the characteristics of air pollution and their relationship with meteorological elements in the three major urban agglomerations around the Bohai Sea, China, based on the daily report data of air quality index from 2015 to 2019 and the meteorological element data of the same period, the spatial autocorrelation model and Pearson correlation coefficient are used to analyze the correlation between them. The results show that: (1) For the three major pollutants studied, PM2.5, PM10 and O3, the main pollutants are PM2.5 and PM10 in winter and O3 in summer. Except O3, the concentrations of other pollutants show a downward trend from 2015 to 2019. (2) The air quality index (AQI) is the lowest in autumn and the highest in winter, and the air quality is gradually improving interannually. (3) AQI has an obvious spatial agglomeration phenomenon, which gradually changes from the good air quality area to the air pollution area from the coastal area to the inland area. (4) The correlations of AQI, PM2.5, PM10 and O3 with various meteorological elements are different in different seasons, and the correlation intensity shows a non-linear change.

On the basis of breaking the coherence of each subprocess in traditional Monte Carlo method in ray tracing, a flexible reverse forward Monte Carlo method is developed to study vector radiative transfer in media. The distribution of atmospheric refractive index is simulated by establishing a one-dimensional gradient refractive index translucent medium model, and the Stokes vector of the medium considering polarization is studied. The results show that the emissivity of different surfaces of graded-index medium is different, and the emissivity of scattering medium when polarization is considered is different from that of scalar method. It is shown that the reverse forward Monte Carlo method can be used to solve the complex problems such as vector radiative transfer.

Based on the self-developed aerosol extinction spectrometer, continuous observation was carried out at Shouxian Observation Station in Anhui Province, China, and the aerosol optical properties in thisarea from May to December 2016 were studied. Combined with the different atmospheric factors, the time series change and daily change law of observation results were analyzed, the correlation between extinction coefficient and PM2.5 mass concentration, scattering coefficient were compared, and the influence of wind speed and direction on the variation of extinction coefficient in Shouxian was discussed. The results show that the time series and diurnal variation characteristics of atmospheric extinction coefficient in Shouxian area are obvious, and they are affected by different meteorological factors. Compared with other seasons, the air pollution in autumn and winter is more serious, and the number of days of polluted weather is increased significantly. Compared with clean weather, the extinction coefficient, PM2.5 mass concentration and scattering coefficient of polluted weather are increased significantly. Especially in winter, pollution events are more frequent. For example, in December, the average concentration of PM2.5 in 24 h exceeds 150 μg·m-3, and the extinction coefficient, PM2.5 mass concentration and scattering coefficient of the month reach the monthly average maximum value during the observation period. During the observation period, according to the diurnal variation, the extinction coefficient, PM2.5 mass concentration and scattering coefficient fluctuate greatly in the daytime, and the lowest values appear in the afternoon. The maximum extinctioncoefficient appears at about 08:00 a.m., and the minimum value appears at about 16:00 p.m.. In addition, the correlation coefficients of aerosol extinction coefficient measured by extinction spectrometer with PM2.5 concentration and scattering coefficient are 0.91 and 0.83, respectively, indicating that the measurement results of the developed extinction spectrometer are in good agreement with those of other instruments.

A new fast gas recognition algorithm was developed based on passive Fourier transform infrared spectrometer, and then the improved momentum gradient descent method was used to realize the fast fitting the measured brightness temperature spectra. This method does not need to measure the background spectrum in advance, and can directly subtract the background interference of atmospheric gas and sky from the measured spectrum. Besides extracting the composition and concentration of the pollutant gas, it can also obtain the concentration-path-length of the main gases in the atmosphere in real time. This method is suitable for the identification and analysis of the pollutant gas with weak signals in the low altitude background.

The rolling prediction of atmospheric SO2 concentration using radial basis function (RBF) neural network based on principal component analysis (PCA) algorithm is presented. Based on the meteorological data and air quality parameters from September 1, 2019 to October 31, 2020 in Daxing district of Beijing, the PCA-RBF prediction model is constructed by combining the stepwise regression method to select the high correlation parameters between SO2 and meteorological factors as input samples. Then the PCA-RBF prediction model is used to predict the SO2 concentration in Daxing area on a certain day, and the prediction results are retained as the input parameters of the prediction model for the next day. In this way, the predicted value is continuously extended forward for the following analyzation and prediction, so as to realize the rolling prediction of SO2 concentration. Comparing the predicted results of RBF network and PCA-RBF network, the error and correlation coefficient of expected value and predicted value of PCA-RBF model are 0.03 μg·m-3 and 0.9989. It is shown that the PCA-RBF network model can accurately predict the variation trend of SO2 concentration, and provide a new technical support for further solving the air pollution problem.

The long-term impact of aerosol on ocean in the subarctic Pacific was investigated based on satellite remote sensing data from MODIS (2003-2018). The spatial, temporal distributions of aerosol optical depth (AOD) and marine net primary productivity (NPP) and their correlations in the area of interest divided into four areas every 10° in longitude in the subarctic Pacific (40° N～50° N, 160° E～160° W) were analyzed and compared. The results show that AOD and NPP in each ares vary seasonally, with AOD peaking in July and NPP peaking in August or September, and both of them have an obvious decreasing trend from west to east. According to the delay correlation analysis of the monthly-averaged long time series data, it is found that AOD in each area has a strong correlation with NPP with a lag of 1-2 months, and the highest Pearson correlation coefficients of the four aeras from west to east are 0.75, 0.84, 0.79 and 0.74, respectively. Through the classification of aerosol according to the AOD peak in July, it is found that dust is an important aerosol type affecting NPP in this area. And it is also found that there is no obvious change trend from west to east for sea surface temperature (SST) and photosynthetically available radiation (PAR) in each area, which effectively excluded their influence on the spatial variation trend of NPP from west to east.

Because the spaceborne atmospheric composition limb detection spectrometer needs to use the reflection of the reflector when collecting the optical information in the height direction, in order to meet the scanning requirements of the spectrometer, a scanning pendulum mirror drive control system is designed, and the spectral data acquisition in the scanning direction is simulated and verified on the ground. The designed drive control system is basedon STM32 microcontroller. DRV8833C motor drive circuit and closed-loop digital PID regulator are adopted in the system, a finite rotation angle DC brushless torque motor is used as the actuator of the spectrometer swing mirror control system, and an 18-bit single-turn absolute value photoelectric encoder is used as position sensor. Then the swing mirror control system and test platform are built, the performance of the swing mirror system is verified by the photoelectric autocollimator, and the pointing accuracy of the system is tested. The results show that the average pointing accuracy is less than 12.15'', the maximum deviation is less than 19.6'', and the minimum deviation is less than 7.46'', which meets the requirements of the main indicators in the pre-research stage of the project.