For solving the dual-value problem of RSD raindrop size distribution retrieval algorithm using GPMCO DPR data, a solution is put forward: combining the observation data of S-band weather radar and GPM-CO DPR to provide determinant for the dual-value problem. On the basis of completing the retrieval algorithm, the radar simulation program, Quickbeam, is utilized to validate the algorithm, and actual observation data is utilized to validate the algorithm as well. The results of theoretical validation show that the correlation coefficients between retrieved data and input data are over 0.99, and the results of validation with actual observation data show that the retrieved RSD can reflect the raindrop distribution situation of the rainfall process. The algorithm can help to solve the dual-value problem to o certain degree.

Due to the combined effects of lake and Gobi, the atmospheric boundary layer located in Gobi desert regions of the Bosten Lake basin is found to possess some special regional characteristics. Based on the data of the developing meteorological observation experiments conducted in the Bosten Lake of Xinjiang Province, the height characteristics of the atmospheric boundary layer from May 21 to August 28, 2013 are analyzed. The result shows that the most prominent feature of the Gobi area of the Bosten Lake Basin is that the boundary layer often has the deep convection during the typical sunny days in the summer. The statistics also shows that in 45 days out of 100 days, the average height of the atmospheric boundary layer is above 3000 m, and the maximum height reaches 4400 m. The causes of forming the deep convective boundary layer are explored. It is argued that the atmospheric thermodynamic properties and the lake wind shear are important factors to form the deep convective boundary layer.

Cloud occurrence frequency (COF, the ratio between the number of records with detected clouds with respect to the total available records) and the distribution of COF within one day below 8 kilometers in summer from 2007 to 2009 over Beijing is studied. Cloud base height (CBH) of one layer clouds, the characteristics of convective CBH and the relationship between convective CBH and relative humidity (RH) measured on the surface are further studied. Results show that cloud occurrence frequency is about 12% in summer over Beijing. Clouds are more likely to occur at night than in the afternoons. The distribution of CBH has one peak at 700 (±50) meters and low clouds (ranging up to 3000 meters) account for 82% of all clouds. Convective CBH ranges from 400 to 2000 meters for the cases studied, but do not vary much within each day. CBH can be conveniently derived from linear relationship between CBH and surface RH based on Adiabatic Parcel Model. Mean relative standard deviation of CBH is 0.321.

By introducing levy-lindeberg theorem to raingauge networks planning, the authors obtain a linear relationship between the measurement precision of areal rainfall (relative error or permissible error) and raingauge networks density (average station spacing). The slope of relative error growing is directly proportional to the mean square error of random observation error, inversely proportional to square root of the areal rainfall and region area. Based on simulated rainfall fields and highdensity rainfall data during Mei-yu season in Anhui Province, conclusion above is validated by the statistic analysis reducing the raingauge stations method. The slope of the linear relationship is also investigated. This research is significant to provide certain reference for optimization of raingauge networks location planning. Based on the relationship, the authors calculate 2005-2008 rainfall data during Mei-yu season in Anhui Province and conclude that correlation coefficient between the measurement precision of areal rainfall and raingauge networks density in Huaibei Plain is 0.49 to 0.80, in the mountains of Southern Anhui Province is 0.70 to 1.41. Assuming that permissible error is 20%, the minimum average station spacing (the maximum Raingauge networks density) in Huaibei Plain is 25 km and in the mountains of Southern Anhui Province is 14 km.

The three-dimensional water vapor field is obtained using Monte Carlo tomography algorithm based on the small-scale network of ground-based GPS. After the proof of the accuracy of the tomography water vapor field through the comparison between the profiles from the radiosonde and tomography field, the three-dimensional water vapor field and the data of Weather Radar are used to study the variation of the water vapor in the strong convection and heavy precipitation processes which took place on July 29, 2011. For the first time, the authors propose the continuous change and fine structure of the small-scale water vapor field during the convective processes. For local convection process, the tomography vapor field can capture the water vapor enrichment phenomenon and provide its location 20 minutes ahead the precipitation and it is an indication to the precipitation forecast. For rainstorm process, the water vapor field can explicitly show the position and intensity of the vapor transport source which meet well with the conclusion from the analysis of weather radar and meteorological situation.

The data of the network of 13 GPS are used to perform the sensitivity tests for the important parameters of the Monte Carlo tomography algorithm such as the average distance of the GPS stations, tomography temporal resolution and Monte Carlo random times. The accuracy of tomography rises with the decrease of the average station distance and the increase of the random times. However the decrease of the temporal resolution of the tomography does not help much to the accuracy of the tomography. A new way is also proposed to obtain the final tomography field. The final tomography field is obtained by averaging “sub-optimal” tomography fields instead of the one “optimal”, and it is proved to reduce the error by around 30%.

Radiosonde (RS) data over Beijing in 2008 was employed to analyze vertical distributions of water vapor in this area. A parameterization scheme of water vapor profiles is proposed according to different conditions of Planetary Boundary Layer (PBL). In the condition of Stable Boundary Layer (SBL) or shallow boundary layer, the parameterization scheme of water vapor profiles can be formulated by q=q₀e^z/2624, here q₀ is specific humidity at surface. In the condition of Convective Boundary Layer (CBL), the vertical distribution of water vapor is significantly influenced by CBL. The parameterization scheme of water vapor profiles can be expressed as 1) within the CBL (0≤ z0; 2) near the top of CBL (|z-PBL|<200), q=q₀(A-B(z-PBL)); 3) above the CBL (PBL+200≤z), q=Cq₀(z-5000-PBL), here z is height (unit is metre), PBL is the height of PBL (unit is metre). A=0.72, B=1.4×10^-3m^-1, C=?9.17×10^?5m^-1. For CBL, the IWV would be underestimated by 10% using the exponential distribution. A better estimation is achieved with this parameterization scheme proposed.

The retrieval product of COSMIC (constellation observing system for meteorology, ionosphere, and climate) radio occultation sounding system was verified using the Chinese radiosonde from 2007 to 2010. The horizontal distance between radiosonde station and the occultation event is within 100 km, and the time window is 1 hour. The comparison is performed from 925 hPa to 10 hPa. Result shows that the temperature of COSMIC agrees well with Chinese radiosonde. The mean temperature bias is ?0.08 K, and the standard deviation is 1.67 K. The water vapor pressure of COSMIC has notable systematic bias from radiosonde on layers above 200 hPa. On layers below 200 hPa, the mean absolute deviation of specific humidity is about +0.080 g/kg, and the standard deviation is about 0.692 g/kg, the mean relative error of water vapor pressure is about +69.6%, and the standard deviation is about 160.0%. The COSMIC quality control process fails to detect some extremely small humidity data which causes large relative error. Here also provides a comparison of 6 radiosonde types with COSMIC product. As result is affected by the background bias from radiosonde which differs on different regions, the COSMIC retrieval product could be used as benchmark if precision requirement is not strictness.

This study was based on two-year MODIS aerosol optical depth (AOD) product and PM₁₀ mass concentrations calculated from Air Pollutant Index (API) in Beijing and the hourly averaged measurements at Peking University in Beijing and Yuen Long station in Hong Kong. The direct correlation between AOD and PM₁₀ mass concentration was found relatively low. The correlation coefficient between aerosol surface extinction ( achieved from dividing AOD by seasonal aerosol scalar heights) and PM₁₀ mass concentration increased to some extent. The correlation further improved after considering the influence of relative hfumidity (RH) on aerosol optical properties. This comparison verified that the MODIS AOD can be applied in the evaluation of surface PM₁₀ mass concentration after taking aerosol vertical distribution and influence of RH into consideration.

The optimum interpolation method is used to estimate radar measured rainfall which then be applied to TOPMODEL to simulate discharge of SHIGUANHE catchment during the summer of 1998 in GAME/HUBEX project. Comparison of simulated discharges between radar and rain gauge implements over a 1500-hour series. The results indicate that: (1) Only with a few rain gauges to adjust the radar estimated rainfall, authors could simulate the runoff as good as that from rain-gauge-network-measured rainfall; (2) The accuracy of radar rainfall estimation to heavy rain acts as a more important role than that of light rain; (3) The estimation error of amount precipitation over a period can be treated as an index for runoff simulation ability.