Eight forecast factors of convective initiation (CI) are calculated by using IR1, IR2 and WV TBB of MTSAT-1R satellite data based on target clouds identification. The critical values of forecast factors are proposed in Beijing-Tianjin region, and the 17 days test of convective initiation forecast in Beijing-Tianjin region during 2006 and 2007 is carried out. Results of the test show that this method of CI forecast can well forecast the CI occurs in the next one hour. The method of CI forecast results evaluation is proposed base on the atmospheric motion vectors, and a statistical analysis of the CI forecast results is conducted. Statistical results show that this CI forecast method has good TS score of 0.545, high POD of 0.654 and low FAR of 0.435.

Using a scoring method presented by Mecikalski et al, which uses eight indicators to forecast convective initiation (CI), a convective weather event in Beijing-Tianjin region was analyzed in the experiment based on the MTSAT satellite data. The experiment results show that after necessary modifications of the indicator thresholds, the above method is feasible to forecast CI ~30 min in advance in Beijing-Tianjin region. The principle component analysis (PCA) method also verifies that every indicator among the eight indicators plays an important role in forecasting CI.

Based on the derivation of the formula of channel IR4 albedo, the data including the brightness temperature and the albedo of channel IR4 was applied to analyze a case of the precipitation in summer in Beijing. The results show that channel IR4 albedo will increase and then decrease significantly before the precipitation is generated. Channel IR4 has an earlier feedback on the precipitation than long-wave infrared (IR1) channel and water vapor (WV) channel. In the weak echo phase after heavy precipitation, the brightness temperatures of channel IR1 and WV will continue to decrease while the temperature and albedo of channel IR4 continue to increase. Correspondingly, the appearance of the minimum brightness temperatures of channel IR1 and WV is later than the appearance of the maximum echo of the weather radar.

In order to generate the good initial perturbations of ensemble prediction and use limited ensemble members to simulate the time evolution of the atmosphere probability density function in the phase space, the singular vectors (SVs) method was introduced into non-hydrostatic GRAPES meso-scale model. The basic problem of GRAEPS SVs and the method to construct ensemble members based on SVs were researched. GRAPES SVs were solved for a heavy rainfall case by a southwest China vortex moving on July 2008, and an ensemble simulation experiment was made based on GRAPES SVs. The results indicate that the first 27 GRAPES SVs reflect the key information of analysis error; the root mean square error ( RMSE) of ensemble mean about forecast elements have better forecast skills than control forecast, and its ensemble spread is gradually growth with time evolution, reflecting the key information of forecast error; brier score and ROC skill of rainfall show that this ensemble prediction has higher probability skill, and can provide some guidance for the heavy rainfall forecast.

The reflectivity fields of Beij ing-Tianjin region during 2005-2007 were classified into convective and stratiform components using an automatic algorithm, and then the vertical profiles and diurnal variation of convective and stratiform echoes in this region were analyzed. The results are as follows. 1) An obvious bright band was in the stratiform profile of Beijing-Tianjin region at about the 4 km altitude. The bright band was strongest in the night local time, while weakest in the morning. 2) In this region, the diurnal variations of both convective area and stratiform area were unimodal. The maximum areas occurred in the local evening, while the minimum occurred near local noon. The variation of stratiform area was about 1-2 hours later than convective area. 3) There were two peaks in the diurnal variation of the ratio between convective and stratiform area. One was associated with the thermal convection in the local afternoon, and the other was connected with the moist mass convection happens in the local evening. Both these two types of convection are tconvective systems in Beijing-Tianjin region.

An optical flow method instead of the cross correlation method is proposed for convective precipitation system Compared with the simple algorithm of the cross correlation method, the optical flow method utilizes partial differential equations to compute the optical flow field, and uses strict constraints to get a smooth optical flow field The results show that the optical flow is superior to the cross correlation method for convective precipitation systems which evolve fast

3-D clustering technique is the fundamental part of storm identification. One shortcoming of the clustering technique is that the connection area between adjacent storms may cause the problem of false merger. To surmount this difficulty, a special erosion operation is used. Instead of the constant convolution mask, it uses a dynamic convolution mask which depends on the reflectivity distribution of the storm boundary. The results of the experiment show that this method can successfully recognize false merger. These advantages result in better forecast evaluation scores.

The dynamical and thermodynamic characteristics of typhoon Matsa (0509) were analyzed in vertical circulation, vorticity, divergence and temperature/humidity structure, based on the successful simulation in terms of track, intensity and internal structure of Matsa by using the high resolution mesoscale model MM5.The results show that the obviously asymmetric distribution of cyclone low-level structure effectes the development of mesoscale systems and the precipitation. The modified non-geostrophic wet Q vector convergence centre and the positive vorticity filed on the 500hPa upper level is directly associated with the position of heavy rain caused by the typhoon.

A new and accurate method based on cloud movements, spherical and plane triangular relationships of satellites, earth's core, projected-cloud and true-cloud for determining cloud-top height is developed. Synthetic stereo images from a single satellite are used to test this method. It is demonstrated that the cloud-top height can be determined from stereoscopic measurements of geo-synchronous satellites with vertical accuracy of approximately 500 m. Error analysis of 1-pixel incorrect match was presented.

Based on the conventional observational data, NCEP final analysis and infrared images of GOES-9, the characteristics of cloud systems, large-scale environment, water vapor sources, and precipitation of persisting rainstorms over Huaihe River Basin during June 21-July 22 of 2003 were studied. This persisting-precipitation period can be separated to three stages, and the first stage has the maximum precipitation. The cloud systems are characterized by a large-scale cloud band associated with the Meiyu front and embodied smaller-scale active meso-scale complex systems (MCSs). According to the analysis from temperatures of black body (TBB) of infrared images, the MCSs with TBB less than -52℃ are the key synoptic systems caused heavy rains. It is found that there were 34 MCSs over Huaihe River Basin during June 21-July 22 of 2003. Most of these MCSs only locally occurred and dissipated over Huaihe River Basin and neighbor regions, and no MCS was found moving from Tibetan Plateau. In the lower troposphere, there are three currents transporting water vapor. One of the currents is from southwest. It crosses the equator and is most important for transporting water vapor. The second one is from South China Sea. It is the weakest one. The third is from the tropical Pacific Ocean to the east of Philippines.

Surface observations and satellite data are successfully incorporated to a mesoscale model using the ARPS data assimilation system (ADAS) developed by the Center for Analysis and Prediction of Storms (CAPS). Satellite data are used to improve the initial humidity condition with comprehensive cloud analysis. Results show that both forecast and simulation of a storm process are improved than that when satellite data and complex cloud analysis are not applied.

The developing process of a cyclone over the Yellow Sea is simulated with the numerical model MM5. The model output are demonstrated using visual software, LiveView, which is a five dimension visualization system for large data sets. Using this visual system, the authors are able to study the dry intrusion of air flows from the lower stratosphere and to analyze the reason of its occurrence and its effects on such a fast developing cyclone in detail. Horizontal and vertical cross sections are presented to reveal the thermal and kinetic structures of the dry intrusion. Air parcel motions during the dry intrusion are extensively studied with the technique of trajectory analysis. The mechanism of the occurrence of the dry intrusion and its connection with the cyclone development are discussed. The results suggest that dry intrusion is an important factor in influencing on this cyclone development.

The simulative cloud image, made from the mixing ratios of cloud liquid water, rain water, ice crystal, snow, graupel which were outputted by the Mesoscale Model MM5, resemble the GMS satellite infrared image very well and also show the cloud evolution of typhoon Winnie from tropical cyclone to extratropical cyclone after landing. The simulated image, which provide the three-dimensional distribution of water contents of cloud, can make up the insufficiency of satellite observation which can only examine the top of cloud.

Using the Pennsylvania State University/NCAR mesoscale model (MM5), a 36h simulation of the heavy rainfall process in Hubei province on July 20, 1998 is conducted. The simulation reproduced the evolution of meso-β-scale convective systems which were responsible for the heavy rainfall quite well. And the MCSs' visual images produced by the visualization software LiveView can reveal the life cycle, the features of water materials' distribution and 3-d stream structure of MCS.

Global observation of atmospheric wind fields are potentially the most impotent data in the analysis for numerical prediction. This paper describes the computer image recognition technique of deriving cloud motion winds (CMW) from the IR images of GMS. The first step in the processing is objects classification analysis in an image segment corresponding to an area of 32×32 IR pixels. In this paper, we use direct classification and K-means algorithm to distinguish the high cloud, meddle cloud, low cloud and earth's surface. At the second step we adopt IR cross-correlation technique in the calculation of CMW, the correlation values between the brightness of the template window and the search window are calculated in the search area of the later time image, a cross-correlation coefficient surface is obtained. For the determination of CMW height, the average weight value of IR brightness temperature of object cloud is compared with the climatic statistic vertical temperature profile in this season, and CMW height, i.e. high cloud, meddle cloud or low cloud, is determined.