Data Analysisof Millimeter-wave Cloud Radar And Its Detection Ability Improvement Research

At present, the cloud is one of the most important and difficult to determine the meteorological elements in the global climate model, cloud observation is an important issue in the field of weather detection. As cloud automatic observation achieved breakthrough progress both at home and abroad. The detection of cloud height, cloud amount and other cloud elements have been automated, but not yet regular operation, and also a variety of cloud observation devices based on different theory are in a long-term comparison test. This paper mainly describes the comparison of cloud observation data, from test on a Ka-band millimeter-wave(35)GHz cloud radar(KaCR) and a vaisala laser ceilometers(VCEIL) in Beijing which are installed in the Meteorological Observation Center of CMA and L-band rawinsonde observations(LRAOBS) in Beijing Weather Observatory, means of data fusion for the integration of satellite and ground-based cloud radar, and the study on how to improve the detection capability of KaCR. The main research contents of this paper are as follows:Data acquisition ratio measured by the KaCR and the VCEIL installed in Meteorological Observation Center of CMA from 20 Nov to 31 Dec in 2014 under different visibility conditions are compared, and a comparison test of cloud base heights and cloud top heights measured by KaCR and VCEIL, and a simple analysis of the reasons for the differences between the various devicesare carried out. The results indicate that the detection ability of the KaCR is better than the VCEIL under the low visibility condition, and their difference of detection ability reduces with the visibility increasing. The cloud base heights measured by the KaCR is slightly lower than that measured by the VCEIL and the LRAOBS, and cloud top height measured by the KaCR is slightly lower than that measured by the LRAOBS. Compared with the VCEIL and the LRAOBS.The KaCR can clearly show the process of cloud information and dissipation and the structure changes of cloud, but cannot accurately identify the cloud base position when rain.Coherent integrationis an effective method to improve the echo signal-to-noise ratio(SNR) of cloud radars, and the time of Coherent integration selected in appropriately probably significantly decreases the effects of coherent integration and even makes the SNR drop down. The simulation of SNR gain variety for object with a certain velocity is carried out according to the case that Coherent integration canimprove the echo signal-to-noise ratio(SNR) of cloud radars, and based on the theoretical simulation, generally analyses the vertical IQ data observed by cloud radar and studies the method to simply select time of Coherent integration, combined with the data of neighboring wind profiler at the test base. Since the empirical formula contained the influencing factors, including the intrinsic parameters of the radar(pulse repetition frequency and wavelength) and several physical parameters of the target(altitude, vertical velocity, and horizontal velocity), which could indicate to some extent the cause of changes in the target's coherent structure during the coherent integration and accurately calculate the maximum number of coherent integration that provides the highest power in the signal processing system. Therefore, the method probably has a important reference value to improve the ability of cloud radar detection.The satellite can obtain a large range of cloud level distribution, and the ground-based cloud radar can obtain the vertical structure of cloud, both have their own advantages and disadvantages in terms of cloud observation. There is a special internal relationship with between upward radiation intensity of cloud body and the vertical structure of cloud body. Training samples of multi-infrared brightness temperature data remote sensed by satellite and multi-sites cloud radar data with the cloud base height, the cloud top height and the cloud reflectivity are fed into the BP Neural Network, which can carry out a network model that reflect thespecial internal relationship in maximum. Then the new multi-infrared brightness temperature data are fed into the exiting network model, which invert the correspondingcloud information about the cloud base height, the cloud top height and the cloud reflectivity each range resolution. The preliminary test results indicate that the method is reasonable although the accuracy is not enough. After a long period of diversified test and improvement, the method will become an effective way to solve the cloud observation of the integration of satellite and ground-based devices.