【作者】 李学华；

【导师】 何子述；

【作者基本信息】 电子科技大学 ， 信号与信息处理， 2013， 博士

【摘要】 多普勒天气雷达分辨率的提高，将有利于提高天气雷达对中小尺度气旋和龙卷风等灾害性的天气探测能力，有利于通过天气雷达获得更精细的天气过程回波特征，为预警和预报提供更加准确的数据依据。要提高天气雷达距离分辨率和方位分辨率，直接办法是增大发射信号带宽和增大天线直径；但这涉及天气雷达发射机、接收机及天线的硬件改动，改造周期长、成本高；对于已布网、正在业务运行的新一代多普勒天气雷达网来说，这是不现实的。本文针对现在正在使用的多普勒天气雷达系统，着重研究在天气雷达接收机后端，通过信号处理方法、数据处理方法及数据获取方式等方面进行改进，来实现多普勒天气雷达分辨率的改善与提高。本文的主要工作和贡献体现在以下几个方面：（1）多普勒天气雷达在现有机械天线连续扫描和谱矩处理方式下，天线有效波束宽度较由雷达天线尺寸和波长等因素决定的固有波束宽度有较大的展宽，从而导致天气雷达实际方位分辨率的降低。本文研究了天气雷达天线有效波束宽度的计算方法，并从理论上分析了天气雷达在不同扫描方式下有效波束宽度的展宽程度；研究了降低雷达方位采样间隔和方位重叠采样以及数据加权两种方法，来减少天线有效波束的展宽，以实现天气雷达实际方位分辨率的提高，并用实验仿真和雷达实际数据验证了两种方法的有效性。（2）针对分辨率提高后，因脉冲积累数减少或信噪比降低而带来的多普勒天气雷达谱矩估计精度降低的实际问题，本文给出了距离过采样和白化滤波算法，与方位重叠采样以及数据加权这两种方法进行有机结合的处理方法，既使方位分辨率得到提高，又使谱矩估计的精度得到改善。论文理论分析了采用该方法后谱矩估计精度的改善程度，并使用实验仿真和雷达实际数据对方法的有效性进行验证。结果表明，使用距离过采样和白化滤波算法，可弥补因减少有效波束宽度展宽、提高方位分辨率而带来的精度降低问题。（3）论文结合天气雷达回波信号模型和距离过采样方法，建立了基于距离过采样回波信号进行距离高分辨率重构的数学模型；把发射脉冲宽度决定的距离分辨单元细分成更小的距离分辨子单元，相当于对距离分辨单元内分布式散射体中心进行更细划分。并在最小方差无失真响应（MVDR）算法研究基础上，提出了利用最陡下降法的改进MVDR算法来实现距离向高分辨率回波信号的重构。仿真结果表明，对点目标和不均匀分布的天气目标，该方法能重构距离分辨率单元内部散射目标的平均功率和径向速度变化特征，实现距离分辨率的提高。（4）针对天气雷达反射率因子的分辨率提高问题，本文从雷达气象方程入手，结合分布式气象目标特点，推导了天气雷达分辨率体积内反射率因子与雷达回波平均功率之间的数学模型，并经离散化处理后建立了反射率因子在距离向和方位向二维分辨率提高的矩阵方程；提出了同时利用距离过采样和增加方位观测密度两种数据获取方式，来获得距离向和方位向部分相关的雷达观测数据样本；并提出将一维、二维截断奇异值分解法（TSVD）和Tikhonov方法两种正则化算法引入到了天气雷达反射率因子的分辨率提高应用中。仿真实验结果表明，基于正则化算法来提高天气雷达反射率因子的距离和方位分辨率具有一定效果，可获得更精细的反射率因子回波结构。更多还原

【Abstract】 The resolution improvement of Doppler weather radar will be helpful to enhancethe detecting ability for these small-scale disaster weather systems, to acquire muchfiner echo structures and to provide more precise radar data for forecasting and earlywarning.To improve the range resolution and angular resolution of weather radar,the directmethods are to increase the radar signal bandwidth and increase the diameter of antenna.However, This methods will be related to a great deal of hardware modifications ofradar transmitter, receiver and antenna, which will result in much longer alterationperiod and huge cost of reconstruction, thus it can be not feasible for theoperation-working weather radar networks of CINRAD. This dissertation will befocused on the research of methods to enhance weather radar resolution through theimprovements of signal processing or data processing at the back-end of radar receiver.The main work and contribution of this dissertation are as follows.(1) The antenna motion combined with time averaging in the spectral estimation ofDoppler weather radar will create an effective broadened beamwidth, which is widerthan that of determined by radar antenna diameter and radar wavelength, and results inthe degradation of effective azimuthal resolution. This dissertation investigates thetheoretical computation method of antenna effective beamwidth for scanning weatherradar and theoretically calculates the effective beamwidth under different volumecoverage patterns (VCP); investigates two methods of decreasing the radar azimuthalsampling interval, overlapping radials and data weighting to decrease the broaden ofeffective beamwidth and improve the azimuthal resolution of weather radar,experimental results from radar simulation and actual radar data have proved theavailability of these two methods.(2) Aiming at the actual problems of the errors in the spectral processing due to thedecrease of pulse average number and signal to noise ratio, this dissertation gives theprocessing methods of combining the methods of range oversampling and whiteningfilter with the ones of overlapping radials and data weighting, which can both enhancethe azimuthal resolution and also improve the accuracy of spectral estimation.Theoretically analyzes their detailed performance of improvement degree, and teststheir availability from radar simulation and actual data. Experimental results indicate that these methods can remedy the accuracy decrease due to the processing of azimuthalresolution and meet the need of radar operation application.(3) Combined with the echo signal model of weather radar and the technique ofrange oversampling, This dissertation builds on the mathematical model for higherrange resolution reconstruction, which divides the basic cell of range resolution intomuch smaller cells and it is equivalent to the divisions for the distributed weather scattercenters in one range resolution cells along range direction. Based on the investigation ofthe algorithm of minimum variance distortionless response (MVDR), this dissertationdevelops a robust MVDR algorithm based on the steepest descent method to realize theecho signal reconstruction of higher range resolution. Experimental results indicate thatfor the target objects and asymmetrical weather objects, these methods can reconstructthe gradient changes of echo power and radial velocity for the scatter centers within therange resolution cell and can improve the range resolution of radar.(4) Aiming at the resolution improvement of weather radar reflectivity factor, thisdissertation derives a mathematical model between reflectivity factor and radar echopower based on the radar meteorology equation and a matrix equation is set up for theimprovement of range and azimuthal resolution, and then an oversampling technique inrange and angular is proposed to acquire multiple partially correlated measurements inphysical weather radar system, and two regularization algorithms of one and twodimensional truncated singular value decomposition (TSVD) and Tikhonovregularization are first provided to be applied in weather radar resolution improvementof reflectivity factor. Experiments results have shown that the proposed methods basedon regularization algorithm are efficient for range and angular resolution enhancementof reflectivity factor and can reconstruct much finer weather echo structure.更多还原