用雷达观测资料改进MM5初始场的初步试验研究

采用同时调整热力学和动力学变量，并考虑湿度与温度的变化相协调的初始化方法，引进雷达资料改善MM5模式的初始场。试验结果表明：经过优化处理的初始场，显著提高了模式对降水特别是甚短时(0～6h)降水的落区和量的预报准确率。其中调整热力学变量对改进降水落区预报比动力学变量的调整更为重要。     

Long-Term Trend and Abrupt Change for Major Climate Variables in the Upper Yellow River Basin

On the basis of the mean air temperature, precipitation, sunshine duration, and pan evaporation from 23 meteorological stations in the upper Yellow River Basin from 1960 to 2001, the feasibility of using hypothesis test techniques to detect the long-term trend for major climate variables has been investigated. Parametric tests are limited by the assumptions such as the normality and constant variance of the error terms. Nonparametric tests have not these additional assumptions and are better adapted to the trend test for hydro-meteorological time series. The possible trends of annual and monthly climatic time series are detected by using a non-parametric method and the abrupt changes have been examined in terms of 5-yr moving averaged seasonal and annual series by using moving T-test (MTT) method, Yamamoto method, and Mann-Kendall method. The results show that the annual mean temperature has increased by 0.8℃in the upper Yellow River Basin during the past 42 years. The warmest center was located in the northern part of the basin. The nonlinear tendency for annual precipitation was negative during the same period. The declining center for annual precipitation was located in the eastern part and the center of the basin. The variation of annual precipitation in the upper Yellow River Basin during the past 42 years exhibited an increasing tendency from 1972 to 1989 and a decreasing tendency from 1990 to 2001. The nonlinear tendencies for annual sunshine duration and pan evaporation were also negative. They have decreased by 125.6 h and 161.3 mm during the past 42 years, respectively. The test for abrupt changes by using MTT method shows that an abrupt wanning occurred in the late 1980s. An abrupt change of the annual mean precipitation occurred in the middle 1980s and an abrupt change of the mean sunshine duration took place in the early 1980s. For the annual mean pan evaporation, two abrupt changes took place in the 1980s and the early 1990s. The test results of the Yamamoto method show that the abrupt changes mostly occurred in the 1980s, and two acute abrupt changes were tested for the spring pan evaporation in 1981 and for the annual mean temperature in 1985. According to the Mann-Kendall method, the abrupt changes of the temperature mainly occurred in the 1990s, the pan evaporation abrupt changes mostly occurred in the 1960s, and the abrupt changes of the sunshine duration primarily took place in the 1980s. Although the results obtained by using three methods are different, it is undoubted that jumps have indeed occurred in the last four decades.     

汉江流域极端水文事件时空分布特征

利用1960-2012年汉江流域15个气象站点的日降雨资料和3个水文站同时期日径流资料,分析了9个极端降雨指数的空间分布规律,运用广义极值分布（GEV）、Gamma分布两种极值统计模型对各站点的最大1 d降雨、最大3 d降雨极值样本进行拟合,遴选描述降雨极值分布规律最优概率模型,进而推算给定重现期下的降雨设计值,并分析其空间分布规律;选用Gumbel、Clayton和Frank这3种Copula函数建立降雨-洪量极值联合分布模型,优选最合适的Copula函数,由此计算给定重现期下的洪量设计值。结果表明：GEV分布模型能更好地模拟降雨极值序列,不同重现期下的降雨极值在空间上均呈西低东高的特征;3种Copula函数中,Frank Copula函数能更好地拟合降雨-洪量相关关系,由此推求的洪量设计值大于单变量拟合设计值。     

带有变量中误差非正态模型中的Box—Cox变换

Ｂｏｘ－Ｃｏｘ变换主要用于得到近似正态分布的随机变量，线性模型中的响应（因）变量Ｙ（或经过变换的Ｙ）通常建立在精确测量无误差的独立变量集上，本文研究独立变量中的测量误差（即变量中误差）对Ｂｏｘ－Ｃｏｘ变换中参数估计影响，研究结果表明，忽略变量中的误差将产生变换参数估计的偏离，该结论由分析论证和随机模拟得以支持。     

四元组标靶辅助下的RGB-D室内点云配准

面向室内弱纹理三维重建需求,本文以RGB-D摄影测量技术获取室内点云为基础,提出了四元组标靶辅助的点云配准方法。该方法首先通过阈值筛选大曲率点,自动识别邻接点云中的辅助标靶,然后采用随机采样一致性表达方法,拟合标靶参数及其中心坐标,并根据拟合参数匹配同名标靶中心,通过刚性转换完成邻接点云粗配准。在此基础上,迭代估算邻接点云间的重叠区域,优化点云间的配准参数,从而实现点云精配准。利用Kinect相机获取两类室内场景各12站点云对本文方法进行测试,试验结果表明,配准后的多站点云间距最大均方根误差优于一个采样间隔,证明了该方法在弱纹理室内点云配准中的可靠性。     

数字城市三维可视化技术的研究

城市三维的可视化作为数字城市核心技术的重要组成部分，已成为GIS界和IT界的一个热点课题。本文利用高分辨率遥感影象提取地物数据，并采取面向对象的空间数据模型对数据进行管理，使用基于图形和基于图象的两种建模技术对建筑物和其他一些复杂的模型如树木、路灯等进行三维重建，然后利用有理函数模型表示遥感影象与地面之间的构像关系，使用纹理映射技术，构建具有高度真实感的城市三维景观图。     

双目视觉与惯导融合的移动机器人定位方法

为了提高移动机器人的定位精度,提出一种双目视觉与惯导融合的视觉SLAM算法.在视觉SLAM前端部分,为了保持直接法计算速度快及特征法精度高的特点,提出一种融合直接法和特征法的半直接法双目视觉里程计.在后端优化阶段,将视觉数据与IMU数据相互融合,在滑动窗口中以非线性优化的方式构建误差函数,优化位姿计算精度.在EuRoc数据集中对本文提出的算法进行试验验证.结果表明,与开源的视觉惯导融合的SLAM系统OKVIS、ROVIO和VINS-Mono相比,本文系统在Machine Hal l与Vi con Room两个场景中的定位精度均得到了明显的提升,同时可以保持较高的运行效率.     

基于区域特征的小波差值变化检测方法

小波变换在图像处理与分析方面的应用已取得很大的成功,如基于小波变换的图像编码、去噪、融合等,但在多时相的遥感影像变化检测方面的研究还很少.提出并实验了一种基于区域特征的小波差值变化检测方法,该方法对于高分辨率的遥感影像具有较好的检测效果,同时提高了运算速度.实验结果表明了该方法的可行性和有效性.     

基于车载序列影像的建筑立面纹理的快速重建

提出了一种基于车载序列影像的建筑立面纹理快速重建的方法。首先进行影像预处理,包括畸变差改正、树木遮挡去除等,然后对影像进行自动纠正,在此基础上,自动选择拼接点对影像进行拼接。实验证明了本文方法的有效性。     

一种正射影像拉花现象的处理方法

在地形变化剧烈的正射影像中,受摄影时中心投影对影像获取的影响,数字微分纠正过程中部分成像信息不足的图像区域会出现被过度拉伸的现象,称之为拉花现象.基于此,深入分析了拉花现象的形成机理,提出了一种基于Z-Buffer的拉花区域检测方法.首先对拉花区域进行标记,再采用对向互补策略,基于相邻未拉花影像对该区域重新进行数字微分...