直线特征的多层次仿射不变匹配方法研究

针对直线两侧表面视差非连续、影像几何变形大等问题,提出了一种直线特征的多层次、仿射不变匹配方法。首先利用核线几何、相对方位、重叠距离、长度比等约束条件确定候选直线,根据仿射不变参数纠正直线支撑窗口并计算两侧邻域灰度加权相似性,得到线特征两侧表面视差连续情况下的直线匹配结果,并利用左右双向一致性检验剔除错误匹配;然后对未匹配直线采用顾及线特征两侧表面视差不连续的单侧邻域灰度加权相似性准则和双向一致性检验得到最终匹配结果。不同纹理类型的倾斜立体影像数据验证了方法的有效性,为倾斜立体影像直线特征匹配提供了一种可行的思路。     

一种协同时空地理加权回归PM2.5浓度估算方法

针对PM2.5浓度估算中时空特征考虑不足和样本量较少的问题,该文将协同训练和时空地理加权回归相结合,提出了协同时空地理加权回归。采用两个不同参数的时空地理加权回归模型作为回归器,利用一个回归器训练另一个回归器的未标注样本,选择最优结果作为标注样本加入标注样本,通过不断学习扩大标注样本量提升模型的回归性能。以京津冀地区2015年3-7月的PM2.5浓度数据为实验数据,利用气溶胶光学厚度产品、温度、风速和相对湿度进行建模,采用不同核函数的时空地理加权回归作为对比方法进行实验。结果显示,协同时空地理加权回归性能比基于Gauss核函数时空地理加权回归提升了10%,比基于bi-square核函数时空地理加权回归提升了6.25%,证明该文方法能够提升时空样本数量不足时的PM2.5浓度估算精度。     

西藏错那洞淡色花岗岩地球化学特征、成岩时代及岩石成因

错那洞淡色花岗岩是西藏北喜马拉雅淡色花岗岩带的重要组成部分。通过地球化学分析揭示其具有富硅(SiO_2含量为74.20%~74.52%)、贫铁(Fe_2O_3含量为0.04%~0.20%,FeO含量为0.40%~0.58%)、贫镁(MgO含量为0.06%~0.14%)、钙碱性(σ为2.15~2.32)、强过铝质(A/CNK为1.11~1.15)的地球化学特征。稀土元素总量较低(∑REE为47.24×10~(-6)~57.59×10~(-6)),轻稀土元素富集(LREE为39.85×10~(-6)~49.23×10~(-6)),重稀土元素亏损(HREE为6.91×10~(-6)~8.68×10~(-6)),有明显负Eu异常(0.49~0.80);富集Rb、Th、U、K等大离子亲石元素,亏损Nb、Ta、Zr、Ti等高场强元素。锆石U-Pb测年结果显示,错那洞淡色花岗岩形成中新世(21 Ma),属北喜马拉雅淡色花岗岩晚阶段峰值期(24~12 Ma)产物。锆石εHf(t)值为负值,且变化较大(-3.92~-17.64),说明其岩浆源区为壳源,以变泥质岩为主,可能存在多种物质组分的混合。初始岩浆结晶温度应不超过675~702℃,构造背景为后碰撞环境,是高喜马拉雅结晶岩系在板片快速折返过程中发生减压熔融而形成的产物。     

淤泥质酸性土水泥土强度试验研究

水泥搅拌桩加固有机质含量较高和呈酸性的土层时,我国规范建议采用现场试验的方法。通过深层搅拌法加固太湖应天河工程富含有机质的淤泥质酸性土地基的多组现场水泥土强度试验,分析了淤泥质酸性土中影响水泥土强度的各种因素。试验和工程实践表明,使用高标号水泥(525#)并掺入适量的外加剂可大幅提高酸性淤泥水泥土的强度, 复合地基承载力可达120 kPa以上,水泥搅拌桩可以加固淤泥质酸性土地基。     

数字化体应变仪观测值突跳的分析

山东长清台数字化体应变仪观测曲线自1999年以来出现几次形态类似的大幅度突跳,为正确认识这类现象,在长清台做了交直流切换对观测值影响试验,并对水位、气压资料进行综合分析。研究结果表明,水位、气压、电源切换等因素不会引起体应变仪观测曲线的突跳。对比其他类似观测分析推测,这类突跳可能与“慢地震”事件或“岩孔崩落”现象有关。     

全国水环境信息数据库的设计与实现

运用地理信息系统技术与数据库技术,通过分析全国160多个水质监测站11年的历史监测数据,将空间数据库与数值数据库联合集成,提出了一种基于SuperMap地理信息系统软件和SQLServer关系数据库管理系统的全国水环境数据库的建立方法,并阐述了全国水环境信息数据库的设计思想及实现过程。      

High-Resolution Mesoscale Analysis Data from the South China Heavy Rainfall Experiment (SCHeREX): Data Generation and Quality Evaluation

In this study,the observational data acquired in the South China Heavy Rainfall Experiment (SCHeREX)from May to July 2008 and 2009 were integrated and assimilated with the US National Oceanic and Atmospheric Administration's (NOAA) Local Analysis and Prediction System (LAPS; information available online at http://laps.fsl.noaa.gov).A high-resolution mesoscale analysis dataset was then generated at a spatial resolution of 5 km and a temporal resolution of 3 h in four observational areas:South China,Central China,Jianghuai area,and Yangtze River Delta area.The quality of this dataset was evaluated as follows.First,the dataset was qualitatively compared with radar reflectivity and TBB image for specific heavy rainfall events so as to examine its capability in reproduction of mesoscale systems.The results show that the SCHeREX analysis dataset has a strong capability in capturing severe mesoscale convective systems.Second,the mean deviation and root mean square error of the SCHeREX mesoscale analysis fields were analyzed and compared with radiosonde data.The results reveal that the errors of geopotential height,temperature,relative humidity,and wind of the SCHeREX analysis were within the acceptable range of observation errors.In particular,the average error was 45 m for geopotential height between 700 and 925 hPa,1.0-1.1℃ for temperature,less than 20％ for relative humidity,1.5-2.0 m s-1 for wind speed,and 20°-25° for wind direction.The above results clearly indicate that the SCHeREX mesoscale analysis dataset is of high quality and sufficient reliability,and it is applicable to refined mesoscale weather studies.     

High-Resolution Mesoscale Analysis Data from the South China Heavy Rainfall Experiment (SCHeREX): Data Generation and Quality Evaluation

In this study, the observational data acquired in the South China Heavy Rainfall Experiment (SCHeREX) from May to July 2008 and 2009 were integrated and assimilated with the US National Oceanic and Atmospheric Administration’s (NOAA) Local Analysis and Prediction System (LAPS; information available online at http://laps.fsl.noaa.gov). A high-resolution mesoscale analysis dataset was then generated at a spatial resolution of 5 km and a temporal resolution of 3 h in four observational areas: South China, Cent...