Minimum-variance unbiased quadratic estimation of covariances of regionalized variables

The parameters of covariance functions (or variograms) of regionalized variables must be determined before linear unbiased estimation can be applied. This work examines the problem of minimum-variance unbiased quadratic estimation of the parameters of ordinary or generalized covariance functions of regionalized variables. Attention is limited to covariance functions that are linear in the parameters and the normality assumption is invoked when fourth moments of the data need to be calculated. The main contributions of this work are (1) it shows when and in what sense minimum-variance unbiased quadratic estimation can be achieved, and (2) it yields a well-founded, practicable, and easy-to-automate methodology for the estimation of parameters of covariance functions. Results of simulation studies are very encouraging.     

On the joint robust estimation of parameters and variance

A method of joint estimation of the parameters and the variance σ² in the linear model is presented in this paper. The M-estimators with finite rejection points are adopted for estimating the parameters. In order to estimate the variance σ², the Grubbs statistics and the Kurtosis test statistics are adopted to test the residual sequence {;r_i};. And the sample variance of {;r_i}; after discarding the outliers is taken as the estimation of σ². This method of estimating σ² is less computational demanding and more accurate, compared with the well-known method which takes 1.483 med_i |r_i| as the estimation of σ. The breakdown point of the above estimation of σ² is more than 20%, while the brekdown point is less than 10%, if only the Grubbs statistics is used.     

基于雷达资料的上海地区暴雨面雨量计算及应用

根据上海地区小时雷达资料的定量降水估测(quantitative precipitation estimation,QPE)网格产品和经过质量控制的自动站雨量资料进行暴雨面雨量计算方法研究,计算方法为:将暴雨区内自动站周围9个网格点的QPE平均值作为自动站的雷达估测雨量,计算各测站雨量与雷达估测雨量的差值,用克里金插值方法将差值场插值到雷达降水估算产品相同的网格点上,再将网格点上残差插值数据加上网格点上雷达估测雨量,得到各网格点用自动站订正后的雷达估测小时雨量资料。文中24个典型暴雨各测站订正后QPE与实测雨量的平均绝对误差比订正前减小了27%;两个典型暴雨主要降水阶段经国家气象站资料订正后QPE与实测雨量的平均误差比订正前减小了33%～39%,暴雨过程总雨量经国家气象站资料订正后减小了34%～59%。根据以上方法得到的网格点雨量计算和绘制上海地区2007—2015年24个典型暴雨降水区域的小时面雨量值和图、过程面雨量值和图、行政区和水利片面雨量值和图。最后基于.NET Framework 4.0基础架构软件开发平台,使用Microsoft Visual Studio 2012和Microsoft Visual Studio软件开发工具制作"基于雷达资料暴雨面雨量自动化计算查询系统",供业务科研人员实时查询和计算。     

2006―2016年粤港澳大湾区城市不透水面 时空变化与驱动力分析

以遥感和地理信息系统技术为支撑,从城市地理学视角对粤港澳大湾区城市不透水面的时空变化状况进行分析,并探讨了不透水面变化的驱动力。结果表明：1）湾区内城市不透水面扩张现象明显,面积从2006年的12 127.69 km2增加到2016年的20 188.87 km2,年均增长速度高达6.65%;2）区域内城市不透水面空间分布越来越集中,呈现高密度区域聚集;3）交通道路（省道及以上级别）与城市不透水面扩张的速度和方向关系密切,其中与国道的关系最为紧密;4）国民生产总值（GDP）和常住人口的增长都对城市不透水面的扩张产生深刻的正向影响,相关系数分别为0.840 7和0.834 1;5）近10年来,粤港澳大湾区城市不透水面的空间格局变化显著,湾区各城市的发展阶段存在差异性,在未来城市发展中应该协调好各城市不透水面的扩张范围和速度,避免过高不透水面带来生态环境问题。     

基于城市不透水面—人口关联的粤港澳大湾区人口密度时空分异规律与特征

城市人口数据是社会经济各领域的基础数据,高分辨率的空间化城市人口数据则对社会经济各领域的分析和研究具有重要意义。本文首先通过多源遥感技术提取空间分辨率为30 m的粤港澳大湾区2007-2015年间城市不透水面的变化,再利用Dasymetric映射方法得到30 m分辨率的网格化人口密度分布,从而分析大湾区2007-2015年间城市人口的时间和空间变化。通过Google Earth时间序列高分辨率影像采集的样本验证,粤港澳大湾区城市不透水面提取精度均在80%以上;通过统计年鉴中县级人口统计数据,分析大湾区网格化城市人口与统计数据之间的一致性,得到决定系数R²总体在0.7以上。研究表明,粤港澳大湾区城市人口具有特殊的时间和空间分异规律和特征：①大湾区内除了香港和澳门人口分布较稳定,其他城市人口都有不同程度和不同方向的扩张,其中广州、深圳、东莞的人口扩张最为明显;②大湾区城市人口空间分布具有明显的多尺度和多中心特征。总体上,大湾区人口集中在以珠江口为中心的城市群核心区内,离核心区较远的肇庆、江门、惠州人口较为稀疏,城市化程度相对较低,是支撑大湾区经济社会继续深入发展的重要区域。在核心区内,城市人口的分布则在城市尺度和城市群尺度上都体现了多中心分布特征,香港和广州都有多个城市中心,而香港、澳门、深圳、广州则是整个大湾区的4个中心。地理位置上4个中心分布在大湾区的不同地方,可以带动整个粤港澳大湾区的全面发展。研究结果可为粤港澳大湾区在社会经济各领域的分析与规划提供决策支持。     

Where Should the Start Zone Be Located for Xiong’an New Area? A Land Use Perspective

China announced plans on April 1, 2017 to create Xiong’an New Area, a new economic zone about 110 km southwest of Beijing. The new area of national significance covers the three counties of Xiongxian, Rongcheng and Anxin in Hebei Province, and is home to Baiyangdian, a major wetland in northern China. It will cover around 100 km² initially (called “Starting Zone”) and will be expanded to 200 km² in the mid-term and about 2000 km² in the long term. In this context, it is important to assess the land use pattern for the new area’s planning and development. Based on remote sensing interpretation of Landsat OLI images, we examine the current land use features, the potential for built-up land saving and intensive use, and the spatial variation of arable land quality. The results indicate that the arable lands, built-up lands and wetlands account for around 60%, 20% and 10%, respectively. The potential for rural built-up land saving and intensive use is quite large. The arable land quality and use intensity in Rongcheng and Anxin counties are much better than that of Xiongxian. The pros and cons for three potential options for the Starting Zone are listed and compared. The eastern area of Rongcheng county, to the north of Baiyangdian, is believed to be the most suitable Starting Zone. Policy recommendations for built-up land expansion, farmland saving and flood risk mitigation for Xiong’an New Area are also proposed in this paper.     

Characteristics of nonpoint source pollution load from crop farming in the context of livelihood diversification

Based on objective data collected from interviews in typical villages of the Three Gorges Reservoir Area, the present study devised three livelihood scenarios related to rural transformation development: agriculturally dominant livelihood, multiple-type livelihood and non-agriculturally dominant livelihood. Moreover, the present study reports the trend characteristics of nonpoint source pollution load of crop farming in relation to the transformation of dominant livelihood types, and discussed the primary factors which affect livelihood type transformations. Results indicated the following: (1) The current farmland pattern shows a trend of diversification as self-cultivation, cropland transfer and fallow in the sample region. Dynamic characteristics of cultivated land present a special feature that is more “transfer-into” than “transfer-out”. Various scales of planting are represented among the various households, according to the following decreasing order: half-labor household > non-labor household > adequate labor household. (2) The highest pollution loading produced by crop farming occurs in half-labor households while the lowest occurs in non-labor households. With increasing labor, the pollution load per unit area tends to first increase and then decrease within families with enough labor. (3) As the type of livelihood transitions from agriculturally dominant to non-agriculturally dominant, the maximum reduction of total pollution loading produced by the agricultural industry can reach 72.01%. Compared to agriculturally dominant livelihoods, multiple-type livelihoods produce a pollution load reduction yield of 19.61%–29.85%, and non-agriculturally dominant livelihoods reduce the pollution load yield by 35.20%–72.01%. However, the rate of reduction of total nitrogen is not the same as total phosphorus. (4) The non-agricultural characteristics of labor allocation and income promote the transformation from dominant livelihood types to non-agricultural livelihoods, while potential revenue conversion follows a similar trend. In addition, different household types do not display identical conversion rates, according to the following decreasing order: enough labor household > half-labor household > non-labor household. (5) During rapid urbanization and the building of new industrial systems, the livelihood types of rural households have been further transformed to off-farm household types in the mountainous region; this process will lead to the further reduction of pollution load generated by planting and agriculture. Hence, significant decreases in the planting pollution load necessitate the development of control measures to enhance transformations from agricultural to off-farm livelihoods.     

MLP在雷达定量降水估测中的应用

通过临近时次、临近空间降水回波性质相似假设,以人工智能技术结合快速动态分级法(Fast Dynamic Categorical method,FDC)为核心,设计广西区域分钟级雷达定量降水估测产品模型。在单部雷达估测降水时,分两层三次使用感知器寻求合理的降水类型分类Z-I关系,首先将FDC看为一种多分类算法,根据单站雷达各级回波的降水估测结果误差将回波区内的站点分为强站点和弱站点两类,然后分别对这两类站点再次使用FDC建立新的强、弱两类Z-I关系。在多部雷达组网联合估测定量降水时,将各雷达估测值等权重看待,将单部雷达估测作为一个分支,通过连结方式构建一个多层感知器(Multilayer Perceptron,MLP)。无站点回波区采用K近邻算法(K-Nearest Neighbor,KNN)选择合适的MLP求得的Z-I关系估算降水量。对2019年3—10月试验产品进行检验分析,结果表明以区域站组成的训练组小时降水相关系数达0.973 7,以国家级气象台站组成的测试组相关系数达0.825 6。     

相关观测值双因子抗差估计的改进算法

由于一个观测值的粗差将影响多个观测值的残差估值,对双因子抗差估计的迭代计算方法作如下改进：当最大标准化残差的检验结果为粗差或者明显是可疑粗差时,只对粗差观测值构建双因子等价权,其他观测值的权不变。除上述两种情况外,可同时对所有观测值构建双因子等价权进行抗差估计,并基于改进的双因子等价权构建方法简化计算步骤。利用GPS控制网的观测数据进行模拟计算分析表明,该改进算法在含有明显粗差时可提高参数估值的精度;在粗差不明显或个数较少时,可在保持参数估计精度的基础上提高计算效率。     

地下水加坡理论在井筒涌水量预算中的应用

地下水加坡理论可有效的消除抽水试验引起的无效降深,结合云驾岭煤矿井筒检查孔抽水试验资料,对该井简检查孔涌水量进行了预算,预算副井筒啮。砂岩含水层涌水量为523．03m^3／h,实际涌水量为455m^3／h．误差率14．95％,可见运用加坡理论计算的矿井涌水量与井筒实际涌水量基本相符。