部分变量误差模型的整体抗差最小二乘估计

部分变量误差模型(partial EIV model)的加权整体最小二乘(weighted total least-squares,WTLS)估计不具备抵御粗差的能力。鉴于粗差可能同时出现在观测值和系数矩阵中,本文在提出部分变量误差模型WTLS估计的两步迭代解法的基础上,运用抗差M估计的等价权方法,发展了一种整体抗差最小二乘(TRLS)估计方法,并采用一致最大功效统计量确定降权因子。针对WTLS估计两步迭代解法的特点,设计了两个不同的降权方案:第1个方案是在估计系数矩阵元素时,不对观测值降权,仅对系数矩阵降权;第2个方案是在估计系数矩阵元素时,既对系数矩阵降权,同时也对观测值降权。通过对模拟2D仿射变换和线性拟合实例进行计算和分析,结果表明第1方案优于第2方案,并且优于基于残差和验后单位权方差的抗差估计和现有的变量误差模型抗差估计。     

SAR图像海岸线检测的区域距离正则化几何主动轮廓模型

合成孔径雷达(SAR)卫星遥感图像可以极大地提高全国海岸线覆盖频率,然而受到海洋波浪所引起的随机海水表面粗糙度的影响,海岸目标与海水背景边界易混淆不清,因此本文提出了基于区域距离正则化几何主动轮廓模型(RDRGAC),引入距离正则项,解决重复初始化水平集函数为符号距离函数的问题,提高了算法收敛速度。此外,将区域面积项系数与SAR图像等效视数(ENL)建立非线性拟合关系,实现RDRGAC模型根据不同SAR遥感图像的自适应调整,改善海岸线自动提取精度。通过河北省北戴河和大连市金州湾SAR数据海岸线提取对比试验,验证了所提方法的有效性。     

回归-Elman网络在矿坝变形预测中的应用

金矿开展矿坝的变形监测工作,引入多层递阶回归分析模型,有较高的预测精度,但方法较繁琐,计算较复杂。由于变形数据可分离成趋势项与随机项,趋势项可用多元线性回归良好地拟合;随机项的预测,文中采用Elman网络建模计算,最后利用矿坝的实测高程位移数据进行验证,并与多层递阶回归进行比较。结果表明:回归-Elman网络模型比多层递阶回归的预测精度更高,效果更好,且方法简洁实用性强。     

混合核特征加权SVM遥感湿地空间分类

针对湿地空间信息的复杂性和SVM的分类性能,设计一种基于混合核函数的特征加权SVM分类模型,综合利用多种特征信息,避免被弱相关特征所支配,从而提供更佳的映射性能和泛化能力。实验结果表明,该分类模型兼具良好的外推和内推能力,能够有效地融合不同信息源特征,得到更完整和准确的分类结果,在总体精度、Kappa系数等多项指标上都表现出更高的水平。     

加权Logistic回归模型在火山岩型铜矿预测中的应用:以宁芜盆地中段为例

文中探讨了加权Logistic回归模型在宁芜盆地中段火山岩型铜矿预测中的应用。首先,结合研究区的成矿地质背景,提取地质体、构造、围岩蚀变三大类证据因子;其次,分析各证据因子与铜矿点之间的空间关系,认为姑山旋回、娘娘山旋回火山机构控制了本区火山岩型铜矿的空间分布,根据计算结果,选取与火山岩型铜矿密切相关的龙王山组、姑山组地层,姑山旋回粗面斑岩、娘娘山旋回二长斑岩、NW向构造1.5 km缓冲区、NE向构造1.3 km缓冲区、EW向构造4.5 km缓冲区、硅化、褐铁矿化、黄铜矿化等作为模型自变量;最后采用加权Logistic回归模型进行成矿概率计算,并结合成矿地质背景,圈定四个成矿远景区,分别为P1、P2、P3、P4,其中P1、P2、P3呈北东向展布,主要受娘娘山和姑山火山机构控制,P4为东西向分布,主要受龙王山火山机构控制,在这些预测区中,均存在已发现的铜矿体,说明预测可信度较高。     

长春地区粉质黏土导热系数与其物性参数相关性研究

为研究导热系数与影响因素之间的相关关系,建立导热系数的推算公式,以长春地区粉质黏土为研究对象,对原状土样的导热系数与其物理参数之间的相关性进行回归分析。制作9个重塑土样,测其相关的参数值,以验证回归方程的适用性。结果表明,回归分析建立导热系数与2个物理参数之间的关系式不成立;考虑天然密度、含水率和孔隙度为自变量,其分别对应的相关性系数T检验显著值(Sig)都0. 05,复决定系数为0. 886,建立的回归方程成立,自变量能准确解释因变量的变化,且含水率与导热系数呈负相关,天然密度和孔隙度呈正相关。重塑土样相关参数代入回归方程得到的导热系数值与实验实测值之间相对误差低于4%,验证了该回归方程的普遍性和适用性。     

逻辑回归模型在边坡稳定性分析中的应用

采用逻辑回归模型和确定性系数CF建立概率模型,使用收集到的22个边坡实例,将重度、内聚力、坡角、坡高、内摩擦角和孔隙压力比六个影响因子作为自变量,边坡状态作为因变量,应用SPSS取不同的滑坡概率P与不滑坡概率1-P的比值的自然对数Z进行回归分析。将获得的滑坡概率拟合模型反馈到样本参数上,对比其误差,选取Z=±10时的函数模型对另外8组边坡进行验证,证明预测模型的合理性。     

Application of Geochemistry and VNIR Spectroscopy in Mapping Heavy Metal Pollution of Stream Sediments in the Takab Mining Area, NW of Iran

This study considered the possibility of using visible and near infrared (VNIR) spectral absorption feature parameters (SAFPs) in predicting the concentration and mapping the distribution of heavy metals in sediments of the Takab area. In total, 60 sediment samples were collected along main streams draining from the mining districts and tailing sites, in order to measure the concentration of As, Co, V, Cu, Cr, Ni, Hg, Ti, Pb and Zn and the reflectance spectra (350–2500 nm). The quantitative relationship between SAFPs (Depth500nm, R610/500nm, R1344/778nm, Area500nm, Depth2200nm, Area2200nm, Asym2200nm) and geochemical data were assessed using stepwise multiple linear regression (SMLR) and enter multiple linear regression (EMLR) methods. The results showed a strong negative correlation between Ni and Cr with Area2200nm, a significant positive correlation between As and Asym2200nm, Ni and Co with Depth2200nm, as well as Co, V and total values with Depth500nm. The EMLR method eventuated in a significant prediction result for Ni, Cr, Co and As concentrations based on spectral parameters, whereas the prediction for Zn, V and total value was relatively weak. The spatial distribution pattern of geochemical data showed that mining activities, along with the natural weathering of base metal occurrences and rock units, has caused high concentrations of heavy metals in sediments of the Sarough River tributaries.     

Dynamic land use change simulation using cellular automata with spatially nonstationary transition rules

The dynamic relationships between land use change and its driving forces vary spatially and can be identified by geographically weighted regression (GWR). We present a novel cellular automata (GWR-CA) model that incorporates GWR-derived spatially varying relationships to simulate land use change. Our GWR-CA model is characterized by spatially nonstationary transition rules that fully address local interactions in land use change. More importantly, each driving factor in our GWR model contains effects that both promote and resist land use change. We applied GWR-CA to simulate rapid land use change in Suzhou City on the Yangtze River Delta from 2000 to 2015. The GWR coefficients were visualized to highlight their spatial patterns and local variation, which are closely associated with their effects on land use change. The transition rules indicate low land conversion potential in the city’s center and outer suburbs, but higher land conversion potential in the inner near suburbs along the belt expressway. Residual statistics show that GWR fits the input data better than logistic regression (LR). Compared with an LR-based CA model, GWR-CA improves overall accuracy by 4.1% and captures 5.5% more urban growth, suggesting that GWR-CA may be superior in modeling land use change. Our results demonstrate that the GWR-CA model is effective in capturing spatially varying land transition rules to produce more realistic results, and is suitable for simulating land use change and urban expansion in rapidly urbanizing regions.     

基于逻辑回归的四川青川县区域滑坡灾害预警模型

四川省青川县滑坡灾害群发,点多面广,区域滑坡灾害预警是有效防灾减灾的重要手段,预警模型是成功预警的核心。由于研究区滑坡诱发机理复杂、调查监测大数据及分析方法不足等原因,传统区域地质灾害预警模型存在预警精度有限、精细化不足等问题。文章在青川县地质灾害调查监测和降水监测成果集成整理与数据清洗基础上,构建了青川县区域滑坡灾害训练样本集,样本集包括地质环境、降雨等27个输入特征属性和1个输出特征属性,涵盖了青川县近9年（2010—2018年）全部样本,数量达1 826个（其中,正样本613个,负样本1 213个）。基于逻辑回归算法,对样本集进行5折交叉验证学习训练,采用贝叶斯优化算法进行模型优化,采用精确度、ROC曲线和AUC值等指标校验模型准确度和模型泛化能力。其中,ROC曲线也称为“受试者工作特征”曲线;AUC值表示ROC曲线下的面积。校验结果显示,基于逻辑回归算法的模型训练结果准确率和泛化能力均较好（准确率94.3%,AUC为0.980）。开展区域滑坡实际预警时,按训练样本特征属性格式,输入研究区各预警单元27个特征属性,调用预先学习训练好的模型,输出滑坡灾害发生概率,根据输出概率分段确定滑坡灾害预警等级。当输出概率P≥40%且P<60%时,发布黄色预警;当输出概率P≥60%且P<80%时,发布橙色预警;当输出概率P≥80%时,发布红色预警。