SVM 方法在北京地区夏季晴雨预报中的初步应用

以常规观测降水资料、T639 模式输出产品为基础,利用支持向量机（SVM）方法分别建立了北京地区延庆、密云、房山和观象台4 个站点的夏季晴雨预报模型,并用独立的样本对预报模型进行了业务试运行检验。检验结果表明,经过SVM 方法学习后建立的站点晴雨预报效果优于T639 模式直接输出的晴雨预报效果,08—20 时段和20—08 时段4 站晴雨预报平均准确率分别提高了20.06%和9.45%。SVM 方法对T639 的晴雨预报效果有显著的改进,对北京地区夏季晴雨预报有一定的客观参考价值。     

时空Kriging算法在区域地磁场插值中的应用及改进

本文以区域地磁场为应用背景,研究了时间域信息对地磁场插值效果的影响.用时空Kriging算法进行插值时,在时空Kriging变差函数的拟合过程中,将角度信息引入向量距离,定义了新的向量距离形式,提出了改进的时空Kriging算法,并对权重的选取进行讨论.对经度87.2°E—126.6°E,纬度19.0°N—49.6°N范围内的32个台站时均值地磁场数据进行插值实验.在添加了时间信息之后,插值结果在兼顾经度方向和纬度方向变化的同时,边值问题也得以缓解,明显优于传统方法.交叉验证结果表明,随着时间域信息和向量角度的加入,Kriging算法的性能得到改进,地磁场的插值精度依次提高.对比实验中,改进Kriging算法的精度最高、性能最稳定,MAE集中在1 nT左右,MSE集中在[0,5]的范围内.     

求解弹性波解耦方程的一种优化拟解析方法

常规伪谱方法二阶时间差分格式时间精度较低,且对于大步长时间采样间隔,常规伪谱方法不稳定.拟解析方法对于速度变化剧烈的模型,在时间和空间上均有较大误差.本文提出了一种基于解耦的二阶位移弹性波方程波场模拟及矢量波场分解的优化拟解析方法,将归一化的拟拉普拉斯算子分别应用于P波和S波波场延拓,延拓矢量波场的同时,可分解并延拓纯纵波和纯横波波场.利用弹性波优化拟微分算子表示拟拉普拉斯算子,该拟微分算子不仅包括原始微分算子的谱估计,而且还包含一个时间补偿项,其可在波数-空间域精确地补偿波动方程在时间方向上采用二阶有限差分引起的误差.利用低秩分解近似求解弹性波优化拟微分算子,可有效提高计算效率.2D均匀模型、层状模型以及部分盐丘模型数值正演模拟结果表明：相比较于常规的伪谱法和拟解析法,本文方法在时间和空间上均有很高的精度,并且稳定性条件比较宽松.

     

一种适用于长弧段的初轨计算方法

根据单位矢量法测轨原理，在人造卫星初轨计算的单位矢量法基础上，给出了一种适用于长弧段的初轨计算方法．该方法既适用于长弧段，也适用于短弧段；适用于各种不同类型的观测资料和任意偏心率、任意轨道倾角的人造地球卫星；有利于提高初轨测定精度；并改善整个计算收敛性．特别需要指出的是，该方法与有摄初轨计算的单位矢量法相结合，为单位矢量法从初轨计算推广到轨道改进打下了坚实的基础．     

总体最小二乘法在初轨计算中的应用

在分析了UVM2方法条件方程特点的基础上,将总体最小二乘法(TLS)引入轨道计算中,避免了原算法中对活力公式的线性化处理．仿真和实测计算结果表明,将TLS应用于UVM2是有效的．     

基于SVM的含约束地球化学单元素异常圈定方法及应用

地球化学元素异常的空间分布受限于地质背景、矿产空间分布和景观特征等因素,单元素异常图的绘制不应仅依据采样数据由计算机追踪等值线图的方法自动勾绘。传统的手工圈定方法效率低、受人为因素影响大。基于SVM(Support Vector Machine,支持向量机)提出了一种可以考虑地质约束的单元素异常圈定方法,并加入地质单元分区、地质构造等约束条件。该方法可以高效绘制出单元素异常图,并成功应用于2个实例中。     

基于SVM的资源三号测绘卫星影像分类

以江苏省宜兴市为研究区,利用支持向量机(SVM)方法对资源三号测绘卫星影像进行了分类,其总分类精度为97.76%,Kappa精度为0.968 7。为了评价算法的适用性,同时应用最大似然法与最小距离法对同一影像进行分类测试,支持向量机分类法精度高于其他2种方法,可以满足土地覆盖分类调查需求。     

A Review of: Environmental Modelling with GIS.

Cartographic modelling operations provide powerful tools for analysing and manipulating geographic data in the raster data model. This research extends these operations to the vector data model. It first discusses how the spatial scopes of analysis can be defined for point, line, and polygon features analogous to the raster cell. Then it introduces the local, focal, and zonal operations available for vector features, followed by providing a prototype syntax that might guide the implementation of these operations. Through example applications, this research also demonstrates the usefulness of these operations by comparing them with traditional vector spatial analysis.     

A proposed framework for feature‐level geospatial data sharing: a case study for transportation network data

Current data sharing in the Internet environment is supported using metadata at the file level. This approach has three fundamental shortcomings. First, sharing data from different sources with different semantics, data models, and acquisition methods usually requires data conversion and/or integration like data conflation. This can be tedious and error‐prone. Second, data updated from one source cannot be automatically propagated to other related data or applications. Finally, data sharing at the file level makes it difficult to provide feature‐level data for searching, accessing, and exchanging in real time over the Internet. This paper addresses these three issues by proposing a standards‐based framework for sharing geospatial data in the transportation application domain. The proposed framework uses a standard data model—geospatial data model proposed by the Geospatial One‐Stop initiative to harmonize the semantics and data models without the use of data integration methods. It uses Geography Markup Language (GML) for geospatial data coding and feature relationship, which provides a basis to propagate the data update from one source to related other sources and applications, and to search and extract data at the feature level. The framework uses the Web Feature Service (WFS) to search, access and extract data at the feature level from distributed sources. Finally, the Scalable Vector Graphics (SVG) standard was used for data display on the Web browser. Two transportation network datasets are used in the prototype case study to implement the proposed framework. The prototype allows the user to access and extract data at the feature level on the Web from distributed sources without downloading the full data file. It shows that the proposed standards‐based feature‐level data‐sharing system is capable of sharing data without data conflation, accessing, and exchanging data in real time at the feature level. The prototype also shows that changes in one database can be automatically reflected or propagated in another related database without data downloading.     

Application of support vector machine and relevance vector machine to determine evaporative losses in reservoirs

This article employs Support Vector Machine (SVM) and Relevance Vector Machine (RVM) for prediction of Evaporation Losses (E) in reservoirs. SVM that is firmly based on the theory of statistical learning theory, uses regression technique by introducing ε‐insensitive loss function has been adopted. RVM is based on a Bayesian formulation of a linear model with an appropriate prior that results in a sparse representation. The input of SVM and RVM models are mean air temperature (T) ( °C), average wind speed (WS) (m/sec), sunshine hours (SH)(hrs/day), and mean relative humidity (RH) (%). Equations have been also developed for prediction of E. The developed RVM model gives variance of the predicted E. A comparative study has also been presented between SVM, RVM and ANN models. The results indicate that the developed SVM and RVM can be used as a practical tool for prediction of E. Copyright © 2011 John Wiley & Sons, Ltd.