摘要
建立地球全空间数据库需要研究空间数据融合的方法和智能技术.本文以布格重力异常数据的融合为例,介绍一种基于奇异值分解的二维空间数据融合方法.此方法在原理上可以实现融合后的数据集误差模不大于融合前的子数据集的最小误差模,相较于传统的融合处理方法原理更加先进.在算法中应用了权重矩阵,可使融合之后数据区域与原始数据的边界上梯度变化较小;运用有限差分法计算重力水平梯度,检验数据融合边界的效果.理论模型和实际数据试验表明,此方法可以保证数据隐含的重要属性完整保存,具有计算过程简单、适用范围广、融合效果好等特点,可以应用于位场等地球物理数据融合,也可以用于遥感和其他地理信息的数据融合.
In order to build the global space database,we need to integrate the local area data by the method of spatial data fusion,and study the corresponding intelligent technology.We used a twodimensional spatial data fusion method based on singular value decomposition with taking the fusion of Bouguer gravity anomaly data as an example.This method can make the error mode of the merged data set not larger than the minimum error mode of the subdata set before the fusion theoretically.Meanwhile,the weight coefficient matrix was used in the algorithm,which could make the gradient change on the boundary between the overlapped area and the original data area small after fusion;the finite difference method was used to calculate the horizontal gradient of gravity to test the effect of the data fusion boundary.In the theoretical model and the actual data test,we selected the number of singular values whose contribution rate of singular value was more than 99%,and eliminated the small singular value items,so as to eliminate the error interference caused by the small singular value.The three matrices after SVD were weighted average,and the weight came from the weight coefficient matrix,which could ensure the smooth boundary of the fusion region and reduce the artificial phenomenon.When the spatial resolution of the initial local data was quite different,wavelet multi-scale decomposition could be used to remove the high-resolution layer and then,fused the remaining anomalies.The result was more consistent on the spatial scale and more perfect.Results indicate that this method can ensure the integrity of the important attributes implied in the data,and has the characteristics of simple calculation process,wide range of application,good fusion effect,and can be applied to geophysical data fusion such as potential field,as well as remote sensing and other geographic information data fusion.
作者
陈勐韬
杨文采
颜萍
施慧丽
CHAN Meng-tao;YANG Wen-cai;YAN Ping;SHI Hui-li(School of Earth Sciences,Zhejiang University,Hangzhou 310027,China)
出处
《地球物理学进展》
CSCD
北大核心
2020年第3期940-949,共10页
Progress in Geophysics
基金
国家自然科学基金资助项目(41574111)资助.
关键词
数据融合
二维空间数据
布格重力
奇异值分解
权重矩阵
有限差分法
Data Fusion
Two-dimensional spatial data
Bouguer gravity
Singular value decomposition
Weight matrix
Finite difference method
