基于K-Means++的省内子块体划分及中国大陆水平相对运动速度场模型的建立与分析

The sub-block demarcation with K-Means++in each province′s interior and establishment analysis of the relative horizontal velocity field model in China's Mainland

省市级区域CORS(Continuously Operating Reference Stations)系统是当代城市数字化、信息化和智慧化的重要组成部分,便于获取各类物体的时空信息及其相关动态变化.为了进一步实现区域框架基准的现代化与自主化,全面提升现代测绘基准综合服务水平和应急保障能力,同时为了提高中国大陆区域水平速度场的精度,并精细地刻画其自身的局部运动特征,本文利用陆态网上千站2011-2017年的连续观测数据,采用GAMIT/GLOBK软件,获得高精度的定位和速度成果,进而提出和构建了基于欧拉矢量模型的中国大陆省级块体相对运动模型和部分省内子块体相对运动模型,并与欧亚板块、大陆整体和二级板块相对运动模型进行对比分析;结果表明,欧亚板块相对运动模型仅能描述大陆的部分运动趋势,中国大陆整体板块相对运动模型能够较好地展现大陆整体运动趋势,二级板块和省级块体相对运动模型均能够较为精细地反映区域的局部运动特征,且两者水平相对速度的内符合精度均小于2mm·a^-1,其外符合精度均小于3mm·a^-1,其中前者物理意义更为明显,后者使用更为简便,但在青藏、川滇等地壳运动复杂的地区两者精度仍有欠缺.因此,本文提出利用K-Means++算法对地壳运动复杂区域的水平速度场进行聚类分析,以快速准确地对这些区域进行子块体划分;结果表明,划分成果与现有部分二级块体成果相符合.为了兼顾省内复杂地质构造与地形地貌的影响,同时提高省级块体划分的物理意义,对地壳运动复杂的省份再细分块体,进而对各子块体构建欧拉矢量模型;结果发现,该模型平均误差和中误差均小于2mm·a^-1,提高新疆、西藏、川滇等地区的速度场模型精度至2mm·a^-1左右,在确保精确度的同时,满足使用简便性的要求.

Provincial CORS system is an important part of modern city digitization,informatization and intellectualization,which is easy to obtain the spatiotemporal information of various objects and their related dynamic changes.In order to further realize the modernization and autonomy of the regional framework benchmark,improve the comprehensive service level and emergency support capacity of the modern surveying benchmark,upgrade the accuracy of the regional velocity field and accurately depict its own local motion characteristics in the mainland of China,the velocity field is derived from continuous observation data of the Crustal Movement Observation Network of China(CMONOC)that was surveyed in 2011-2017.Both high precise coordinate and velocity of these CMONOC stations have been calculated with GAMIT/GLOBK software.In addition,the Euler vector of the whole Chinese continent and two-stage plates in the China mainland are reanalyzed based on the above velocity field.On the basis of the Euler vectors of the above-mentioned,i.e.the whole Chinese continent,two-stage plates,provincial plates,sub-blocks in some provinces′interior,as well as that of Eurasia plate derived from NNR-NUVEL1A plate motion model,the contribution of the average movement of each block to the horizontal velocity field could be available and then the residual velocity fields in China's Mainland are further analyzed and compared.The results show that the NNR-NUVEL1A model only deducts the partial movement trend of the velocity field in China's Mainland.In contrast,the overall movement model of China's Mainland plate can better reflect the overall movement trend.The movement model of two-stage plates and provincial plates in Chinese mainland,both of whose inner and outer average precision is less than 2 mm·a^-1 and 3 mm·a^-1,respectively,can both depict the local movement characteristics of China's Mainland more precisely,and the only difference between them is that the former is easier to understand in the sense of physics while the latter is simpler in using.However,both of them is not perfect for researching the relative motion of some regions such as Xinjiang,Tibetan Plateau,Sichuan,Yunnan where exist complex crustal movement.Hence,through cluster analysis with K-Means++on horizontal velocity field in these complex areas,these provincial blocks are further decomposed into some sub-blocks quickly and accurately.Our results are in accordance with those of present-day blocks of second order.Consideration to the impact of complex geological structure,topography and geomorphology in these provinces and the physical significance in respect to each corresponding provincial plate,we also give the Euler vectors of these sub-blocks in some provinces′interior.The result can be concluded that not only the average error and mean square error are less than 2 mm·a^-1 in China's Mainland but also the precision of horizontal velocity fields can reach 2 mm·a^-1 in some complex regions,that is Xinjiang,Tibet,Sichuan,Yunnan.Obviously,the scheme of the sub-block demarcation in some provinces′interior by the method of K-Means++is effective and feasible in terms of high precision and convenience.