【作者】 王伟；

【导师】 王琪；

【作者基本信息】 中国地震局地球物理研究所 ， 固体地球物理学， 2013， 博士

【摘要】 中国大陆位于欧亚板块东南部,受印度板块、菲律宾板块和西伯利亚一蒙古亚板块的夹持,是全球岩石圈变形和地壳运动最强烈的地区之一,也是全球大陆地震多发区。因而研究我国大陆内部的变形对于认识长期的岩石圈动力学过程与变形模式,了解断层的作用机制以及进行地震危险性评估都有十分重要的意义。我国于20世纪80年代末开始用GPS监测地壳形变,之后的20多年,先后在青藏高原、川滇、天山等地区陆续开展了流动GPS观测,获取了大量的GPS测站资料。随着中国地壳运动观测网络(简称网络工程)和中国大陆构造环境监测网络(简称陆态网络)的实施和运行,我国的地壳运动监测能力和相关研究取得了长足进步。近年来,随着GPS处理软件的模型精度和参考框架精度的显著提高,较以往基于较短时期内解算的速度场结果,新的参考框架、软件处理模型和观测时间更长的GPS数据资料为我们获取更为精确的速度场结果提供了可靠保障。本文工作以网络工程的GPS资料为主,辅以陆态网络部分站点的GPS区域站资料和国内外其它研究小组公开发表的速度场结果,对中国大陆的地壳运动和变形场进行研究。在对GPS数据进行处理时,与以往采用GIPSY或GAMIT软件不同的是,本文的GPS数据处理采用的是BERNESE软件,并使用最新的软件模型进行统一处理。具体研究内容和结果包括以下3个方面：1)使用最大似然估计(MLE)方法对网络工程1999-2010年的GPS基准站位置时间序列进行了噪声分析,研究了基准站GPS时间序列的时间和空间相关性,定量评估了基准站速率的可靠精度。此外除周年、半周年信号以外,还评估了其他异常周期信号对噪声分析结果的影响,主要内容和结论有：对网络工程基准站位置时间序列空间滤波前、滤波后以及共模信号进行了噪声分析。在空间滤波时使用了改进的加权叠加滤波方法进行空间滤波以去除各测站位置时间序列之间的空间相关性,并且主要考虑了1000km以内的测站残差时间序列来形成共模误差。通过MLE分析发现共模误差主要由白噪声和闪烁噪声组成,其中白噪声在北向、东向和垂向幅度的均值分别为0.65、0.72、1.39mm,闪烁噪声在北向、东向和垂直方向的幅度均值为2.86、2.98、10.11mm/a1/4。此外,在“白噪声+闪烁噪声”的组合噪声模型假设下,空间滤波前、后测站速率之差在北向、东向和垂向的均值(及其相应的标准偏差)分别为0.04±0.09,0.06±0.15,-0.06±0.34mm/a。说明在空间滤波前、后,测站速率的结果具有很好的一致性。通过MLE方法比较、分析各类不同噪声模型,滤波前GPS位置时间序列有色噪声的主要成分是闪烁噪声,而滤波后时间序列有色噪声的主要成分为幂指数噪声。总体来说,对于滤波前的原始时间序列,纯白噪声模型得到的速率误差与“白噪声+闪烁噪声”或“白噪声+幂指数噪声”相比,可能会被低估8-10倍；对滤波后的时间序列,纯白噪声模型得到的速率误差与“白噪声+闪烁噪声”或“白噪声+幂指数噪声”相比,可能会被低估8-16倍。当使用“白噪声+闪烁噪声”的模型时,滤波前GPS位置时间序列在北向、东向和垂直方向速率误差中值分别为0.16、0.17、0.58mm/a,滤波后时间序列在北向、东向和垂直方向速率误差的中值分别为0.09、0.10、0.40mm/a。2)采用网络工程1999-2011年6期区域网GPS数据和陆态网络2009-2011年部分测站的2期GPS数据构建中国大陆地壳运动的水平速度场；利用网络工程近10年区域站GPS资料构建中国大陆长期的垂直运动速度场；综合其它区域已公开发表速度场结果,获取了整个中国大陆及周边地区更为完整、空间分辨率更高的现今地壳运动速度场。本文垂直速度场结果表明：中国大陆大部分地区的地壳垂直运动速率较低,其中青藏高原地区地壳垂直运动趋势整体表现为较低速率的隆升。青藏高原东北部的祁连山-阿拉善地区测站的垂直速率均值约为1.0mm/a；藏南拉萨地块内部测站垂向速率均值约为1.6mm/a；巴颜喀拉地块东部、龙门山以西的川西地区测站垂直速率均值约为1.9mm/a；天山块体的内部整体上表现为低速的隆升,该地区测站的垂直运动速率约为1.8mm/a;华南地区以及塔里木盆地内部的地壳垂直运动十分微弱,测站垂直运动速率小于1mm/a；在华北平原,有部分地区(如河北、山东、江苏、天津等地)由于长期开采地下水,一些测站的垂直运动表现为十分显著的沉降,个别测站下沉速率达到6-10cm/a东北地区的北部地壳垂直速率相对较高,部分测站达到3-5mm/a。3)利用本文处理得到的中国大陆水平运动速度场结果,精化中国大陆活动地块的运动学模型,在顾及地块旋转和块体边界断层滑移所产生的弹性应变条件下,整体性反演实测GPS速度场,获取了各活动地块与边界断裂带的运动速率。在GPS观测精度内,该模型整体上较好地反映了中国大陆构造变形的主要特征,模型给出的主要断层滑动速率在中国西部的青藏高原及周边为5-20mm/a,在中国东部地区为1-2mm/a。例如：华南块体的运动速率为6.4mm/a,华北平原运动速率为4.1mm/a,准噶尔块体为4.5mm/a,塔里木块体的运动速率为13.1mm/a。阿尔金断裂带中东段的左旋走滑速率约为6mm/a,海原断裂带左旋走滑速率约为6mm/a,东昆仑断裂带左旋走滑速率约为8-10mm/a,甘孜-玉树-鲜水河断裂带的左旋走滑速率约为12-15mm/a,喀啦昆仑断裂带右旋走滑速率约为7mm/a,红河断裂带右旋走滑速率约为8-12mm/a。更多还原

【Abstract】 The Chinese mainland, which is located in the southeastern part of Eurasia plate, is one of the most actively deoforming areas with significant crustal movement and widespread earthquakes, due to its contact with India plate, Philippine plate and Siberia-Mongolia subplate. Therefore, the study of the deformation of Chinese mainland is critical to understand long-term crustal deformation and dynamic processes, and has important implications for faulting mechanism and earthquake hazard assessment.In the late1980s, China began to utilize GPS measurements in order to monitor the crustal deformation. Over the next20years, several campaign-mode GPS surveys were carried out in the Tibet plateau, Sichuan-Yunan and Tienshan, etc., and abundant GPS data had been obtained since then. With the implementation and operation of "Crustal Movement Observation Network of China "(CMONOC) and " Continental Tectonic Environment Monitoring Network of China "(CTEMNC), the capability of monitoring the crustal movement and its related research has made considerable progresses in China.Nowadays, owing to the improvements for the models related to the GPS data processing and the precisions of the GPS reference frame, the new software models, new reference frame and data with much longer time span provide us the oppotunity to obtain more precise velocity field. This work utilize the GPS data mainly from CMONOC, combined with the data from CTEMNC and the published velocity field results from other research groups, in order to study the crustal movement and active deformation in the Chinese mainland. The GPS data in this study are processed homogeneously using the BERNESE GPS software, which is deferent from previously used GAMIT or GIPSY softwares. The main content and conclusions are as follows:1) Decade-long coordinate time series of26continuous GPS sites of CMONOC are analyzed for their noise content using maximum likelihood estimation (MLE). The noise properties of continuous GPS time series of CMONOC are studied for both the temporally and spatially correlated characteristics, and the velocity uncertainties of the continuous GPS sites are quantified. Besides, in addition to the annual and semiannual signals, the influence of other periodic signals on the noise assessment is also analyzed.The main conclusions are:This work study the noise properties of continuous GPS time series of CMONOC for the unfiltered, filtered solutions and also the common mode signals in terms of power law plus white noise model. In the spatial filtering, a common mode error was removed for every time series from a modified stacking of position residuals from other sites within～1000km of the selected site. The common mode signal in our network has a combination of spatially correlated flicker noise and a common white noise with large spatial extent. The common mode signal removed in the spatial filtering has a mean magnitude of0.65(north),0.72(east),1.39(vertical) mm of white noise and a mean magnitude of2.86(north),2.98(east),10.11(vertical) mm/a1/4of flicker noise, respectively. In addition, the mean of the velocity differences (and their corresponding standard deviations) obtained from the white noise plus flicker noise model between unfiltered and filtered solutions is0.04±0.09,0.06±0.15,-0.06±0.34mm/a, for north, east and vertical components, respectively, indicating an overall agreement between the velocity estimates of the unfiltered and filtered coordinate time series.Through the comparison of different noise models, we can draw that for the unfiltered solutions the main coloured noise is a flicker process; whereas for the filtered solutions the main coloured noise is a general power law noise process with any estimated spectral index. On the whole, for the unfiltered solutions, when a pure white noise model is assumed, the velocity error in CMONOC continuous GPS position time series may be underestimated by factors of8-10compared with that of the white plus flicker noise or white plus power law noise model. For the filtered solutions, when a pure white noise model is assumed, the velocity error may be underestimated by factors of8-16compared with that of the white plus flicker noise or white plus power law noise model. In addition, when using a white plus flicker noise model, the median values of the velocity errors for the unfiltered solutions are0.16,0.17and0.58mm/a for the north, east and vertical components, respectively; and the median values for the filtered solutions are0.09,0.10and0.40mm/a for the north, east and vertical components, respectively.2) The horizontal velocity field was derived using the campaign-mode GPS data from CMONOC surveyed in1999-2011and CTEMNC surveyed in2009-2011. The vertical velocity field was obtained using the campaign-mode GPS data from CMONOC surveyed in1999-2011. Besides, combined with other published results, a more complete and high resolution horizontal velocity field covering the whole Chinese mainland and its surrounding areas has also been given.Preliminary results from vertical velocity field demonstrate that most of the Chinese mainland has a low-speed vertical motions and the general trend of Tibetan Plateau is toward uplift with low rates.Specifically speaking, the GPS sites in the Qilian-Alashan of northeastern Tibet have a mean vertical rate of1.0mm/a; the GPS sites in the Himalaya and Lhasa blocks have a mean vertical rate of1.6mm/a; the sites in the western Sichuan have a mean vertical rate of1.9mm/a; the sites within the Tienshan have a low uplift trend with a mean rate of1.8mm/a; the sites within the South China and Tarim blocks have insignificant vertical rates of<1mm/a; in the North China plain, many sites show significant subsidence due to the effect of groundwater pumping, and several sites even have subsidence rates of6-10cm/a; the sites in the northeastern China have a relatively high vertical uplift rates of3-5mm/a.3) Taking advantage of the velocity field obtained above, this work refine the active block model for the Chinese mainland. To interpret the GPS velocity field, the GPS velocities are inverted simultaneously for the effects of block rotations, elastic strain induced by the block-bounding faults and the block movement and fault slip rates are calculated from this block model. Within the precision of GPS measurements, this block model provides a first-order description of the active tectonics of Chinese mainland.The slip rates inferred from this model is in a range of5-20mm/a for the major faults in Tibet and its margins and1-2mm/a in eastern China. For instance, the velocities of South China, North China plain, Dzunggar, Tarim blocks are～6.4mm/a,～4.1mm/a,～4.5mm/a and～13.1mm/a, respectively.The left-lateral slip rate for Altyn Tagh is～6mm/a; the left-lateral slip rate for Kunlun is about8-10mm/a; the left-lateral slip rate for Ganzi-Yushu-Xianshuihe is about12-15mm/a; the right-lateral slip rate for Karakorum is～7mm/a; the right-lateral slip rate for red revier is about8-12mm/a, etc.更多还原