Determination of reservoir induced earthquake using support vector machine and gaussian process regression

The prediction of magnitude （M） of reservoir induced earthquake is an important task in earthquake engineering. In this article, we employ a Support Vector Machine （SVM） and Gaussian Process Regression （GPR） for prediction of reservoir induced earthquake M based on reservoir parameters. Comprehensive parameter （E） and maximum reservoir depth] （H） are considered as inputs to the SVM and GPR. We give an equation for determination oil reservoir induced earthquake M. The developed SVM and GPR have been compared with] the Artificial Neural Network （ANN） method. The results show that the developed SVM and] GPR are efficient tools for prediction of reservoir induced earthquake M. /

Induced Earthquake Monitoring of the Three Gorges Reservoir

The Three Gorges Project(TGP) on the Yangtze River(YR) is the largest water power station in the world;it is now attracting the worldwide attention. Possessing comprehensive utilization benefits mainly for flood control,power generation and navigation improvement,TGP will be a vital and backbone project in harnessing and developing of the YR. TGP has been commenced in 1993,the main channel of Yangtze was diverted on Nov.8th,1997 and whole project was completed in 2009. The Three Gorges(TG) reservoir has been formed since June 2003 and impounded to 175 m on Oct.26th,2010. The reservoir has a total length of 660 km and an average width of 1.0-1.5 km,with a total waters area of 1048 km2. In pace with the raised pool level,the enlargedstorage,the probability of reservoir earthquake would be enlarged. According to the magnitude records of historical maximum earthquake and properly weighted,defined the maximum reliable earthquake of TG reservoir would be ML6 magnitude,existed the probability of ML 5.0-5.8 induced earthquake around the Xiannvshan and Jiuwanxi fault zone(18 km upstream of the dam) . However,the highest intensity around the dam site would not exceed Degree VI,whereas the main structures of TGP were designed with an anti-seismic intensity of Degree VII,hence the reservoir induced earthquake would not form direct threat to the main structures of TGP. After the impoundment and operation of TG reservoir,the frequency and intensity of earthquake has been increased in some extent,yet the earthquake activities at TG region still kept in an original weak earthquake activities status. We should strengthen the TG reservoir induced earthquake monitoring and forecast,prevent the earthquake and geological calamity,ensure the safety of dam construction and operation,form the harmonious society as well as guarantee the long term security. This paper explained the mechanism,kinds,causes,development and calamities of the earthquakes in the world,collected a lot of TGP,China and World's earthquake and geology materials. Such as 105 year(Jan.1906-June2011) earthquake records in the world(Tab.1) ,13 main reservoir induced earthquake records in China(Tab. 2) ,19 main reservoir induced earthquake records in the world(Tab.3) ,14 year(1996-2009) reservoir earthquake monitoring records in TGP(Tab.4) . Analyzed the cause,calamities of the main earthquakes as well as its alarm and forecast. May it provide reference for readers and beg your guide.

Tectonic environment and cause of earthquakes in the Three Gorges reservoir area

Seismotectonics in the Three Gorges reservoir area is investigated by using the P-wave tomography with earthquakes that occurred before the impoundment of the reservoir. The result indicates that most of these events occurred in or around the velocity-gradient belts between high-velocity and low-velocity anomalies. These belts have similar characteristics to buried-fauh zones. Stresses generated by movement of partially molten upper-mantle materials and thermal stress may have jointly contributed to the seismic activities along the faults and such buried faults, and possibly activated new earthquake ruptures.

Three-Gorges Reservoir Area on the Yangtze River Faced with Risk of Induced Destructive Earthquakes

The Three-Gorge Water Conservancy Project on the Yangtze River started to impound water by locking gates on June 1, 2003. A swarm of more than 2000 small earthquakes suddenly occurred densely along the river section in Xinling Town north of Badong, Hubei Province, on June 7. This caused grave concern in the following years, but, with the completion of the second- and third-phase engineering construction, the water level in the reservoir will rise up to 156 m and 175 m respectively, no matter whether it can induce larger and stronger earthquakes. After an analysis of the distribution of active faults in the eastern part of the reservoir area, their intersections and capability to generate earthquakes from a seismo-tectonic viewpoint, we consider that after the reservoir impounding, two potential focal zones existing in Badong and Zigui counties may generate M 5.5 earthquake, the seismic intensity of which may reach Ⅷ. It will induce landslide bodies in the reservoir area to be reactivated and slide. The evidence is the large-scale landslide occurring on the Shazhenxi Creek river in Zigui County on June 12, 2003.

Study on Precise Location and Structure of Earthquakes in the Shanxi Reservoir,Zhejiang Province

The Shanxi reservoir earthquakes are significant seismic events in southern Zhejiang Province in recent years, an area with fewer and weaker earthquakes. The seismicity showed an intermittent characteristic and group distribution. The epicenters located by the seismic network did not show a predominant direction and the seismogenic structure is not clear. In the study, the nonlinear imaginary wave travel time equation was linearized and solved, and the source position, initial imaginary velocity and travel time residuals were obtained. Then by doubling the standard deviation as time residuals, the maximum error generated from longitude, latitude, depth and imaginary velocity was calculated. The genetic population was structured using the maximum error and the end result of earthquake location was obtained by genetic algorithm. The result of relocation of the Shanxi reservoir earthquakes with this method shows that earthquakes are largely concentrated on a near-vertical, northwest oriented fault plane, and the included angles between the normal of the plane and the due north, due east and vertical directions are 46~, 44~, and 87~, respectively. The result is in agreement with that of comprehensive fault plane solutions of small earthquakes. The average depth of the earthquakes was 4.7km, the maximum depth 9.5kin, and the minimum depth 1.7km. The epicenters showed a northwestward narrow banded distribution, and the focal depth increased along the northwest direction. There was a discontinuous seismic gap of about 3.5km long at the northwest end of the strip. The characteristics of source parameters obtained by using the Borun model were not significantly different from that of tectonic earthquakes. Seismic stress drop was about 0.33MPa, and the average stress drop was 0.88MPa. According to the stress drop＇ spatial distribution, the seismic discontinuities segment at the northwest end of the strip is in a low stress drop zone.

Discussion on 'Was the 2008 Wenchuan earthquake triggered by Zipingpu Reservoir?'

Some recent publications presented a result suggesting that Zipingpu reservoir hastened the occurrence of the 2008 Ms8.0 Wenchuan earthquake by tens to hundreds of years. Their researches calculated the Coulomb stress change induced by Zipingpu reservoir on the rupturing fault of Wenchuan earthquake. Their results, however, are critically dependent upon the 3-D event location, reservoir location, and the fault plane orientation. We repeated Ge et al.＇s work in this paper and found that an improper dip angle parameter of their 2-D fault model might lead to a wrong conclusion. Both the modeling results based on the 2-D model and 3-D model with proper fault parameters will show Coulomb stress changes alone were neither large enough nor had the correct orientation to affect the occurrence of Wenchuan earthquake, which supports our recent argument based on the local seismicity analysis and the induced Coulomb stress change calculation with a 3-D model.

Research and Practice on the Crustal Deformation Mobile Monitoring Network Layout in the Hydropower Station Reservoir Area

According to the construction project of the crustal deformation mobile monitoring network in the cascade hydropower stations built in the lower reaches of Jinsha River,this paper analyzes the design ideas and layout principles of crustal deformation mobile monitoring used in the monitoring of reservoir induced earthquakes. This paper introduces three types of monitoring networks used in the Xiluodu reservoir and Xiangjiaba reservoir, as well as the work already undertaken,in order to provide a kind of reference for the related engineering construction and comprehensive monitoring of reservoir induced earthquakes.

深部地热能系统主要挑战与耦合储能的增强型创新开发模式

中国在中低温地热能直接利用方面早已领先全球,但在深部地热能发电方面却发展缓慢。深部高温高压环境下岩石渗透性降低,深部地热能开采需要建立工程型地热系统(Engineered Geothermal System,EGS),通过水力压裂对储层进行改造,以获得具有较高渗透性的人工地热储层。由于目前常用的深部地热能储层改造技术主要借鉴油气增产领域的水力压裂工艺,在热储改造效果、地震风险控制、高效取热等方面受到限制。首先总结深部地热能水力压裂的特点为:裂缝破坏主要以剪切机理为主;冷水回灌引起的温差效应产生拉应力会促使裂缝向更远处扩展;持续的注水使注入井井筒压力高于地层压力,有助于保持裂缝处于张开状态。因此,EGS水力压裂不需要使用支撑剂,与依靠支撑剂的油气井增产压裂完全不同。同时,系统剖析EGS面临的发电产能低、注采连通差、诱发破坏性地震以及无补贴难盈利4大难题与挑战。从创新压裂和循环利用层面提出耦合储能的增强型创新开发模式(Regenerative Engineered Geothermal System,REGS),通过数值模拟研究REGS的优点。结果表明,采用水平井非等距、非等面积、非等注水量分段压裂,可以提高注水井与生产井的连通能力。通过优化压裂工艺,采用多段压裂模式,每次压裂初始期间快速提高注水速率,而后期缓慢降低注水速率,避免井筒压力的突然波动,可以达到控制诱发地震震级的目的,避免产生实质性伤害地震。结合可再生能源大规模地下存储,既能实现多能互补,又能提高REGS项目生产寿命和盈利能力。研究成果有助于为我国深部地热能热电联产和储能一体化技术的试点和标准化推广奠定基础。

构造型水库诱发地震震级预测与探讨

国内外大量水库发生过不同震级的水库诱发地震,给社会经济环境和生态环境等都造成了不同程度的损失,自20世纪70年代以来,国内学者从不同角度对水库诱发地震的诱发原因和预测方法展开了研究。在前人研究的基础上,对水库诱发地震基本特征和诱发主要因素进行了分析、研究,探索性提出了构造水库地震影响因素的隶属函数,该构造函数综合考虑了对水库规模、构造类型、构造活动性、地层岩性等因素对水库诱发地震的影响,并构建了新的水库诱发地震数学模型对国内外100多座水库诱发地震震例进行计算。经对比分析,发现采用本模型计算的水库诱发地震震级与各水库实际地震震级相接近,可见利用该数学模型预测水库诱发地震震级是基本可行的,希望能为构造型水库诱发地震预测与防治提供一点启示;此外,还发现地震动参数对水库诱发地震的震级贡献相对较大,但水库规模对震级贡献相对较小,特别是水深超过20 m时,水库规模导致的震级差异不明显。

基于PALM自动检测方法的三峡库区微震活动研究

选用三峡工程水库诱发地震监测系统在库首区布置的24个台站采集的2016~2018年数据,利用PALM微震检测定位方法对三峡库区进行微震检测定位分析。共检测到12 814次地震动事件,约为同时段三峡地震台网目录的4倍,完备震级从ML0.8下降至ML0.3,表明PALM方法可提高台网监测能力。基于构建的更完善的极微震-微震目录,分析库区地震活动空间分布特征。结果表明,巴东地区震源深度较浅,多在5 km以内,主要与岩溶、煤矿塌陷和浅表卸荷等非构造因素有关;秭归地区震源深度较深,多在5~15 km,主要与库区九畹溪和仙女山两条断裂活动有关。

四川大岗山水库蓄水对2022年泸定M_(S)6.8地震及余震的影响

2022年9月5日四川省泸定县发生M_(S)6.8地震,2023年1月发生MS 5.6余震。该次地震发生在鲜水河断裂南段,位于泸定地震震中东南部70 km的大岗山水库。该水库自建成蓄水后,引起库区北西侧磨西断裂的地震活动性显著变化:2014年11月水库达到预定水位之前的3年内,磨西段地震震级较小,多为0~2级微震;11月之后的3年内该区域地震震级和数量明显增加。文章利用高精度DEM数据、精准的地表断裂信息和地层地质信息,建立了三维孔隙弹性有限元数值模型,并根据震源机制解得到断层参数,定量计算了水库水位变化对地层孔隙压力、断层库仑应力的影响。在相关参数约束下的结果表明:水库蓄水造成M_(S)6.8地震震源位置在发震时刻的地层孔隙压力增加5 kPa,库仑应力增加3.6 kPa,对其发生具有一定促进作用。对于MS 5.6正断层型余震,水库蓄水造成其震源位置孔隙压力增加0.32 kPa,库仑应力降低0.69 kPa,对其所处断层面的滑动具有抑制作用。

长阳页岩气试验井压裂微震检测及特征

基于三峡地震台网2017年7—8月记录到的长阳地区发生的震群资料,通过PhaseNet进行震相识别、REAL震相关联、HypoInverse绝对定位、HypoDD相对定位等定位流程,构建了长阳页岩气试验井压裂震群的高分辨率地震目录。震群重定位结果表明:地震序列整体上沿着地表由南向北呈带状展布;在长阳页岩气试验井水力压裂前后,长阳地区的地震活动性非常低,但在试验井开始压裂期间,整体地震活动性猛增,最大震级可达ML3.4,但震群活动对周边区域地震活动性的影响较小;该震群成因可能为页岩气开采试验井水力压裂导致天阳坪断裂活化产生的。

紫坪铺水库与汶川地震关系的讨论

利用紫坪铺水库专用地震台站和加密地震台站资料,使用双差法完成了震前930次地震、汶川8级主震和5789次余震的重新定位,探讨了紫坪铺水库与汶川地震之间的关系,获得了有利于证明它们之间密切关联的事实:1)紫坪铺水库2005年蓄水之后出现了水磨、深溪沟和都江堰震群,地震应变能释放增加了200%,与库水位变化密切相关,在汶川地震前呈现出加速释放的现象;2)野外调查表明存在1条贯穿紫坪铺水库库区、走向NE的同震地表破裂带,与中滩铺断层位置一致;3)余震分布以中滩铺断层为界,西北上盘多,东南下盘少;4)利用P波初始到时重新确定出的汶川主震发震时刻为27分59.5秒,初始破裂深度在6～9km左右,直接位于水磨震群之下;5)汶川主震与发生在2008年4月5日水磨震群中的1次小震的各个台站到时差相同。

龙滩库区水库地震震源机制及应力场特征

文中使用龙滩水库地震监测台网记录的波形数据,采用P波初动、SH波和P渡位移振幅比数据计算震源机制解的FOCMEC方法,获取了龙滩库区2006年9月蓄水至2008年底发生的73次M_L2.0以上地震的震源机制解,并在此基础上反演了库区应力场。龙滩库区2006年10月蓄水以来发生的2级以上地震以逆断层型为主,由震源机制解获得的2个地震丛集区应力场的主压应力都近于水平,取向都为NWW-SEE。反映出蓄水后库区仍为以水平NWW-SEE向压应力为主的应力场结构,且最大主压应力倾角更为水平。而最大主张应力及中等应力轴的分布则不一致,显示出在近水平的主压应力背景下,龙滩库区局部应力场的非一致性。通过对龙滩水库地震机制解特征及应力场的认识和讨论,初步提出了龙滩水库诱发地震的发震机理,认为载荷作用所引起的剪应力增大不是龙滩水库蓄水诱发地震的主要因素,而蓄水所产生的孔隙压力作用和库水渗透的润滑弱化作用的耦合作用可能是主要的成因。

长江三峡水库诱发地震加密观测及地震成因初步分析

从2009年3月起在长江三峡水库湖北段建立了26个地震台站组成的加密台网。2009年3—12月记录到2995次ML-0.8～2.9地震。利用双差定位软件重新确定了2837次地震的精确位置。结果显示:长江三峡水库库区小震震群分布图像呈线性分布或团块状丛集分布,团块状丛集一般在距离库水边5km范围内,地震线性分布可以延伸到距库水边16km远的地方;长江三峡水库湖北段地震主要集中分布在香溪河附近的仙女山断裂北端及九湾溪断裂、泄滩乡以西的长江两岸和巴东北岸神龙溪及附近地区,震源深度<10km,平均在4km左右;库区地震活动频次与库水位升降过程正相关,说明属于水库诱发地震。巴东库区神龙溪两岸地震明显呈现出3条线性分布,通过对比该地区碳酸盐岩的分布特征,发现是由于水库蓄水后,库水从神龙溪两岸等地下暗河渗入而诱发地震的;而仙女山断裂过江段、九湾溪断裂和泄滩乡、沙镇溪镇西部地区等的地震可能与仙女山断裂带、牛口断裂或顺层解理等不连续结构面软化,导致岩体失稳而诱发了水库地震,但诱发机制仍需要进一步详细研究才能获得令人信服的结论。另外,在秭归县文化南和杨林、巴东县东壤口镇以东等地存在着煤矿开采引起的矿山诱发地震,在长江三峡库区两岸存在着一些塌陷地震。

水库诱发地震的震例比较与分析

以水库诱发地震震例数据库为基础,借鉴统计学原理,分析了全球已发震水库的150个震例在地震活动性、震源机制、构造地质和水文地质等方面的主要特征.研究结果表明：水库诱发地震发生在水库周围并在蓄水过程中,多数在距库岸10km范围之内、震源深度＜10km,少数沿着断裂带、溶洞、（温）泉等渗水通道向外延伸至十多km,甚至更远;水库诱发地震活动的频次和震级与库水水位变化呈现相关关系,大部分最大震级地震发生在首次达到最高水位之前或者之后的2～3个蓄水周期内,之后发生最高震级的水库诱发地震概率较少;水库诱发地震序列完整,一般多为“前震-主震-余震”或“前震-震群-余震”型,以1～3级微震居多,因此b值多在1左右;水库诱发地震具有较高的震中烈度、地震动频率和地面峰值加速度,但随距离增加烈度衰减很快;水库地震应力降明显小于同一级别的天然地震应力降,震源尺度大于同一级别的天然地震;水库诱发地震的发震地段具有良好的库水渗透的岩性条件和构造条件,张性断裂带、未胶结的断裂面、岩溶、（温）泉等成为库水的渗透通道.

三峡水库地区震源参数特征研究

利用2009年3—11月三峡水库地区26个流动台站记录到的数字地震观测资料,首先采用Atk inson方法得到该地区地震波衰减特征,再用Moya方法得到26个台站的场地响应。在此基础上,精确测定得到了该地区1,020个ML≥0.1地震的震源参数,开展了定标关系的研究。结果表明:1)三峡水库地区的非弹性衰减结果为Q(f)=112.0f0.918,具有较小的Q0值和较大的η值。除个别台站外,绝大部分流动台站的场地响应值位于1～10倍之间。2)地震矩M0随震级ML的增大而增大,两者之间存在较好的线性关系,统计关系为LogM0=1.07ML+9.76。应力降与地震大小之间的关系与Nuttli(1983)的板内地震为增加应力降(ISD)模型的结果比较吻合,统计关系为LogΔσ=0.66ML-3.08。3)三峡水库地区地震辐射能量和地震视应力均随震级的增大而增大,后者意味着大地震是比小地震具更高效率的地震能量辐射体。4)与同震级的构造地震相比,水库地震的应力降值比前者明显偏低,约小10倍。这可能是由于水库蓄水造成地下介质孔隙压力增大,从而导致在一个比较低的构造应力情况下发生水库诱发地震。

水库地震安全问题分析

通过分析水库周围社会及大坝安全的地震影响、世界水库的发展情况、目前对水库地震安全的认识及其研究技术,认为水库周围的地震威胁有天然地震和水库诱发地震两类,表现为对水库大坝及其附属工程和库区社会均产生影响,因此水库地震安全主要由大坝地震安全、大坝附属工程地震安全和库区社会安全构成。随着水库大坝建设速度的不断加快和大型水库的不断增加,尤其是复杂地质构造区水库建设的增多,有必要在对大坝地震安全进行研究的基础上,重视水库地震安全的研究。

水库诱发地震与构造地震震源参数特征差异性研究——以龙滩水库为例

首先讨论龙滩库区水库蓄水与地震活动之间的关系,发现龙滩水库诱发地震特征十分明显,地震共分5丛呈丛集分布.利用库区架设的24个固定和流动台站记录的数字记录资料,在研究得到龙滩库区非弹性衰减和台站场地响应的基础上,精确测定得到了该地区总共1616个ML≥0.1级地震的震源参数,比较了水库诱发地震与构造地震震源参数特征的差异,得到了以下主要结论:1)龙滩水库地震活动与水库蓄水关系密切,不同蓄水阶段5丛的地震活动状态不同,局部断裂构造发育以及岩石透水性能影响着地震活动对蓄水过程的响应.2)龙滩水库诱发地震的地震矩M0随震级ML的增大而增大,两者之间存在较好的线性关系,统计关系为LogM0=1.07 ML+10.17.应力降与地震大小之间的关系和Nuttli的板内地震为增加应力降(ISD)模型的结果比较吻合,统计关系为LogΔσ=0.71 ML-2.89.3)龙滩水库地区地震辐射能量和地震视应力均随震级的增大而增大,后者意味着大地震是比小地震更高效率的地震能量辐射体.4)总体上不同丛地震应力降水平存在差异.地震应力降空间分布上与库水深度有较好的一致性,即库水深的区域应力降水平高.5)与同震级的构造地震相比,水库诱发地震的应力降值比前者明显偏低,大约小10倍.这可能是由于水库蓄水造成地下介质孔隙压力增大或者水的润滑作用,从而导致在一个比较低的构造应力情况下发生水库诱发地震.

三峡水库地震类型:构造触发和非构造诱发地震及发震机理研究

选取三峡水库2003年6月135 m和2012年175 m蓄水后的典型水库地震活动为例进行水库地震类型分析。135 m蓄水后首发微震群活动为非构造型水库"诱发"地震的例证,成因类型包括岩溶型、矿塌型和浅表应力调整型;2012年175 m蓄水后秭归屈原镇ML3.8地震序列与仙女山断裂和九畹溪断裂关系密切,为构造型水库"触发"地震震例。从库水荷载和渗透作用两方面探讨水库地震发震机理,库水荷载与快速响应型地震关系密切,滞后响应型则主要由库水渗透引起。