Short- and Long-Term Earthquake Forecasts for California and Nevada

We present estimates of future earthquake rate density (probability per unit area, time, and magnitude) on a 0.1-degree grid for a region including California and Nevada, based only on data from past earthquakes. Our long-term forecast is not explicitly time-dependent, but it can be updated at any time to incorporate information from recent earthquakes. The present version, founded on several decades worth of data, is suitable for testing without updating over a five-year period as part of the experiment conducted by the Collaboratory for Study of Earthquake Predictability  (CSEP). The short-term forecast is meant to be updated daily and tested against similar models by CSEP. The short-term forecast includes a fraction of our long-term one plus time-dependent contributions from all previous earthquakes. Those contributions decrease with time according to the Omori law: proportional to the reciprocal of the elapsed time. Both forecasts estimate rate density using a radially symmetric spatial smoothing kernel decreasing approximately as the reciprocal of the square of epicentral distance, weighted according to the magnitude of each past earthquake. We made two versions of both the long- and short-term forecasts, based on the Advanced National Seismic System  (ANSS) and Preliminary Determinations of Epicenters (PDE) catalogs, respectively. The two versions are quite consistent, but for testing purposes we prefer those based on the ANSS catalog since it covers a longer time interval, is complete to a lower magnitude threshold and has more precise locations. Both forecasts apply to shallow earthquakes only (depth 25 km or less) and assume a tapered Gutenberg-Richter magnitude distribution extending to a lower threshold of 4.0.     

Probability gains of an epidemic-type aftershock sequence model in retrospective forecasting of <Emphasis Type="Italic">M</Emphasis> ≥ 5 earthquakes in Italy

A stochastic triggering (epidemic) model incorporating short-term clustering was fitted to the instrumental earthquake catalog of Italy for event with local magnitudes 2.6 and greater to optimize its ability to retrospectively forecast 33 target events of magnitude 5.0 and greater that occurred in the period 1990–2006. To obtain an unbiased evaluation of the information value of the model, forecasts of each event use parameter values obtained from data up to the end of the year preceding the target event. The results of the test are given in terms of the probability gain of the epidemic-type aftershock sequence (ETAS) model relative to a time-invariant Poisson model for each of the 33 target events. These probability gains range from 0.93 to 32000, with ten of the target events yielding a probability gain of at least 10. As the forecasting capability of the ETAS model is based on seismic activity recorded prior to the target earthquakes, the highest probability gains are associated with the occurrence of secondary mainshocks during seismic sequences. However, in nine of these cases, the largest mainshock of the sequence was marked by a probability gain larger than 50, having been preceded by previous smaller magnitude earthquakes. The overall evaluation of the performance of the epidemic model has been carried out by means of four popular statistical criteria: the relative operating characteristic diagram, the R score, the probability gain, and the log-likelihood ratio. These tests confirm the superior performance of the method with respect to a spatially varying, time-invariant Poisson model. Nevertheless, this method is characterized by a high false alarm rate, which would make its application in real circumstances problematic.     

Space- and Time-Dependent Probabilities for Earthquake Fault Systems from Numerical Simulations: Feasibility Study and First Results

In weather forecasting, current and past observational data are routinely assimilated into numerical simulations to produce ensemble forecasts of future events in a process termed “model steering”. Here we describe a similar approach that is motivated by analyses of previous forecasts of the Working Group on California Earthquake Probabilities (WGCEP). Our approach is adapted to the problem of earthquake forecasting using topologically realistic numerical simulations for the strike-slip fault system in California. By systematically comparing simulation data to observed paleoseismic data, a series of spatial probability density functions (PDFs) can be computed that describe the probable locations of future large earthquakes. We develop this approach and show examples of PDFs associated with magnitude M > 6.5 and M > 7.0 earthquakes in California.     

Earthquake Forecasting Using Hidden Markov Models

This paper develops a novel method, based on hidden Markov models, to forecast earthquakes and applies the method to mainshock seismic activity in southern California and western Nevada. The forecasts are of the probability of a mainshock within 1, 5, and 10 days in the entire study region or in specific subregions and are based on the observations available at the forecast time, namely the interevent times and locations of the previous mainshocks and the elapsed time since the most recent one. Hidden Markov models have been applied to many problems, including earthquake classification; this is the first application to earthquake forecasting.     

Optimization of Seismicity-Based Forecasts

In this paper, the extent to which some improvement can be made in seismicity-based earthquake forecasting methods are examined. Two methods that employ the statistics and locations for past smaller earthquakes to determine the location of future large earthquakes, the pattern informatics (PI) index and the Benioff relative intensity (RI), are employed for both global and regional forecasting. Two approaches for forecast parameter estimation, the TM metric and threshold optimization, are applied to these methods and the results evaluated. Application of the TM metric allows for estimation of both the training and forecast time intervals as well as the minimum magnitude cutoff and spatial discretization. The threshold optimization scheme is employed in order to formulate a binary forecast that maximizes the Pierce’s skill score. The combined application of these techniques is successful in forecasting those large events that occurred in Haiti, Chile, and California in 2010, on both global and regional scales.     

Using Grey System Theory for Earthquake Forecast

By combining conventional grey correlation analysis,grey clustering method and grey forecasting methods with our multi-goal forecast thoughts and the techniques of grey time series processing,we develop six different grey earthquake forecast models in this paper,Using the record of major earthquakes in Japan from 1872 to 1995,we forecast future earthquakes in Japan.We develop an earthquake forecast model.By using the major earthquakes in Japan from 1872 to 1984,we forecast earthquakes from 1985 to 1995 and check the precision of the grey earthquake models.We find that the grey system theory can be applied to earthquake forecast.We introduce the above analysis methods and give a real example to evaluate and forecast.We also further discuss the problems of how to improve the precision of earthquake forecast and how to strengthen the forecast models in future research.     

Pattern Informatics and its Application for Optimal Forecasting of Large Earthquakes in Japan

Pattern Informatics (PI) technique can be used to detect precursory seismic activation or quiescence and make an earthquake forecast. Here we apply the PI method for optimal forecasting of large earthquakes in Japan, using the data catalogue maintained by the Japan Meteorological Agency. The PI method is tested to forecast large (magnitude m ≥ 5) earthquakes spanning the time period 1995–2004 in the Kobe region. Visual inspection and statistical testing show that the optimized PI method has forecasting skill, relative to the seismic intensity data often used as a standard null hypothesis. Moreover, we find in a retrospective forecast that the 1995 Kobe earthquake (m = 7.2) falls in a seismically anomalous area. Another approach to test the forecasting algorithm is to create a future potential map for large (m ≥ 5) earthquake events. This is illustrated using the Kobe and Tokyo regions for the forecast period 2000–2009. Based on the resulting Kobe map we point out several forecasted areas: The epicentral area of the 1995 Kobe earthquake, the Wakayama area, the Mie area, and the Aichi area. The Tokyo forecast map was created prior to the occurrence of the Oct. 23, 2004 Niigata earthquake (m = 6.8) and the principal aftershocks with 5.0 ≤ m. We find that these events were close to in a forecasted area on the Tokyo map. The PI technique for regional seismicity observation substantiates an example showing considerable promise as an intermediate-term earthquake forecasting in Japan.     

Retrospective Forecasting of M ≥ 4.0 Earthquakes in New Zealand

We have applied a variation of the Epidemic Type Aftershock Sequence (ETAS) model, which is a stochastic triggering epidemic model incorporating short-term clustering, to data collected by the New Zealand Seismological Observatory-Wellington (Geonet) for forecasting earthquakes of moderate and large magnitude in the New Zealand region. The model uses earthquake data only, with no explicit use of tectonic, geologic, or geodetic information. In this epidemic-type model every earthquake is regarded, at the same time, as being triggered by previous events and triggering following earthquakes. A maximum likelihood estimate of the model parameters has been performed on the learning period from 1960 to 2005 for earthquakes of magnitude 4.0 and larger. Forecast verification procedures have been carried out in a forward-retrospective way on the January 2006 to April 2008 data set, making use of statistical tools as the log-likelihood ratio, the Relative Operating Characteristics (ROC) diagrams, the Molchan error diagrams, the probability gain and the R-score. These procedures show that the clustering epidemic model achieves a log-likelihood ratio per event of the order of some units, and a probability gain up to several hundred times larger than a time-independent spatially uniform random forecasting hypothesis. The results show also that a significant component of the probability gain is linked to the time-independent spatial distribution of the seismicity used in the model.     

用响应比波速比监测预测瀑布沟水库诱发地震

介绍对瀑布沟水库用加卸载响应比和波速比监测预测水库诱发地震的实践过程,通过课题以往的预测和对应其后期实际发生的地震来分析,结果显示:地震频度库水加卸载响应比Yn和Cr值、库水下方地震波速比这两种方法都可以作为监测预测水库诱发地震的新方法。     

华北水氡异常与地震关系的Bayes判别分析

选取华北地区１９９０￣１９９８年８月较完整的水氡观测资料,笔者采用ｘ＾２统计检验法识别前兆异常,利用笔者建立的Ｂａｙｅｓ判别分析方法,对该地区水氡异常与中强震活动性的关系进行了内符检验和外推预测。在风险代价比Ｋｄｎ取４的情况下,内符检验的有震报准率ｃ为０．７１,预报占时率ｂ为０．３３,Ｒ值可达０．３８;外推有震报效率ｃ为０．５,时空占有率０．０５,Ｒ值为０．４５,能够正确预测１９９８年１月１０日张     

汶川8.0级地震前地震趋势分析意见的回顾

该文在系统整理自2001年昆仑山口西8.1级地震到汶川8.0级地震前年度地震趋势预测中提出的地震活动异常和分析意见的基础上, 研究汶川地震未能做出中长期预测的原因。 结果表明: 2001年昆仑山口西8.1级地震发生后, 判定中国大陆处于强震连发阶段, 仍有发生8级大震危险, 但此后的2002—2007年中国大陆周边接连发生大震, 而内部连续6年的7级地震平静, 以及中国大陆5、 6级地震相继出现的显著平静, 是导致2006年以后对中国大陆地震活动水平预测明显偏低的原因。 南北地震带中段一直是作为近几年可能发生强震的危险地区, 但2007年云南宁洱6.4级地震后, 对西南地区强震危险的紧迫性估计不足。 而汶川地震所在的龙门山地震带历史上没有7级以上地震记录, 也是该地震带未作为近几年地震重点危险区的原因之一。     

Conditional Probabilities for Large Events Estimated by Small Earthquake Rate

We examined forecasting quiescence and activation models to obtain the conditional probability that a large earthquake will occur in a specific time period on different scales in Taiwan. The basic idea of the quiescence and activation models is to use earthquakes that have magnitudes larger than the completeness magnitude to compute the expected properties of large earthquakes. We calculated the probability time series for the whole Taiwan region and for three subareas of Taiwan—the western, eastern, and northeastern Taiwan regions—using 40 years of data from the Central Weather Bureau catalog. In the probability time series for the eastern and northeastern Taiwan regions, a high probability value is usually yielded in cluster events such as events with foreshocks and events that all occur in a short time period. In addition to the time series, we produced probability maps by calculating the conditional probability for every grid point at the time just before a large earthquake. The probability maps show that high probability values are yielded around the epicenter before a large earthquake. The receiver operating characteristic (ROC) curves of the probability maps demonstrate that the probability maps are not random forecasts, but also suggest that lowering the magnitude of a forecasted large earthquake may not improve the forecast method itself. From both the probability time series and probability maps, it can be observed that the probability obtained from the quiescence model increases before a large earthquake and the probability obtained from the activation model increases as the large earthquakes occur. The results lead us to conclude that the quiescence model has better forecast potential than the activation model.     

基于小波变换和支持向量机的中国大陆强震预测

将小波变换和支持向量机用于中国大陆年度最大地震震级预测。 先用小波变换把中国大陆年度最大地震序列分解成几个不同尺度水平(频率)的子序列, 然后使用支持向量机对分解后的子序列分别进行预测, 最后通过重构几个子序列的支持向量机预测结果得到最终预测结果, 预测次年中国大陆最大地震震级。 与支持向量机和神经网络方法对比, 结果表明小波变换和支持向量机相结合方法具有更高的预测精度, 预测效果很好, 说明此方法可用于地震时间序列预测。     

Long-range Earthquake Forecasting with Every Earthquake a Precursor According to Scale

Scaling relations previously derived from examples of the precursory scale increase before major earthquakes show that the precursor is a long-term predictor of the time, magnitude, and location of the major earthquake. These relations are here taken as the basis of a stochastic forecasting model in which every earthquake is regarded as a precursor. The problem of identifying those earthquakes that are actually precursory is thus set aside, at the cost of limiting the strength of the resulting forecast. The contribution of an individual earthquake to the future distribution of hazard in time, magnitude and location is on a scale determined, through the scaling relations, by its magnitude. Provision is made for a contribution to be affected by other earthquakes close in time and location, e.g., an aftershock may be given low weight. Using the New Zealand catalogue, the model has been fitted to the forecasting of shallow earthquakes exceeding magnitude 5.75 over the period 1965–2000. It fits the data much better than a baseline Poisson model with a location distribution based on proximity to the epicenters of past earthquakes. Further, the model has been applied, with unchanged parameters, to the California region over the period 1975–2001. There also, it performs much better than the baseline model fitted to the same region over the period 1951–1974; the likelihood ratio is 10¹⁵ in favor of the present model. These results lend credence to the precursory scale increase phenomenon, and show that the scaling relations are pervasive in earthquake catalogues. The forecasting model provides a new baseline model against which future refinements, and other proposed models, can be tested. It may also prove to be useful in practice. Its applicability to other regions has still to be established.     

中国大陆强震时间序列预测的支持向量机方法

统计学习理论（Statistical Learning Theory或SLT）是研究有限样本情况下机器学习规律的理论。支持向量机（Support Vector Machines或SVM）是基于统计学习理论框架下的一种新的通用机器学习方法。它不但较好地解决了以往困扰很多学习方法的小样本、过学习、高维数、局部最小等实际难题，而且具有很强的泛化（预测）能力。文中使用支持向量机对中国大陆最大地震时间序列进行预测，预测次年的我国大陆最大地震震级，结果表明该方法具有较好的预报效果。研究结果还表明我国大陆强震活动除了与强震时间序列本身有关外，还与全球的强震活动、太阳黑子活动等有密切的关系。尽管这种关系还不清楚，但是通过支持向量机可以很好地反应出这种非线性关系。     

Forecasting aftershock activity: 1. Adaptive estimates based on the Omori and Gutenberg–Richter laws

The method for forecasting the intensity of the aftershock processes after strong earthquakes in different magnitude intervals is considered. The method is based on the joint use of the time model of the aftershock process and the Gutenberg–Richter law. The time model serves for estimating the intensity of the aftershock flow with a magnitude larger than or equal to the magnitude of completeness. The Gutenberg–Richter law is used for magnitude scaling. The suggested approach implements successive refinement of the parameters of both components of the method, which is the main novelty distinguishing it from the previous ones. This approach, to a significant extent, takes into account the variations in the parameters of the frequency–magnitude distribution, which often show themselves by the decreasing fraction of stronger aftershocks with time. Testing the method on eight aftershock sequences in the regions with different patterns of seismicity demonstrates the high probability of successful forecasts. The suggested technique can be employed in seismological monitoring centers for forecasting the aftershock activity of a strong earthquake based on the results of operational processing.     

考虑震源机制解的地震丛集区搜索方法

构造带区段的划分对地震区划与构造演化等研究有着重要作用,而地震丛集区可以为构造带区段的划分提供参考依据.本文发展了一种考虑震源机制解的地震丛集区搜索方法：基于震源机制解的四元数表示方法来定量评估不同地震的震源机制解的相似程度,并以满足给定相似度阈值条件的地震所占比例作为判据来确定地震丛集区.合成数据的测试结果显示,该方法能够对地震丛集区做有效搜索并区分具有不同震源机制特征的丛集区.将该方法应用于2011年日本M9.0东北大地震显示,搜索得到的地震丛集区与实际观测到的余震特征分布具有较好的一致性,表明该方法可以为实际的地震丛集区划分提供一条有效的途径.     

1993年7—9月的全球地震动态

１９９３年第三季度，全球地震活动水平为中等偏高，明显高于上半年平均水平。日本北海道西南近海发生７．６级浅源地震，但不属于日本海沟地震。埃及西奈半岛发生５．７级地震，为今年亚欧带西段之最大地震。马里亚纳群岛发生８．１级中深震，使西北太平洋地区地震水平达到全球第一。兴都库什地区接连发生三次较大中深震，可能对我国西部地区地震活动有影响。墨西哥恰帕斯州近海发生７．３级地震，美洲带新的地震活动轮回正式开始。印度南部发生６．３级中强震，属于板内地震。     

A new approach on seismic mortality estimations based on average population density

This study examines a new methodology to predict the final seismic mortality from earthquakes in China. Most studies established the association between mortality estimation and seismic intensity without considering the population density. In China, however, the data are not always available, especially when it comes to the very urgent relief situation in the disaster. And the population density varies greatly from region to region. This motivates the development of empirical models that use historical death data to provide the path to analyze the death tolls for earthquakes. The present paper employs the average population density to predict the final death tolls in earthquakes using a case-based reasoning model from realistic perspective. To validate the forecasting results, historical data from 18 large-scale earthquakes occurred in China are used to estimate the seismic morality of each case. And a typical earthquake case occurred in the northwest of Sichuan Province is employed to demonstrate the estimation of final death toll. The strength of this paper is that it provides scientific methods with overall forecast errors lower than 20 %, and opens the door for conducting final death forecasts with a qualitative and quantitative approach. Limitations and future research are also analyzed and discussed in the conclusion.     

Temporal and Spatial Variation of b-value before Maduo MS7.4 Earthquake

大量研究表明地震孕育过程中存在相对平静期,该阶段b值会相对下降,b值变化在地震活动性研究中起着十分重要的作用。2021年5月22日青海玛多发生M_S7.4地震,为研究地震前b值时空变化特征,本文截取2009年至震前地震目录,将地震目录分为去除余震目录、完整地震目录,对比研究b值变化特征。研究发现,玛多M_S7.4地震前1年b值开始低于均值且不断下降,至b值有上升趋势时地震发生,扩大到区域内其它5级以上地震,也符合此规律,地震发生后b值明显上升,短时间内又下降至较低位置,并一直处于较低位置直至下次地震发生;从b值空间扫描结果看,玛多M_S7.4地震前,震中位于低b值区域,该位置为b值最低处;通过不同时间段的b值空间扫描结果,可发现玛多M_S7.4地震发生前低b值区域向震中不断迁移,表明地震发生前震中附近应力不断集中;b值空间扫描时,完整地震目录掩盖了中强震震中区域低b值特性,去余震地震目录较好的凸显了中强震震中区域低b值特性。