“…Varotsos et al [1996] examine the history of the application of seismic electric signals (known as the VAN or SES precursors) for earthquake prediction, as well as efforts by several groups [Hamada, 1993;Gasperini, 1992, 1993; $hnirman et al, 1993; Takayama, 1993] to verify the prediction results statistically. These validation attempts lead to conflicting conclusions: on the one hand, Mulargia and Gasperini [1992,1993] state that the successful VAN predictions can be attributed to a random chance, whereas Hamada [1993], $hnirman et al [1993], and Takayama [1993] find a significant correlation between the VAN signals and ensuing earthquakes. Varotsos et al [1996] summarize the discussion, propose more detailed and specific prediction criteria and offer basic rules for testing the predictions.…”
Statistical verification of the VAN or SES (seismic electric signals) predictions from 1987–1989 is considered. The test is carried out with the updated rules proposed by Varotsos et al. [1996]. Although for the Greek (SI‐NOA) earthquake catalog the VAN method formally is successful, this high rate of success is due either to the retroactive adjustment of prediction rules or to the non‐randomness of seismicity. A simple prediction algorithm accounting for earthquake clustering (foreshock‐mainshock‐aftershock sequences), yields similar or even better forecast results. If we remove dependent events from the catalog, the ‘prediction effect’ becomes statistically insignificant. For the PDE (NOAA) catalog the test shows that the VAN predictions' rate of success can be attributed to chance.
“…Varotsos et al [1996] examine the history of the application of seismic electric signals (known as the VAN or SES precursors) for earthquake prediction, as well as efforts by several groups [Hamada, 1993;Gasperini, 1992, 1993; $hnirman et al, 1993; Takayama, 1993] to verify the prediction results statistically. These validation attempts lead to conflicting conclusions: on the one hand, Mulargia and Gasperini [1992,1993] state that the successful VAN predictions can be attributed to a random chance, whereas Hamada [1993], $hnirman et al [1993], and Takayama [1993] find a significant correlation between the VAN signals and ensuing earthquakes. Varotsos et al [1996] summarize the discussion, propose more detailed and specific prediction criteria and offer basic rules for testing the predictions.…”
Statistical verification of the VAN or SES (seismic electric signals) predictions from 1987–1989 is considered. The test is carried out with the updated rules proposed by Varotsos et al. [1996]. Although for the Greek (SI‐NOA) earthquake catalog the VAN method formally is successful, this high rate of success is due either to the retroactive adjustment of prediction rules or to the non‐randomness of seismicity. A simple prediction algorithm accounting for earthquake clustering (foreshock‐mainshock‐aftershock sequences), yields similar or even better forecast results. If we remove dependent events from the catalog, the ‘prediction effect’ becomes statistically insignificant. For the PDE (NOAA) catalog the test shows that the VAN predictions' rate of success can be attributed to chance.
“…The published contributions, although testing significance and not performance, have nevertheless raised some substantial questions. Thus in the criticism by Takayama [1993] of the statistical tests carried out by MG, in the reply by Mulargia and Gasperini [1993], and in further comments by VEVL, the arguments have centred on what is a fair test of significance, using a binomial formulation where the test statistic is the number of correct predictions of earthquakes with magnitude above a given threshold. The problem is to specify the probability p that a given prediction is successful "by chance," a term which has been used by many authors to represent some version of a null hypothesis.…”
Section: Testing the Van Hypothesismentioning
confidence: 99%
“…Dealing with inhomogeneity is difficult. There is little agreement on the details of how it should be done, and the disagreement is reflected in the widely differing significance levels under the various tests by MG, Mulargia and Gasperini [1993] and Takayama [1993].…”
Section: Testing the Van Hypothesismentioning
confidence: 99%
“…A test based on the binomial distribution, as employed by Hamada [1993], MG, Mulargia and Gasperini [1993], and Takayama [1993], is rather unsuitable anyway, because the test statistic treats all earthquakes exceeding the stated magnitude threshold equally, regardless of their actual magnitudes. In fact, larger earthquakes occur less frequently than smaller ones according to the well known Gutenberg-Richter relationship.…”
Assessment of the proposed VAN method for predicting earthquakes in Greece remains inconclusive. Authors who have attempted to evaluate the method have had to make their own subjective decisions about some features of the hypothesis, and to propose their own algorithms for testing against a null hypothesis. Different treatments of the inhomogeneity in space and time have lead to widely different conclusions. The binomial distribution has been misused in considering whether predictions may have been satisfied “by chance.” Objective tests on the performance of the method, using independent data, cannot begin until the VAN hypothesis and the proposed null hypothesis have been fully formulated.
“…One crucial point in the debate is the difference between the claims of Mulargia and Gasperini [1992] and Takayama [1993]. The difference stems from the question of how the spatial probability should be taken into account.…”
The VAN earthquake predictions were made on the basis of seismic electric signals (SES), but the debate seems to be directed toward the statistical significance of the predictions from seismic data only. Accordingly, applying a logistic regression model to seismicity, we present our estimation of the probability of earthquake occurrence in Greece. The main purpose of our study is to examine whether or not we can find a specific seismicity pattern that can be used to considerably increase probability estimates. Our estimation of the probability of occurrence of an earthquake of Ms ≥ 5.0 is less than 0.25 in all the cases that we have examined. If we lower the threshold magnitude from 5.0 to 4.3, we can find cases in which the probability becomes as high as 0.75, comparable to the success rate of the VAN method estimated by Hamada [1993]. In these cases, however, such a high probability is due mostly to aftershocks, and if aftershocks are removed from the data set, the probability falls below 0.5.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.