论文部分内容阅读
Honkura 和 Tanaka(1996)以及 Aceves 等(1996)的计算为阐明这一论点的主要问题,即是否 VAN 预报可以归因于偶然,提供了重要的工具。Honkura 和 Tanaka(1996)的计算表明,对半径为120km 的圆形区域和22天的时间窗,在希腊 M_s≥5.0地震(EQ)发生的概率 P 小于0.25,甚至对于11天的时间窗来说更小。对更大震级阈值,即 M_s≥5.5或 M_s≥5.8(及由古登堡-里克特关系式来看),它们的 P 值更小。简单比较这些 P 值与 Mulargia 和 Gasperini(1992;1996a)的表格会立刻显示出 VAN 预报不能归因于偶然。记住,检查后者的表格会得到:1)VAN 成功率为40%～45%(当仅考虑与 M_s≥5.0地震相关时);2)VAN 警报率随地震震级闶值的提高而增加,对 M_(EQ)≥5.5和5.8,分别达到50%和60%。Honkura 和 Tanaka(1996)计算中出现的另一要点是“余震必须仔细处理。”这一点加强了我们在 Varotsos 等(1996a)的原则4和原则5中对 Mulargia 和 Gasperini(1992)对余震的不适当处理(基于泊松分布)的评论:①大大改变了显著性水平值;②预报与地震间的“正向时间相关”转变为“逆向时间相关”。这后一观点分别也被Honkura 和 Tanaka(1996)检查过,他们得出结论:“……考虑这种逆向时间相关的观点……,我们不能在目标区域里找到 M_s≥5.0的地震发生的高概率事件区。”在此答复中,我们还进行一些必要的阐述即在预报方法中,正如所料想的那样,在震级确定中存在试验误差时计算“成功率”和“警报率”有关的内容。
The calculations of Honkura and Tanaka (1996) and Aceves et al. (1996) are the main questions that elucidate this argument, namely, whether VAN forecasts can be attributed to chance and provide important tools. The calculations of Honkura and Tanaka (1996) show that the probability P occurring in a Greek M_s ≥5 earthquake (EQ) is less than 0.25 for a circular region of 120 km radius and a time window of 22 days, even for an 11 day time window smaller. For larger magnitude thresholds, that is, M_s ≧ 5.5 or M_s ≧ 5.8 (and from the Gutenberg-Richter relation), their P values are smaller. A simple comparison of these P values with those of Mulargia and Gasperini (1992; 1996a) will immediately show that VAN forecasts can not be attributed to chance. Remember that checking the latter table yields: 1) VAN success rates of 40% -45% (when considering only M_s ≥5 earthquakes); 2) VAN alarm rates increase with increasing earthquake magnitude 闶, For M_ (EQ) ≥5.5 and 5.8, they reach 50% and 60% respectively. Another point that emerges from the calculations of Honkura and Tanaka (1996) is that “aftershocks must be carefully handled.” This reinforces our observation of the effect of Mulargia and Gasperini (1992) on aftershocks in principles 4 and 5 of Varotsos et al. (1996a) Appropriate treatment (based on Poisson distribution) comments: ① Significantly changed the level of significance; ② The “forward-time correlation” between forecasts and earthquakes was changed to “reverse-time correlation.” This latter point of view has also been examined by Honkura and Tanaka (1996), respectively, and they conclude: “... considering this reverse-time-related point of view ... we can not find high occurrences of earthquakes with M_s ≧ 5.0 in the target area Probability event area. ”In this reply, we also make some necessary clarifications that, in the forecasting method, as expected, the calculation of“ success rate ”and“ alarm rate ”when there is a trial error in the magnitude determination .