一次地震预报有效性的“概率统计”评价法——中短期前兆及预报效能评价之二

Probability statistic method of evaluating efficiency of an earthquake prediction— Part 2 of evaluation of Mid- short term precursor and prediction efficiency

该文提出了评价一次地震预报有效性的概率统计法(套圈模型)。设预报4维空间中一个有限封闭的范围,实际地震落入其范围内即为"报准",否则为"未报准"。在报准的情况下,预报区间的大小就反映预报水平的高低。这可用预报区域内地震发生的自然概率P来衡量。与"打靶模型"相似,"套圈模型"评价上限是准理想预报尺度,下限是可容忍误差尺度。根据不同的尺度,计算相应的概率。以相应概率对数值之差为比例计算评价值。实际算例表明,此方法与距准误差评价结果相当吻合。

The paper puts forward the Probability Statistic Method of Evaluation Efficiency of an Earthquake Prediction ('Ringing-circle' Model). We assume that there is a limited close area in four-dimensional prediction space. We call it 'prediction nicety' if one earthquake occurred in such an area, otherwise 'prediction not nicety'. In the case of 'prediction nicety', the size of prediction area reflects prediction level, which could be evaluated by natural probability P of earthquake occurrence in prediction area. Evaluating upper limit of 'ringing-circle' model just like 'target practice' model is scale of quasi-ideal prediction, and lower limit is tolerated-error scale. According to different scales, we calculate the corresponding probability. We get evaluating value with ratio to logarithm difference of corresponding probability. Earthquake cases show that the method is consistent with evaluating method of 'off-center' error.