我读过在《Nature》和《Science》杂志上发表的有关地震学研究的大多数文章，诸多观点经不起推敲。有人说《Nature》是一本“高级科普”式杂志，刊登的文章只要有点数据或现象支持，如果内容具有轰动性和社会性，很容易发表。下面以在《Nature》上发表的一篇论文为例，看看其发表的论文质量是否靠谱。

The disastrous magnitude-7.9 earthquake in Sichuan, China, in May 2008 was a surprise because it occurred on a fault that had had little recent seismicity. Such surprises occur often for earthquakes within continents. In contrast to plate boundaries where large (magnitude M>7) earthquakes occur at expected locations along the boundary faults, continental interiors like Sichuan or the central and eastern United States contain many old faults, most of which show little seismicity over the past hundred years for which we have seismological data. Hence, we are uncertain of the times and locations of future large earthquakes. Present earthquake hazard assessments typically assume that recent small earthquakes indicate the location of large future earthquakes.

点评：汶川地震被世界上许多地震学家认为是“意外”和“不可思议”的“事故”，其主要原因是：

（1）    过去对强震孕震规律不清楚，没有建立岩石破裂前兆过程的力学准则；

（2）    没有孕震时空范围的概念，也不知道如何划分孕震时空区域；

（3）    忘记了世人公认的能量守恒原理。

我们的研究清晰地表明，汶川地震与世界上其它强震一样，遵循着严格的确定性力学规律，没有“意外”，都是自然界演化的必然结果。

Large earthquakes are typically followed by aftershock activity that decays to a lower level interpreted as ‘normal’ background seismicity^2. This transition is difficult to identify precisely, because defining it depends on the area treated as the aftershock zone and the criterion used. For the majority of large earthquakes, which occur at plate boundaries, the transition generally occurs less than a decade after the main shock, as shown by both studies of individual events^3–5  and a large global compilation⁶ (Fig. 1a). This duration is therefore regarded as the norm. In contrast, aftershock sequences can be much longer in other tectonic settings. In diffuse plate boundary zones such as the Basin and Range, aftershocks often continue for fifty years or more^7–10 (Fig. 1b). Within continental interiors, aftershocks may continue for hundreds of years after the main shock. Seismicity in the areas of past large earthquakes, including those in New Madrid, Missouri (1811–1812), Charlevoix, Quebec (1663), and Basel, Switzerland (1356), may be aftershocks¹¹. In the New Madrid seismic zone, the best-studied of these areas, seismicity delineates the areas thought to have ruptured in the main shocks, appears to be decreasing (Fig. 1c) and the largest earthquakes are at the ends of the presumed ruptures; all three observations are often found in aftershock studies¹².

For most aftershock sequences, the number of earthquakes decreases approximately hyperbolically with time after the mainshock, following the empirical Omori’s Law. Hence the aftershock durations may be estimated, using the change in seismicity rates from decaying aftershocks to background seismicity, which can be directly measured for short aftershock durations^2,5,6. However, identifying this transition becomes difficult and then impossible as the aftershock duration approaches and then exceeds the length of the short earthquake record (Fig. 1). For historic earthquakes, further problems arise from incomplete and nonuniform catalogues. Nonetheless, bounds on the duration of some aftershock sequences can be inferred. A lower bound comes from treating today’s earthquakes as aftershocks, and an upper bound comes where there is paleoseismic evidence of large (probably magnitude greater than 7) earthquakes but little seismicity today.

Figure 2 shows aftershock durations for selected events from three tectonic settings: plate boundary faults, broad plate boundary zones, and continental interiors. Despite the uncertainties in estimating both the duration and the rate at which tectonic slip loads the faults, the data indicate an inverse relationship between the aftershock durations and the slip rates. Faults at plate boundaries that are loaded by the rapid (typically faster than 10 mmyr^-1) plate motion show aftershock durations of about ten years^3–6. Faults within broad plate boundary zones but off the main boundaries move at only a small fraction of the plate motion (a few millimetres per year^7–9) and have longer aftershock durations. Aftershocks continue today following the 1952 Kern County, California, and 1959  Hebgen Lake, Montana, earthquakes, and have a typical duration of about 100 years in the Central Nevada seismic belt¹⁰. Such a duration is consistent with the absence of aftershocks from large earthquakes on the Wasatch fault, the most recent of which occurred about 600 years ago⁹.

The longest aftershock sequences occur within continental plate interiors, which deform at rates typically less than a millimetre per year^13–15. Today’s New Madrid and Charlevoix seismicity give lower duration bounds. Upper bounds come from the absence of seismicity at sites of past earthquakes such as the Reelfoot Rift’s eastern margin¹⁶ in the New Madrid area and the Meers fault in Oklahoma¹⁷. Thus, within continents, aftershock sequences can last hundreds of years or longer.

It is reasonable that within continents aftershocks should continue long after large earthquakes. Aftershocks result from changes of stress and fault properties induced by the main shock. At a plate boundary, steady plate motion quickly reloads the fault after a large earthquake and overwhelms the effects of the main shock. Within continents, however, the faults are reloaded much more slowly, allowing aftershocks to continue much longer.

点评：大森经验公式虽然在地震学中得到广泛应用，但该公式缺乏严格的物理依据，用它估算余震持续时间只能是“估算”。用大森公式估算余震持续时间，需要知道背景地震活动性，其实背景地震活动性如何确定也是“人为”的，也没有“客观的规则”。再者余震分布范围的确定也需要知道孕震空间范围（地震区）。

我们同意“余震序列的长度和断层加载速率呈反比关系”的观点，因为加载速率快，岩石破裂的程度和速率越快，余震持续时间越短。原文继续说到：位于板块边界由快速(通常大于10 mm／a)板块运动所加载的断层，显示其余震持续时间在10年左右(根据已有文献得出的结论，但过去的研究能确定余震活动何时结束吗？显然不能，因为过去缺乏科学的方法和公认的准则)；位于宽阔的板块边界带且远离主边界的断层，其运动速度只占板块运动的一小部分(每年几毫米)，而其余震持续时间较长，最长的余震序列发生在大陆板块内部，那里的形变速率非常低，低于每年1mm，因而，大陆内部的余震序列可以延续数百年甚至更久。

大陆内部某地震区主震后的“余震持续时间可达数百年甚至更久”的说法值得商榷，因为余震持续时间除与断层加载速率有关外，还与断层性质等有关。一般对逆冲断层而言，余震持续时间较长；但对走滑断层而言，由于能量易释放，余震持续时间较短。

如何确定余震持续时间，需要用严格的力学准则计算。如我们根据孕震断层多锁固段脆性破裂理论并结合数据分析，确定的邢台7.2级地震后余震的持续时间不超过3年；海城1975年2月4日7.3级地震后，余震的持续时间为1978年5月18日。

该文的观点缺乏直接数据的支持，支持其说法的主要依据是一个经验公式，臆测的成分实在太多。由于过去无法确定孕震时空区域，难以分清预震、前震与余震事件。在基本概念都不清楚且无法辨别的条件下，不知道发表这种“吸引眼球、哗众取宠”的文章意欲何为？我甚至怀疑审稿人是否喝多了？至少对这篇文章而言，其研究内容和方法确实没有达到大家预期的《Nature》杂志水平要求，结论不够nature，人为的“水分”太多。

类似的这种文章在《Nature》和《Science》杂志上常见，结论不明确如“或许”、“可能”的文章不胜枚举，是大牛杂志本身还是作者热捧的原因导致这种“垃圾”文章的频频“出镜”，值得科学共同体同仁们的深思。

原文见：

Seth Stein & Mian Liu，Long aftershock sequences within continents and implications for earthquake hazard assessment, Nature, Vol 462|5 November 2009| doi:10.1038/nature08502

Long aftershock sequences within continents and.pdf