那么，1920年海原地震的震级是否被高估了。8.5这个数值的最初来源是哪里呢？

似乎没有多少人深究。搜寻之下，原来是Richter (1958)。后来的人几乎不提，但Chen and Molnar (1977)中写得很清楚。兰州所出的《1920年海原大地震》书中提到震级是Richter得到的，但没有引用相关文献。这个Richter就是里氏震级之冠名人，地震学鼻祖级人物。

Chen and Molnar (1979)中给出的地震矩，对应于8.3的矩震级。也小于8.5之说。

如果我们意识到8.5是指里氏震级的话，问题也不是多大，但是对于1900世纪初期，刚有仪器记录的一些大地震，我们没有这样的清晰概念，使之与后来常用的面波震级或矩震级混为一谈，而引发一系列学术研究层面的谬误。里氏震级适用于中小地震或近震（地震发生地与仪器记录较近）的震级标定，侧重于震波的高频成分的高频检震仪。对于大地震，容易有饱和的偏差缺陷。这是一个共识。有兴趣的人可以翻找Hiroo Kanamori等在引入矩震级时的论述。所以里氏震级的数值不适合于描述大地震。1920年海原大地震并非特例，如1906年旧金山大地震也如此：Richter（1958）给出的震级是8.3，大于现在文献中常用的矩震级（Mw）7.9.【也有的用Ms7.7级】。即：对于1906年旧金山大地震，我们不再用Richter的震级。对此地震的震级，有过较详细的研究和阐述。

似乎到了应该澄清和纠正的时候了，有人/地震学专家感兴趣吗？

正如Chen, Molnar 和我的邮件沟通所示。当初，在断裂产状和位移数据不明的情况下，他们的结果可能有误,而且这是目前可以矫正的。可以重新计算地震矩。

-----
To: Jing Liu
Cc: Peter Hale Molnar
Subject: RE: 1920 Haiyuan M~8 earthquake
Dear Dr. Liu:
Thank you for writing. I have not worked on this subject for a very long time and am pleasantly surprised that people still inquire about this paper from time to time.
Yes, Peter Molnar and I simply quoted the magnitude reported by Richter (which did not appear in the abstract). The idea is that the determination of the seismic moment should give an improved estimate of the true size of the Haiyuan earthquake.

For technical reasons that will not improve with the passage of time, such as unfavorable and poor-known instrument response, finite length of the mechanical recording arm, limitations in digitizing old paper copies, etc., the accuracy (not precision) of our estimate of the moment can easily be off by a factor of two.
Another major source of uncertainty, which could be narrowed down with new data/technology, is the average dip of the rupture at depth. As you may already know, the excitation of Rayleigh waves depends strongly on the dip angle of a dip-slip fault. This is particularly true for shallow-dipping faults where a subhorizonal fault approaches a degenerate case for Rayleigh wave excitation. In our study, we simply assumed a dip of 45 degrees because no other information was available back in the mid-1970's.
I have not followed recent research on the 1920 rupture, but suspect that field studies also suffer from poorly constrained fault width and dip at depth. In principle, deep-penetrating seismic profiles may be able to constrain the rupture's average dip at depth. Of course, mapping out a fault at depth is one thing; telling how much of it ruptured during a particular event is a far more difficult task. Perhaps CEA would be interested in carrying out such a study?
In short, I think it is safe to say that the magnitude of the Haiyuan event is unlikely to be 8.6, but I have no new basis to tell how much smaller the true magnitude is.
Best,
Wang-Ping

----------------

From: Peter Hale Molnar

I had forgotten how little we knew at that time.
The faulting is nearly pure strike-slip, presumably on a vertical plane, except at the east end.
I had forgotten that we assumed reverse (or was it normal) faulting.