VOL. 84, NO.

B5

JOURNAL OF GEOPHYSICAL RESEARCH

MAY 10, 1979

A Moment Magnitude Scale

THOMAS C. HANKS

U.S. GeologicalSurvey, Menlo Park, California 94025

HIROO KANAMORI

Seismological Laboratory,CaliforniaInstituteof Technology, Pasadena, California 91125

The nearly coincidentforms of the relations betweenseismicmoment Mo and the magnitudesM,, Ms, and Mw imply a momentmagnitude scale,M - { log Mo - 10.7which is uniformlyvalid for 3 < M,
< 7,5 < Ms < 71, andMw > 71.

It is well known that the most widely used earthquake magnitude scales,ML (local magnitude), M, (surface wave magnitude),and rn0(body wavemagnitude),are, in principle, unbounded from above.It is equallywell knownthat, in fact, they are sobounded,and the reasons for this are understood in termsof the operationof finite bandwidthinstrumentation on the magnitude-dependent frequency characteristics of the elastic radiation excitedby earthquakesources. Using just these ideas,Hanks [ 1979]demonstrated how the maximumreported rn0 - 7 and maximum reportedM, - 8.3 can be rationalized rather precisely. (The upperlimit to rnoat -7occurswhenrn is determined from body waveamplitudes at periods near 1 s, for example, when rnois determinedfrom World-Wide Standard SeismographNetwork (WWSSN) data. Prior to the establishmentof the WWSSN in the early 1960's,many rnovalues weredetermined from longer-period amplitude measurements, and this is especially so in the caseof the larger earthquakes [Geller and Kanarnori, 1977]. Many of these older, longerperiod values are considerablygreater than 7; this period dependence of rn0 is also understoodin terms of frequencydependent source excitation.) Similarly, the maximum ML reported in southernCalifornia after more than 40 years of magnitude determination is 6.8 [Hilernan et al., 1973], the valuesof 7.1 for the earthquakeof December31, 1934,and 7.7 for the Kern County earthquakeof July 21, 1952,listedthere being M. Although Kanarnoriand Jennings [1978] have obtained M slightlylarger than 6.8 by synthesizing Wood-Anderson seismograms from available strong-motionaccelerogramsand Bolt [1978] has recentlydetermined M = 7.2 for the Kern County earthquakefrom distant stations,it seems likely that the upper limit to M is also near 7, principally because rnoand M are both obtainedfrom amplitudes at 1-s period and because of the form of the correlationbetweenrno and ML [Gutenberg and Richter, 1956].

Hanks and Thatcher[1972] pointed out that a magnitude scale baseddirectly on an estimateof the radiated energy, ratherthan the converse, would not only circumvent the difficulties associated with characterizing earthquake source strength with ,narrow-band time domainamplitudemeasurements, specifically magnitudesaturation,but had become practical with theincreased understanding of thegross spectral characteristics of earthquakesources that developed in the early 1970's. Kanarnori [1977]realized this possibility by independently estimating the radiatedenergy E, with the relation
Ao-

Eo = '-Mo
1

(1)

where Aa is the earthquakestressdrop and # is the shear modulus, reducing(1) to

E= 2X l04M

(2)

by taking advantageof the constancyof earthquakestress drops for shallow earthquakes[Aki, 1972; Thatcherand Hanks, 1973;KanarnoriandAnderson, 1975;Hanks, 1977],and using(2) in the Gutenberg-Richter relationbetween E andM
log E, = 1.5M, + 11.8 (3)

whereE is in ergs.The idea is that if M is bounded, sotoo is E as obtainedfrom (3), but if E, is knownindependently from (2), it may be usedon the left-handsideof (3) to determine a magnitude Mw that will not saturate.A significant featureof Kanarnori's [1977]definitionof Mw by (3) throughuseof (2) is that he found that M so definedis quite similarto M for a numberof earthquakes with M < 8, that is, well below the saturationlevel of M. This agreement atteststo the general

validityof both the Gutenberg-Richter E-M, relation (3) for

M < 8 and the useof (2) to estimate E independently. A second important featureof M as definedby Kanarnori Thus the magnitudescalesM, M, and rnoare'said to a momentmagnitude scale. This saturateat largemagnitude. Justasin the caseof peakacceler- [ 1977]is that it is intrinsically (2) on the ation data at a fixed distanceR [Hanks and Johnson,1976; momentmagnituderelation is, upon substituting sideof Hanks, 1979],M, M, and rnofor crustalearthquakes saturate left-handsideof (3) and M, for M on the right-hand for the samephysical reason: for largeenough earthquakes, all (3),
of thesenarrow-band time domain amplitude measurements log Mo = 1.5M,,,+ 16.1 (4) no longermeasure grossfaultingcharacteristics but onlylimiting conditionson localizedfailure along crustalfault zones. which is remarkablycoincidentwith the Mo-M relationship definedby Purcaru andBerckherner [1978]for 5 < Peak accelerationdata at R - 10 km no more measuregross empirically source propertiesfor an earthquake with seismicmoment M, < 7: Mo > 10 : dyn cm than does rn0or M for an earthquake of Mo log Mo = 1.5Ms + (16.1 4- 0.1) (5) > l0:7dvncm or M for an earthquake of Mo > 10 ag dyn cm. and the Mo-M relationshipempirically definedby Thatcher Copyright 1979by the AmericanGeophysical Union. and Hanks [1973] for southernCalifornia earthquakes (3 <
Paper number 9B0059.
0148-0227/79/009 B-0059501.00

2348

HANKS AND KANAMORI: A MOMENT MAGNITUDE SCALE

2349

ML 7): log Mo = 1.5ML + 16.0 (4), (5), and (6): (6)

TABLE 2. Seismic Moments and Magnitudes for Some Large California Earthquakes(1857-1906)

Thus a singlemomentmagnitudeM may be written from

M = t log Mo - 10.7 (7)

Date
Jan. 9, 1857 March 26, 1872 Feb. 9, 1890 Dec. 25, 1899

X 1025 dyn cm
900, 530-870* 500 15 15

Ms

M
7.9, 7.8-7.9 7.8 6.8 6.8

Apart from the scatterof the observations about the empirically definedrelations(3), (5), and (6), M as definedby (7) is uniformlyvalid with respect to 3 < M,. < 7, 5 <Ms <71,and M,, at larger magnitude.To the extent that coda duration magnitudes usedextensively for M,. < 3 earthquakes are tied to ML [e.g.,Lee et al., 1972],M asgivenby (6) shouldapplyto them as well, althoughit would be desirable to verifythiswith
Mo-coda duration data.

April 18, 1906

400, 350-430*

8t:]:

7.7, 7.7

Unless otherwise specified, Mo entries are from Hankset al. [1975].

*From Sieh [1977]. From Thatcher [1975].

:!:FromGutenberg and Richter [1954].

There are severalinteresting featuresof Table 2, all associatedwith the three'great'Californiaearthquakes of the historical record,the Fort Tejon (1857), OwensValley (1872), and main shock, M,. is 0.5 units lessthan Ms and 0.3 units lessthan San Francisco(1906) earthquakes.First, none of the earthM, almost certainly reflectingthe saturation level of ML quakescan be classified as a great earthquake on the M scale. around7. In the caseof the Point Mugu (1973) earthquake the Kanamori[1977] liststhree dozenearthquakes between1904 largeM, relativeto Ms and M reflects anomalously large 1-s and 1969largerthan the San Francisco earthquake,and in fact excitation for an earthquake with such a small Mo, in turn there are more than this, since the Mo given here for this suggesting a relativelyhigh stress drop [Ellsworth et al., 1973]. earthquakeis approximately2[timessmallerthan that given Similarly,ML for the DesertHot Springs (1948) earthquake is by Kanamori [1977]. Second,thesethree earthquakes all have very similarMo and M, suggesting that an upperlimit of Mo 10 '8 dyn cm andM - 8.0 mayexistfor Californiaearthquakes. TABLE 1. Seismic Momentsand Magnitudes for Southern California This upper limit is physicallyreasonable if, for California Earthquakes(1918-1973) earthquakes, the seismogenic depthof faultingdoesnot exceed 15-20 km and if fault lengthsdo not exceedseveralhundred X 10:5 kilometers.Finally, these three earthquakesare only someDate dyn cm M,. Ma M what larger than the Kern County (1952) earthquake,for which close-ininstrumental groundmotion recordsare availApril 21, 1918 15 6.8 6.8 able. While the coverage is far from ideal, the importance of July23, 1923 1 6t 6.0 theserecords for aseismic designpractices and the mitigation June 29, 1925 20 6t 6.8 Nov. 4, 1927 100, 65* 7.3 7.3, 7.2 of other earthquake hazardsis obviousenough,if indeedthe March 11, 1933 2 6.3 6t 6.2 Kern County earthquakeis near the upper limit in source May19 1940 30 6.4, 6.7 7.0 strengthfor California earthquakes.
July 1, 1941 Oct. 21, 1942 March 15, 1946 April 10, 1947 Dec. 4, 1948 July 21, 1952(main shock) July 21, 1952(aftershock) July 29, 1952(aftershock)
March 19, 1954

Table I presents Mo, ML, Ms, and M calculated from (7) for a number of significantsouthernCalifornia earthquakes between 1918and 1973, and Table 2 presents theseparameters, as available, for five large California earthquakesbetween 1857 and 1906. For the earthquakesbetween 1918 and 1973, there is, on balance,goodagreement between ML, Ms, and M. In the caseof the Imperial Valley (1940) earthquake, M,. is 0.6 unitslessthan M, revealing anomalously low 1-sgroundmotion amplitudes for what is otherwise a fairly largeearthquake by Californiastandards. In thecase of the Kern County(1952)

0.5 unitslargerthan M, whichmay againreflecta high stress drop [Thatcher andHanks, 1973],but Ms, although it is apparently lesswell determined, it also larger than M by the same
amount. As is the case for the observational scatter of the

momentmagnitudepairsfrom the appropriateempiricalrelation (i.e., equations (5) or (6)), deviations of M from Ms or ML can in most casesbe attributed to variable stressdrop or
saturation of M, or

0.9 9
1

7
1

200 3 3
4

5.9 6.5 6.3 6.2 6.5 7.2:!: 6.4 6.1

6.2

5.9 6} 6t 6.4 6.5+ 7.7

6.0 6.6 6.0 6.5 6.0 7.5 6.3 6.3

6.4

Acknowledgments. D.M. Boore, W. R. Thatcher, and R. L. Wessoncriticallyreadthe manuscript and madeseveral usefulsuggestions. This research was supportedin part by NSF grant ENV 76-81816. Publicationapprovedby the Director, U.S. GeologicalSurvey.Contribution 3187, Division of Geologicaland PlanetarySciences, California Instituteof Technology.
REFERENCES

April 9, 1968
Feb. 9, 1971 Feb. 21, 1973

6
10 0.1

6.4
6.4 5.9

6.7
6.6 5.2

6.5
6.6 5.3

Unlessotherwisespecified, Mo entriesare from Hanks et al. [1975], and ML entriesfrom Hileman et al. [1973]. All but the last three Ms entriesare from Gutenberg and Richter [1954], taken as Ms according to Gellerand Kanamori [1977]. Ms for April 9, 1968, and February9, 1971,are from Kanamori andAnderson [1975],and Ms for February21, 1973, is from the National Earthquake Information Service, U.S. Geological Survey. *From Yeh [1975]. From Trifunacand Brune [1970]. :]:FromBolt [1978].

Aki, K., Earthquakemechanism, Tectonophysics, 13, 423-446, 1972. Bolt, B. A., The localmagnitude M, of the Kern Countyearthquake of July 21, 1952,Bull. Seismol.Soc. Amer., 68, 513-515, 1978. Ellsworth,W. L., R. H. Campbell, D. P. Hill, R. A. Page, R. W.
Alewine, T. C. Hanks, T. H. Heaton, J. A. Hileman, H. Kanamori,

B. Minster, and J. H. Whitcomb, Point Mugu, California, earthquake of 21 February 1973and its aftershocks, Science,182, 11271129, 1973.

Geller, R. J., and H. Kanamori, Magnitudesof great shallowearthquakes from 1904-1952, Bull. Seismol.Soc. Amer., 67, 587-598,
1977.

Gutenberg,B., and C. F. Richter,Seismicity of theEarth, 2nd ed., 310 pp., PrincetonUniversity Press,Princeton, N.J.. 1954.

2350

HANKS AND KANAMORI: A MOMENT MAGNITUDE SCALE

Gutenberg,B., and C. F. Richter, Magnitude and energyof earthquakes,,,Inn. Geofis.,9, 1-15, 1956. Hanks, T. C., Earthquakestress drops,ambienttectonicstresses and stresses that driveplatemotions,Pure,4ppl.Geophys., 115, 441-458,
1977.

Kanamori, H., and D. L. Anderson,Theoreticalbasisof someempirical relations in seismology,Bull. Seismol.Soc. Amer., 65, 10731096, 1975.

Kanamori, H., and P. C. Jennings, Determinationof local magnitude,

ML, fromstrong-motion accelerograms, Bull.Seismol. Soc.Ame.,

68, 471-485, 1978. Hanks, T. C., b-valuesand o -* seismicsourcemodels:Implications for tectonicstress variationsalongactivecrustalfault zonesand the Purcaru, G., and H. Berckhemer,A magnitude scalefor very large estimation of high-frequency strong ground motion, J. Geophys. earthquakes, Tectonophysics, 49, 189-198, 1978. Res., this issue, 1979. Sieh, K. E., A study of Holocene displacement history along the south-centralreach of the San Andreas fault, Ph.D. thesis,Stanford Hanks, T. C., and D. A. Johnson,Geophysical assessment of peak accelerations, Bull. Seisrnol.Soc. ,4mer., 66, 959-968, 1976. Univ., Stanford, Calif., 1977. Hanks, T. C., and W. Thatcher,A graphical respresentation of seismic Thatcher, W., Strain accumulation and releasemechanismof the 1906 sourceparameters, J. Geophys. Res., 77, 4393-4405, 1972. San Franciscoearthquake,J. Geophys.Res., 80, 4862-4880, 1975. Hanks, T. C., J. A. Hileman, and W. Thatcher,Seismic moments of Thatcher,W., and T. C. Hanks, Sourceparameters of southern Calithe largerearthquakes of the southernCalifornia region,Geol.Soc. fornia earthquakes, J. Geophys. Res., 78, 8547-8576, 1973. ,4mer. Bull., 86, 1131-1139, 1975. Trifunac, M.D., and J. N. Brune,Complexityof energy release during Hileman, J. A., C. R. Allen, and J. M. Nordquist, Seismicity of the the Imperial Valley, California, earthquakeof 1940,Bull. Seismol. Soc. ,4mer., 60, 137-160, 1970. southernCalifornia region 1 January 1932 to 31 December 1972, report, Seismol. Lab., Calif. Inst. of Technol.,Pasadena, 1973. Yeh, H.-C., Mechanism of the 1927Lompocearthquake from surface Lee, W. H. K., R. E. Bennet, and K. L. M eahger, A method of wave analysis,M.S. thesis,Univ. of Wash., Seattle, 1975. estimatingmagnitudeof local earthquakes from signalduration, (ReceivedAugust 22, 1978; openfile report, U.S. Geol. Surv., Menlo Park, Calif., 1972. revised December 20, 1978; Kanamori, H., The energyreleasein great earthquakes, J. Geophys. Res., 82, 2981-2987, 1977, acceptedJanuary 8, 1979.)