HAL Id: hal-00535499https://hal.archives-ouvertes.fr/hal-00535499

Submitted on 11 Nov 2010

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.

The comparison of macroseismic intensity scalesRoger M. W. Musson, Gottfried Grünthal, Max Stucchi

To cite this version:Roger M. W. Musson, Gottfried Grünthal, Max Stucchi. The comparison of macroseismic intensityscales. Journal of Seismology, Springer Verlag, 2009, 14 (2), pp.413-428. �10.1007/s10950-009-9172-0�.�hal-00535499�

J Seismol (2010) 14:413–428DOI 10.1007/s10950-009-9172-0

ORIGINAL ARTICLE

The comparison of macroseismic intensity scales

Roger M. W. Musson · Gottfried Grünthal ·Max Stucchi

Received: 28 August 2008 / Accepted: 14 May 2009 / Published online: 30 May 2009© Springer Science + Business Media B.V. 2009

Abstract The number of different macroseismicscales that have been used to express earthquakeshaking in the course of the last 200 years is notknown; it may reach three figures. The number ofimportant scales that have been widely adopted ismuch smaller, perhaps about eight, not countingminor variants. Where data sets exist that areexpressed in different scales, it is often necessaryto establish some sort of equivalence betweenthem, although best practice would be to reassignintensity values rather than convert them. Thisis particularly true because difference betweenworkers in assigning intensity is often greater thandifferences between the scales themselves, partic-ularly in cases where one scale may not be verywell defined. The extent to which a scale guides

Electronic supplementary material The online versionof this article (doi:10.1007/s10950-009-9172-0) containssupplementary material, which is available toauthorized users.

R. M. W. Musson (B)British Geological Survey, West Mains Road,Edinburgh, EH9 3LA, UKe-mail: rmwm@bgs.ac.uk

G. GrünthalGFZ German Research Centre for Geosciences,Telegrafenberg, 14473 Potsdam, Germany

M. StucchiINGV Milano, via Bassini 15, 20133 Milan, Italy

the user to arrive at a correct assessment of theintensity is a measure of the quality of the scale.There are a number of reasons why one shouldprefer one scale to another for routine use, andsome of these tend in different directions. If ascale has many tests (diagnostics) for each degree,it is more likely that the scale can be applied inany case that comes to hand, but if the diagnosticsare so numerous that they include ones that do notaccurately indicate any one intensity level, thenthe use of the scale will tend to produce falsevalues. The purpose of this paper is chiefly todiscuss in a general way the principles involved inthe analysis of intensity scales. Conversions fromdifferent scales to the European MacroseismicScale are discussed.

Keywords Intensity · Intensity scales ·Macroseismology · History of seismology

1 Introduction

In the beginning was the scale, and the name ofthe scale was Rossi–Forel.

The above is actually not strictly correct, butserves as a reminder that there was a time, at theend of the nineteenth century, when macroseismicintensities were assessed throughout the worldusing a single recognised intensity scale. This wasthe Rossi–Forel scale (de Rossi 1883), and thus,

414 J Seismol (2010) 14:413–428

“intensity” and “Rossi–Forel” (RF) were effec-tively synonymous, and intensity came in ten de-grees, since this was the number of degrees inthe RF scale. It is rather curious that today, RFintensity is largely forgotten.

Although RF was the first scale ever to bewidely adopted, it was, of course, not the firstintensity scale. It was itself a compromise be-tween the scales proposed independently andnear-simultaneously by de Rossi (1874) and Forel(1881). But the first recognisable intensity scalein the modern understanding of the term goesback to the work of Egen (1828) some 50 yearspreviously. Egen’s innovation was ahead of itstime.

In the course of time, the inadequacies of theRF scale led to the development of modified ver-sions, entirely new scales, and modifications thatso departed from the original that they were ef-fectively new scales. As of 1921, Charles Davisonwas able to identify 27 different intensity scales,though some of these were subjective or classi-fications of earthquakes rather than strength ofshaking at a place (Davison 1921). A follow-upstudy in 1933 increased the number of differentscales from 27 to 39, with between three and 12degrees (Davison 1933). Obviously, the numberhas considerably increased since then.

From the reader’s point of view, the multiplicityof scales can be confusing and cause problems ofinterpretation. This is all the more the case when(as is very frequent) the scale used in any studyis not distinctly identified. The comparison of dif-ferent intensity scales is thus an important issue,which has not really been studied in any depth inthe literature. Under the general heading of “com-parison”, two separate issues can be identified.The first is the vexed practical question of conver-sion between scales. If one wishes to establish acorrelation between (say) epicentral intensity andmagnitude, and one has a table of values from onestudy of RF intensities, and another of Medvedev–Sponheuer–Karník (MSK) intensities, can they becombined, and if so, how? The second is a moretheoretical question of the evaluation of scales.Can one say that one scale is better than another?Both these issues will be addressed in this paper.

It should be understood that intensity scaleshave more than one use in seismology. The orig-inal purpose of using intensity was to parame-terise earthquake effects in a descriptive way fromobservational data. More recently, intensity hasalso been used in a predictive way, in hazard andrisk studies, where loss estimates can be inferredfrom expected intensity values. The present paperis concerned chiefly with the observational useof intensity scales (as, for instance, by institutesresponsible for seismic monitoring, including web-based macroseismic surveys, or in studies of his-torical earthquakes). It should also be stressedthat this paper intends to discuss the general issuesinvolved in comparing and evaluating intensityscales; it is not intended to be a review of anyparticular scales in themselves.

2 Cancani and non-Cancani scales

However, there is a particular matter that requiressome preliminary discussion, and that is the num-ber of degrees a scale may have and, in particular,the division of intensity scales into two classes:firstly, those that have 12 degrees and descend ul-timately from the proposal of Cancani (1904), andother scales. Since the first group form a familyof scales that are broadly compatible with one an-other, comparison issues are different from casesin which totally diverse scales are being compared.Some historical background is helpful here (seealso Musson and Cecic 2002). Throughout thispaper, we use Arabic numerals for intensity ratherthan the now rather antiquated Roman numerals,which are ill-adapted to computer files.

The history of how the Cancani family evolvedis also the story of how the name of Rossi–Forelbecame forgotten, to be replaced to some extentby the name of Mercalli. Mercalli was actually theauthor of two intensity scales. The first of these(Mercalli 1883) is described by Davison (1921) as“merely an adaptation of the de Rossi scale”. Ithas six degrees unlike the ten degrees of de Rossiand is now more or less forgotten.

Mercalli’s second scale, published in 1902, is amodification of the RF Scale (Mercalli 1902). The

J Seismol (2010) 14:413–428 415

ten-degree structure is retained, but the descrip-tions of each degree are considerably amplified.For example, 5 RF is defined:

Felt generally by everyone; disturbance offurniture and beds; ringing of some bells.

In all these extracts, the translations of Davison(1921, 1933) are followed. In this paper, we takeexamples mostly from lower intensity degrees, ashere the general points of comparison can bedemonstrated more succinctly. Compare Mercalli:

Rather strong, felt generally indoors, but byfew outside, with waking of those asleep,with alarm of some persons, rattling of doors,ringing of bells, rather large oscillation ofsuspended objects, stopping of clocks.

Mercalli’s fuller version found favour withusers, and the scale was adopted as standard bythe Central Office of Meteorology and Geody-namics in Rome (Davison 1921). A large numberof incorrect remarks concerning this scale can befound in articles on the Internet, including com-mon assertions that Mercalli invented the idea ofintensity scales and that the 1902 scale had 12degrees.

A modification was proposed by Cancani(1904) for dealing with very strong earthquakes,which consisted of adding two extra degrees atthe top of the scale. Thus, the first 12-degree scalewas born. Mercalli’s (1902) scale already had titlesfor each degree, of which the top three (8–10)were “ruinous–disastrous–very disastrous”. Thetwo new degrees (11–12) were labelled “catastro-phe” and “enormous catastrophe”. This 12-degreestructure has come to dominate most of the scalesin major use since. Ironically (in view of later de-velopments), Cancani proposed that the new 12-degree scale be named the “Forel–Mercalli scale”(thus cutting out de Rossi). In general, it savesconfusion if scales are not named after people whoare not their authors, and it is much better to fol-low Davison (1921) by referring to this innovationof 1904 as the Cancani scale.

Cancani’s scale was never really usable as heproposed it, since his text consists only of the titlesof each degree, i.e. one or two words per degree,

plus a suggested peak ground acceleration equiv-alent. Consequently, the next development wasthe complete overhaul of the definitions of thedegrees by Sieberg (1912). This involved veryconsiderable expansion. Intensity 5 now read asfollows:

Rather strong.

Even during the busiest hours of the day, theearthquake is felt by many persons in theopen air. Indoors, the shaking of the wholebuilding is generally noticed, the feeling be-ing the same as when some heavy object(such as a sack or piece of furniture) falls inthe house; or the observers move, togetherwith chair, bed, etc., as in a ship on an ag-itated sea. Plants, the branches and weakerboughs of shrubs and trees sway visibly, asthey do with a moderate wind. Freely hang-ing objects, such as curtains and lamps, butnot heavy chandeliers, oscillate; small bellsring; the pendulums of clocks are stopped orswing more widely according as the directionof the shock is at right angles or parallel tothe plane of oscillation; similarly, stoppedpendulum clocks are set going; the strikingspring of clocks sounds; electric lights fail toact when contact of the conducting wire ismade; pictures rattle against the walls or aredisplaced; small quantities of liquid are spiltout of well-filled open vessels; ornaments,small standing frames fall and also objectsleaning against the wall; even light furnituremay be somewhat shifted from its place, rat-tling of furniture; doors and window-shuttersopen or shut; window panes crack. Sleepersas a rule are awakened. A few persons runinto the open air.

Clearly, this is changed far more from Mercalli(1902) than Mercalli is from Rossi–Forel. Not onlyis the description greatly expanded in detail, a newclass of information is added—effects on nature(plants sway visibly). And some details are actu-ally changed. Mercalli has few people outside feel-ing the earthquake at intensity 5, whilst Sieberghas many.

416 J Seismol (2010) 14:413–428

This scale is rightly referred to by Davison(1921) as the Sieberg scale, and although the 12-degree structure of later scales is derived fromCancani (hence the title of this section of thispaper), the content of most later scales is over-whelmingly influenced by Sieberg.

The new scale, clearly an advance on all pre-vious scales, became in a later version (Sieberg1923) referred to as the “Mercalli–Cancani scale,formulated by Sieberg”—or more simply, theMercalli–Cancani–Sieberg scale, or MCS. The de-finition of the intensity degrees 1 to 6 of Sieberg(1912) is identical with those of Sieberg (1923).The innovations in Sieberg (1923) concentrate onthe degrees 6 and 7.

When the Sieberg (1923) scale was translatedinto English, with some changes, by Wood andNeumann (1931), whilst acknowledging the workof Sieberg in the text of their paper, in the title,they picked out Mercalli and dropped Cancaniand Sieberg, and the result was the ModifiedMercalli scale of 1931. Davison (1933) is more ac-curate in referring to this as the Wood–Neumannscale. In the condensation of Sieberg’s text, muchhelpful detail is lost, and Wood and Neumann’sversion is not an improvement. Where Sieberg(1923) has:

. . . Plants, the branches and weaker boughsof shrubs and trees sway visibly, as they dowith a moderate wind. Freely hanging ob-jects, such as curtains and lamps, but notheavy chandeliers, oscillate. . . .

Wood and Neumann (1931) have

Hanging objects. . . swing generally or consid-erably. . . .Trees, bushes, shaken slightly.

The former is more vivid and clearer.Problems of nomenclature became more acute

25 years later when the scale was completelyoverhauled by Richter (1958). Richter’s changeswere far reaching—in particular, he introduceda series of building classes based on typology,facilitating the assessment of intensity in caseswhere weak buildings are damaged but strongones are not. This resulted again in a quite newscale, which, Richter realised, should be called

the Richter scale. However, by historical acci-dent, the phrase “Richter scale” was already inwide journalistic use for the local magnitude (ML)scale, and Richter correctly foresaw that havingan intensity scale called after him would causeimmense confusion. He therefore proposed thatthe scale be referred to as the Modified Mercalliscale of 1956.

This proved a bad decision; it would have beenbetter to call it something like the “Californiascale”. What happened was that thereafter, inten-sity data sets appeared with the intensity givenonly as “Modified Mercalli” (or MM or MMI)with no indication of which scale was intended.To make matters even worse, subsequent authorsfollowed the same course in producing furtherscales, and there is now a multiplicity of differentscales calling themselves “Modified Mercalli” ofone version or another, some of which are clearlynot compatible with either the 1931 or 1956 scales(for instance, the version by Brazee 1978). As aresult, in many papers, it is hard to know exactlywhat is meant when “MMI” appears.

The latest version of “Modified Mercalli” ap-pears to be that of Stover and Coffman (1993).One of the changes of this version is to downgradedamage to a few chimneys from intensity 7 tointensity 6.

One also notes a carelessness in some authors,especially on the Internet, in dropping the “mod-ified” and referring to the “Mercalli intensityscale” as if the 1902 version were still in use. Thispractice, justified by those who perpetrate it onthe grounds of “ease of use”, seems sometimes tostem from an ignorance that different scales exist.

In a parallel development, a new 12-degreescale was created in 1962 by Medvedev,Sponheuer and Kárník, published as the MSKscale (Medvedev et al. 1964). This was broadlybased on MCS, Wood and Neumann and Richter’swork, and a previous scale by Medvedev, butarranged in a more systematic way. It was widelyadopted in Europe.

In 1988, it was resolved by the European Seis-mological Commission to bring the MSK scalein line with modern building types and advancesin macroseismology. The result was a new scale,called the European Macroseismic Scale (EMS),published in Grünthal (1998).

J Seismol (2010) 14:413–428 417

Scales in the non-Cancani group can be dividedinto three categories:

1. Old scales predating 1904, e.g. Mercalli(1883);

2. Scales descended from the seven-degree scaleof Omori (1900);

3. Later scales deliberately different from theCancani family.

The second category comprises the differentversions of the Japanese Meteorological Agency(JMA) scale. The most recent version of 1996 hasincreased the number of degrees from seven tonine by dividing intensity 5 into “5 upper” and “5lower” and similarly for intensity 6 (JMA 1996).Present use of the JMA scale is made by convert-ing instrumental ground motion parameters intopseudo-intensity values (Yamazaki et al. 1998),so one has no idea whether a quoted intensityvalue for a modern Japanese earthquake actuallycorresponded to the description of effects for thatdegree of the scale being observed or not. Thereplacement of the original JMA scale with itsinstrumental version means that to all intents andpurposes, macroseismology is no longer practicedin Japan. Intensity is by definition a classificationof observed effects. With earthquakes in Japanbefore 1996, the isoseismal 5 JMA marked theboundary of the area of observed damage to prop-erty. With later earthquakes, intensity 5 JMA in-dicates only a physical ground motion believed tobe sufficient to cause damage to houses. Whetheror not damage actually occurred is unknown.

Scales in the third category are uncommon,but a fine example is found in Ambraseys andMelville (1982). This scale, intended for use forhistorical earthquakes in Iran, has five degrees.These are deliberately arranged so that intensity1 is the most damaging and 5 the least; the inverseorder to most other scales. This emphasises thatthe scale is intended to be unique to the matter inhand (historical Iranian sources) and not an aliasor simplification for MSK or MMI.

3 Conversion between intensity scales

How can one convert between intensity scales?In the case of scalar parameters like magnitude,

it is at least possible to regress one scale againstanother and produce a conversion formula thatcan then be applied to a given data set, to infer,say, Mw from Ms or vice versa. However, inten-sity is not a scalar; it is a classification accordingto the nature of effects. This is often expressedby the practice of writing intensities in Romannumerals to emphasise that these are integersand indivisible (the practice of adding decimalsin Arabic numerals after an intensity written inRoman numerals has, however, been seen—Klügel 2005 is an example). The modern tendencyis to write intensities with Arabic numerals asthese are easier to handle in a computer file, butthey are nonetheless integer classes (there wouldbe some justification in replacing the numbers inan intensity scale with letters: intensity A–L ina 12-degree scale. It should also be mentionedin passing that these strictures apply to intensityassessed from observations and not necessarily topredicted intensity inferred from models).

Ideally, conversion should not be done at all;the correct and best procedure is to return to theoriginal data and reassign values in the desiredintensity scale (Ambraseys et al. 1983, Grünthal1998—see especially Section 4.3). This is fre-quently very difficult or impossible, either becausethe original data no longer exist or because ofpractical constraints of time and effort. In suchcases, some form of conversion is necessary, al-though it will add extra uncertainty to the finalvalues.

One sometimes sees conversion diagrams inwhich different scales are represented as a seriesof blocks (one per degree of the scale) of differentwidths. One scale will be taken as a standard, andthe width and placing of blocks in the other scaleswill vary according to the perceived relationshipbetween the scales. Figure 1 illustrates the type ofscheme (this one is imaginary, but real examplesare found, for example, in Murphy and O’Brien1977; Krinitzsky and Chang 1988; Reiter 1990;Guidoboni and Ebel 2009). In it, one is led toinfer that intensity 5 in scale A is slightly higherthan intensity 5 in scale B, presumably becausesome diagnostics for intensity 5 in scale B appearin the description of intensity 4 in intensity scaleA, and some diagnostics for intensity 6 in scaleB appear under 5 in scale A. Figures of this type

418 J Seismol (2010) 14:413–428

Fig. 1 Overlapping blockcomparison of two scales(partial); this is a fictitiousexample to demonstratethe general principle

express something of the spread of correspon-dences between two scales in the eyes of an au-thor, but do not provide direct practical guidance.When dealing with integer classes, there is no“slightly higher”, and Fig. 1 would not provide onewith a clear means of converting a data set.

A possible application of the approach wouldbe in cases of this sort: imagine that for someearthquake, the only information available forDerby was the following (imaginary) report:

In every place, the earthquake was generallyfelt only by those indoors; even in Derby fewof those who were in the street at the timenoticed anything.

Now, consider two intensity scales, A and B,where the phrase “Felt by few persons outdoors”appears under intensity 4 in scale A, but underintensity 5 in scale B (leading to the sort of overlapshown in Fig. 1). If one supposes that a seismolo-gist would give an unqualified intensity assessmentfor Derby of 4 (scale A) or 5 (scale B) on theslender evidence above, then one can say thatFig. 1 at least alerts one to the possibility that somevalues of 4 (scale A) are the equivalent of 5 (scaleB). But one cannot know which, and one would

hope anyway that one would not encounter firm,unqualified intensity assessments that rested on asingle diagnostic in the way illustrated.

If a practitioner is faced with the task of givinga value to a data point in scale A when all hehas is a value in scale B, the question is whetherhe should assign one value or another. He has tomake a working decision, and from Fig. 1, one canread that intensity 5 B is broadly compatible with5 A, and therefore, one would translate 5 B to 5 Aif one had no other information. There may be acase where the two scales are so mismatched that(for instance) 5 B could be 4 A or 5 A with equallikelihood, and 4–5 A is the only answer. Butslight overlaps are mostly of theoretical interest.A table of equivalence such as that in Willmore(1979) is more practical in giving direct numericalcorrespondences. The author of that table (N.V.Shebalin) leaves the reader in no doubt that if hewishes to convert some data from MMI to MSK,where he has a value 5 MMI, his recommendedcourse of action is to write 5 MSK.

Ostensibly, it appears that there are two pos-sible ways to arrive at such a conversion table—an empirical way and a theoretical way. Theempirical way would be to look at what intensitieshave been assigned to a body of data in scale A,compare with the intensities given to the samedata using scale B and examine the results. Thetheoretical way is to compare the actual wordingof the scale degrees.

It might appear on the surface that the em-pirical approach is likely to be more robust, butin practice, this is not the case, especially notwhere all the scales to be compared are withinthe Cancani family. The problems are as follows.Firstly, an experienced seismologist may developa conception of the Platonic ideal for each inten-sity degree and may then tend to assign intensitiesaccording to this ideal irrespective of what thescale actually says. Consider the following illustra-tion. Suppose a description reads as follows:

In Derby, all over the town, windowsand doors rattled, crockery and other lightitems were shaken and most people wereawakened.

J Seismol (2010) 14:413–428 419

Now we consider two imaginary intensity scalesin the Cancani family. In scale A, intensity 4 isdefined as follows:

Many people indoors feel the earthquake.A few sleepers wake. Windows and doorsrattle.

Scale B has the following definition:

The earthquake is felt generally indoors, andmany or most are awakened. Crockery, win-dows, doors, etc are shaken perceptibly.

What is the intensity in Derby in scale A andB? Clearly, there is a good match for intensity 4in both scales. In the case of scale B, one wouldassign intensity 4 without hesitation. However,if one treats the definition in scale A literally,the intensity is actually higher than 4 because atintensity 4, only a few are awakened, whereas thereport states that most were. So the literal-mindedseismologist would assess an intensity of 4–5 us-ing scale A and 4 using scale B. However, theless-literally minded seismologist would reason asfollows, when using scale A: “This is obviouslyintensity 4. I won’t take into account the fact thatscale A considers that only few should wake—if allthe windows are rattling people will be woken bythe noise. Therefore I will assign intensity 4; thereis no evidence the intensity was any higher.”

This is perfectly good reasoning, but it erodesany distinction between the two scales as regardsvalues. In fact (and this is an important point thatwill be returned to later), the difference betweenscales A and B is not that intensity 4 is “slightlyhigher” in scale B but in the fact that using scaleB, both imaginary seismologists arrive at the samevalue (4), whereas using scale A, they arrive at dif-ferent values (4–5 and 4). As a general rule, differ-ences between seismologists in assessing intensitywith the same scale are greater than differencesassessed by the same seismologist using differentscales, for scales within the Cancani family. Thiscan often be seen in comparing evaluations ofhistorical earthquakes where the same earthquakedata have been assessed by different authors; nospecific examples are cited here, as we do notwish to appear to highlight particular instances,

but wherever the same data have been assignedintensities by different workers, differences willappear.

In fact, this follows directly from the subjectiveelement in the assessment of intensity. To saythat not everyone will make the same intensityassignment to the same data is a trivial obser-vation. Thus, any empirical comparison of, say,MCS and EMS, could only be done on the basisof “MCS assigned by Dr A” compared to “EMSassigned by Dr B”. MCS assigned by Dr C mightbe systematically different according to differentworking practices between A and C. For instance,Dr A might assume that any mention of damageguarantees an intensity of at least 7 MCS, whilstDr C may be prone to assuming that isolatedreports of damage refer to buildings in such poorcondition that they can be disregarded. This willlead to systematic differences in data sets evenusing the same scale. Therefore, the attempt tocompare scales empirically for general purposes isproblematic.

What if the same seismologist attempts to as-sign intensities using two different scales to ex-plore the differences that way (e.g. Neilson et al.1984)? If one treats the wording of each degreevery literally in order to wring out the maximumdifference, one ends up assessing intensity in anartificial way, since most experienced seismolo-gists would tend to interpret the scale in a freerway. If one uses the scales in a natural way, theseismologist’s Platonic ideal for each intensity de-gree starts to have an effect, and any differences inthe wording between two scales becomes eroded,as discussed above. One then tends to reach theconclusion that all scales in the Cancani family arelargely equivalent.

There are some exceptions to this where, evenin scales within the same family, there are someobvious and inescapable differences. For exam-ple, some 12-degree scales define intensity 1 as“the earthquake is not felt by anyone”, and anypositive observation, no matter how feeble, is atleast intensity 2 (e.g. MSK). Others (e.g. Brazee1978) consider that one or two observers feelingthe earthquake on upper floors is intensity 1. Thus,an observation that is unambiguously intensity 1

420 J Seismol (2010) 14:413–428

on the Brazee scale is unambiguously 2 on theMSK scale.

When comparing scales that are not related,however, the anchoring effect that comes fromfamiliarity with 12-degree scales no longer hassuch an influence. If the fictitious description ofeffects in Derby given above is to be assessed interms of the Mercalli (1883) scale, the seismologistmust decide whether it fits better to intensity 2:

Moderate, with general shaking of fasten-ings, chandelier-prisms, etc., and ringing ofsome bells.

Or intensity 3:

Strong, more or less general ringing of bells,stopping of clocks, and oscillation of lamps.

And now having a feeling that the rattling ofwindows and doors “ought” to be intensity 4 (atleast in MMI or MSK or EMS) cannot influencethe result.

Empirical conversion really requires a situationwhere it is desired to translate a particular data setfrom one scale to another, where the data set isbased on a single set of working practices (usuallymeaning it was compiled by one seismologist overa period of time). One might perhaps find fromobservation (assuming the original data exist) thatwherever this seismologist has assigned 7 MCS,the correct EMS assessment appears to be 6 EMS.In that case, and for this data set, one can concludethat 7 MCS = 6 EMS. But for other data setscompiled using different working practices, thismight not necessarily hold, and general empiricalguidelines cannot be given.

Deriving a theoretical conversion by directcomparison of the wording of the scale couldbe done, as already suggested, by disassemblingeach intensity degree description into individualdiagnostics and then seeing where these fall withrespect to a different scale. Thus, if one foundthat for intensity 4 in scale A there were eightdiagnostics, of which six were found repeated inscale B for intensity 4 and two under intensity 5,this would lead to the sort of comparison foundin Fig. 1. It does not answer the question of howsignificant the discrepancy is in terms of actualuse.

A more practical approach is to imagine thatthe description in scale A is a report and to at-tempt to assign an intensity to this report usingscale B. To take the definition of intensity 3 inthe first Mercalli scale as given above and give theequivalence in EMS-98, one would imagine that adescription of effects at some place matched theMercalli definition exactly. This could read, forinstance,

In Geneva, the shock was rather strong,and felt generally by those indoors, but byfew outside. Many sleepers were woken, andsome persons were alarmed by the rattlingof doors and ringing of bells. There was arather large oscillation of suspended objects,and some clocks stopped.

One would have little hesitation in assessing 5EMS. This can then be repeated in the same wayfor all intensity degrees, including those involvingdamage (with some assumptions about likely vul-nerability of buildings). In this way, one can arriveat a table of equivalence for Mercalli and EMS asshown in Table 1.

This process does involve some degree of sub-jectivity, but one can at least set out one’s reasonsfor arriving at the decisions made.

4 Conversion of the major scales to EMS-98

Following in the manner of Table 1, in Table 2,the exercise is repeated for the major intensityscales encountered: Rossi–Forel, MCS, MM56,MSK and JMA (JMA 1996 version). In each case,each degree of the scale was rewritten as an imag-inary news report and intensity assigned usingEMS-98. These imaginary reports are not givenhere, but it is a trivial exercise to imagine eachdegree of a given scale rewritten in the style of

Table 1 Conversion fromMercalli (1883) toEMS-98

Mercalli (1883) EMS-98

1 2 or 32 43 54 6 or 75 8 or 96 10 or 11

J Seismol (2010) 14:413–428 421

Table 2 Non-prescriptive guidelines to conversion from five major scales to EMS-98

RF EMS-98 MCS EMS-98 MMI 56 EMS-98 MSK EMS-98 JMA-96 EMS-98

0 11 1 1 1 1 1 1 1 1 2 or 32 2 2 2 2 2 2 2 2 43 3 3 3 3 3 3 3 3 4 or 54 4 4 4 4 4 4 4 4 55 5 5 5 5 5 5 5 5L 66 5 6 6 6 6 6 6 5U 77 6 7 7 7 7 7 7 6L 88 7 or 8 8 8 8 8 8 8 6U 9 or 109 9 9 9 9 9 9 9 7 1110 –a 10 10 10 10 10 10

11 11 11 –a 11 1112 –a 12 –a 12 –a

aThis intensity is defined in such a way that it relates to phenomena that do not represent strength of shaking, e.g. those dueto surface faulting, or reaches a saturation point in the scale where total damage refers to total damage to buildings withoutantiseismic design

a journalistic report (mostly a matter of changingthe present tense to the past tense). This table ispresented with some cautions:

1. In the first case, it must be recognised thatits use should be thought of as a last resortand that one should always attempt to reassignintensity values from the original data. Atthe very least, one should check conversionsagainst original data if at all possible.

2. These are theoretical relationships. In prac-tice, any intensity data set is a conflation of thescale and the working practices used to applyit. The relationships in Table 2 are proposedguidelines and should not be taken in a rigidand prescriptive way.

3. There might be a particular issue in the case ofMCS. Experience seems to show that MCS in-tensity assignments are frequently higher thanthose in EMS for the same data despite thefact that assigning EMS intensities to the MCSscale descriptions themselves generally leadsto equality in our opinion. The difference maylie in the way in which the scale has beeninterpreted; this point needs to be investigatedfurther.

4. There is no practical value, as already stated,in conversions that portray one scale as“slightly higher” than another. There is no“slightly higher” in an integer scale, so allconversions are given here in integers.

5. Conversions tend to break down at the topend of any scale. Most scales have a maximumdegree which is defined, in some form ofwords, as “everything is destroyed”. Themeaning of this is ambiguous; it depends onwhat existed beforehand. If buildings aresufficiently poor, complete destruction mayoccur at an intensity well below the maximumconceivable earthquake shaking. In suchcases, intensity scales saturate at what maybe only around 9 EMS. This is a particularproblem with the RF scale, which jumps frommoderate damage at 9 RF to total ruin at 10RF. Nor does it help to add extra diagnosticsnot related to building damage, such as large-scale changes in landscape. These effects areoften related to fault rupture or secondaryground effects and generally do not help todefine intensity degrees. Thus, in Table 2, inmost cases, the top degree of older scales isundefined in terms of EMS (but may be pre-sumed to be high). The issue is not critical, asthese top degrees are seldom, if ever, used inpractice.

6. A literal reading of MSK and EMS-98 mightsuggest some differences, especially withregard to the occurrence of slight damage.In practice, experienced users of the MSKscale were not treating it literally, but makingmental adjustments to the wording of the scalein order to arrive at intensity assessments that

422 J Seismol (2010) 14:413–428

were robust and consistent with the overallpattern of the data (e.g. Burton et al. 1984).One of the aims of the ESC Working Groupfor the revision of the MSK scale was tobring the wording of the scale in line withwhat seismologists were already doing as bestpractice (Grünthal 1989). Thus, whateverdifferences there might apparently be inthe wording of the scales, there should beno differences in MSK values and EMS val-ues if the MSK values were correctly assigned.The greatest difference is likely to be thatsome MSK assessments that were given assplit values due to uncertainty, e.g. 5–6 or6–7, in many cases can now be resolved to asingle degree. Most assignments of 5–6 MSKare likely to be 5 EMS, and 6–7 MSK is mostlikely to be 6 EMS.

7. The JMA scale presents particular problems.As mentioned previously, since routinepractice in Japan is to assign this from strongground motion records, whether a JMAintensity value actually corresponds to anysort of observed effect is unknown. Thedescriptive interpretations still exist, butthey are hard to interpret in terms of otherscales because of significant discrepancies.According to the JMA scale, an intensity (3JMA) where most people feel the earthquakeand some are frightened (5 EMS) is thethreshold at which objects begin to rattleslightly (4 EMS if not sometimes 3 EMS). TheJMA scale is unusual in having a degree zero(not felt). In most scales, intensity 1 is not felt.

5 The evaluation of intensity scales

In this section of the paper, we consider com-parisons between scales in the sense of decidingwhether one scale is better for use than another.Superficial comparisons can be made betweenscales in the sense of whether one scale is longerthan another, has more degrees in it, is intendedfor use in one country or another and so on.But there are also features that one can find that

make a scale actually more recommendable or lessrecommendable.

One can, for example, discard from consider-ation what Davison (1921) refers to as “personalscales” which are completely subjective, such asthe very early Macfarlane scale (Milne 1842)which was used by Peter Macfarlane to comparethe strengths of aftershocks of an earthquake on23 October 1839. Rating the mainshock as 10,Macfarlane considered that if he thought a shockwas half as strong, it rated 5, and so on. Obviously,this was a scale that no-one else could use.

The following are some criteria that need to betaken into consideration.

5.1 Applicability

A good scale should be easy to apply to manyindividual cases because the definitions are suffi-ciently copious to meet a range of different cir-cumstances. If the definitions are very short, withfew diagnostics, it may be hard in some cases tomatch a description of effects to the scale. Thus, ifone had a scale that contained only descriptions ofeffects indoors, one could not use it if all the datawere from observers outdoors.

A specific problem is posed by the assess-ment of intensity for modern buildings that haveearthquake-resistant design. For the same degreeof shaking, these buildings should show less dam-age than older types of building. Thus, a scale thattakes into account modern building types is likelyto be more useful than a scale written to dealonly with the building types that were prevalent50 years ago.

However, the applicability issue does not trans-late simply into “more diagnostics equals a betterscale”, as increasing the length of each descrip-tion causes one to run the risk of two pitfalls, asfollows.

5.2 Consistency

The first pitfall is the issue of consistency. Are allthe diagnostics for a given intensity really equiv-alent to the same degree of shaking? If a scale

J Seismol (2010) 14:413–428 423

yokes together two diagnostics in the same degreethat actually occur at different strengths of shak-ing, this will cause uncertainties in the applicationof the scale.

This was a particular fixation of Davison’s, as hemakes very clear in his 1921 review. In the scale hedevised for his own use, each intensity degree hasone diagnostic only, so he could be sure that eachdegree was internally consistent (Davison 1900).Of course, this extreme solution meant that thescale was often impossible to apply.

A novel approach was pursued by Brazee(1978) who attempted to determine empiricallyfrom a large data set which diagnostic should beassociated with which using distance as discrimi-nant. It is not really apparent that the results ofthis experiment are any better than other scaleswhere the arrangement was made using judge-ment, and the scale has not been widely adopted.

5.3 Discrimination

This is an issue similar to that of consistency.Whereas the previous paragraphs dealt with thequestion that a diagnostic might be in the wrongplace in the scale, the discrimination problem re-lates to cases where a diagnostic does not relateuniquely to any degree of the scale and is thusnot useful in establishing the level of intensity. Insome cases, a diagnostic may not be properly anexpression of intensity at all. This is an impor-tant issue, often not correctly appreciated, whichrequires particular consideration. It is importantto understand that intensity is an expression ofearthquake shaking, and therefore, all diagnosticsshould refer to phenomena that are the directresult of earthquake shaking. It is not just a col-lection of earthquake-related effects, and not alleffects produced by an earthquake are suitable tobe put into an intensity scale. For intensity to beuseful as a concept, there must be an evident re-lationship between effect A and intensity Y, suchthat the observation of A occurring is evidencethat the intensity was at least Y and also that theobservation that A did not occur is evidence thatthe intensity was less than Y. If effect A occursat any intensity, it is not useful to have it in an

intensity scale. Equally, if the absence of A (evenif the required sensor is present) is unrelated tointensity, then A does not belong in an intensityscale.

As a very clear example, it is notable that veryfew intensity scales ever include remarks abouthuman casualties, and with good reason (an ex-ception is found in Principia 1982). Clearly, manyother factors, like time of day, control the numberof observed fatalities. One could, however, find acorrelation between fatalities and intensity at leastto the extent that many deaths usually indicatea high intensity. However, the absence of manydeaths may mean only that people were out in thefields.

Direct fault rupture effects are a similar case.Absence of rupturing effects has to do with thecharacteristics of the fault (including its depth)and has nothing to do with shaking or intensity.Ground rupture, and effects contingent on it, thushave no place in an intensity scale. They may cor-relate with intensity (because typically, the inten-sity will be high where the fault breaks) but theyare not expressions of it. The fact that phenom-ena related to ground rupture were sometimesincorporated into earlier intensity scales does notmean that this was ever good practice. A commonexample relates to the bending of railway lines,which first appears in Omori (1900) for intensity6. Clearly, shaking of itself, no matter how strong,does not cause steel rails to bend. This is a sec-ondary effect due to either fault displacement orground failure.

Although effects on nature can be used, withgreat care, as reinforcing evidence of the level ofintensity, it is not possible to rely on them ex-clusively. This has been pointed out several times(e.g. Vogt et al. 1994; Musson et al. 1995). The useof effects of nature has been encouraged by someauthors on the grounds that otherwise, intensityvalues cannot be assigned at all in uninhabitedareas (Dengler and McPherson 1993; Serva 1994).This is a problem one has to live with; neither canone assign intensities in the sea. It is better to leavemountainous areas as blanks on the map than tofill them with unreliable intensity values that maynot actually show strength of shaking.

424 J Seismol (2010) 14:413–428

In conclusion, “applicability” is not always anadvantage if the scale is written in such a waythat real problems are concealed. The fact that aparticular scale is written in such a way that usingit one can reach an intensity assignment for a placethat might not be possible with a different scale isnot necessarily an advantage. The numbers maynot be meaningful.

5.4 Number of degrees

The resolving power of an intensity scale is ob-viously related to the number of degrees in thescale. Montessus de Ballore was sceptical of in-tensity scales and remarked once that the onlytrue scale would be one with two degrees—feltand not felt (Ballore 1916). One could obtain onlyvery limited information using a scale at such anextreme position in the continuum between therudimentary and the sophisticated. However, theextent to which one can resolve intensity degreesdepends on two things: firstly, the quality of thedata and, secondly, a certain irreducible chaos ineven the most detailed macroseismic data set.

If the quality of data is not good, and this isespecially true of historical data sets, one cannotresolve intensities to the same fineness as maybe defined in the scale. Thus, for a report, “InVerona many buildings were damaged,” the in-tensity in a classic 12-degree scale may be any-where from 6 to 8. Thus, a six-degree scale thatruns not felt–weak–moderate–strong–damaging–destructive may be as far as one can really dis-criminate (Musson 1991). On the other hand,with detailed contemporary data sets, such a scalewould be found limiting, which is why early scaleswith few degrees, like Mercalli’s first scale, werephased out in favour of scales of first ten and then12 degrees.

However, even with a very detailed and co-pious data set, it will be found in practice thatobservations of earthquake shaking are somewhatirregular due to the complexity and scattering ofearthquake waves in the upper crust and soil col-umn. As a result, even with many detailed reportsfrom each settlement, the seismologist may find

it hard to decide if the intensity is one degree oranother.

As a result, it seems to be the case that 12degrees is really the limit for a practical workingscale. Any more and the increased fineness ofresolution could not actually be exploited usingreal data. This is a good reason for avoiding half-intensity degrees, which would suppose that onecould have a workable 23-degree scale.

In fact, the top two intensities in a modern 12-degree scale are used so infrequently that theycan be effectively dropped, as is current USGSpractice (Wald 2006, personal communication),and has been proposed in New Zealand also(Dowrick 1996).

Normally, a collection of questionnaires froma place (including Internet-submitted responses)represents a very small and non-random sampleof the total population. Inferences from such asample as to the true intensity at the place maybe questionable even at the resolution of integerintensity values; thus, resolving values down tofractional or decimal degrees is probably a caseof spurious accuracy. Intensities assigned fromquestionnaire data may not even be accurate toa resolution of one degree of a 12-degree scale(however, hypothetical intensity values predictedfrom attenuation models can validly be expressedas decimals, as they do not reflect observationaldata).

An intensity scale is a practical instrument withan end in view, and the number of degrees inany particular scale is likely to be a reflectionof the purposes to which that scale is intendedto be put. Thus, having more or fewer degreesis not necessarily an advantage or disadvantage.It is no criticism of the scale in Ambraseys andMelville (1982) that it has only five degrees; thatis a reflection of what can be discriminated usingPersian historical records, and the scale does thetask for which it was constructed.

5.5 Regularity

Determining correlations between intensity andsimple measures of physical ground motion has

J Seismol (2010) 14:413–428 425

been proven difficult in the past and prone tolarge scatter (e.g. Murphy and O’Brien 1977; Kakaand Atkinson 2004). Even though the conceptof intensity is an integer classification, it is stillto be hoped that intensity will behave more orless as if it were a physical parameter and decaysmoothly with distance. For this to be the case,the degrees of the scale need to be defined insuch a way that the decrease in shaking witheach lower value is regular. This should be clearin intensity data point distributions or isoseismalmaps. If isoseismal maps drawn using a particularscale consistently showed a narrow gap betweenisoseismals 4 and 5, compared to the gap between3 and 4 and between 5 and 6, it would indicate thatthe scale was badly constructed (and similarly withintensity data point distributions).

Failure to find even spacing may reflect poorpractice rather than poor scale design. A commonobservation is to see isoseismal maps in whichthe spacing of isoseismals 5 and 4 is very wideand then the spacing suddenly drops below 4 sothat isoseismals 4, 3 and 2 are very tightly spaced.The most obvious explanation for this is poorpractice in assigning intensity. If intensity valuesare assigned to individual questionnaires insteadof to communities on the basis of several ques-tionnaires, then isolated returns that answer “yes”to “Did windows rattle?” will be rated 4. Thiscan result in a whole settlement being assignedintensity 4 on the basis of one observer with loosewindows.

In general, the 12-degree family of scales does agood job at preserving regular spacing when prop-erly applied, which is one reason why the outlineof the different degrees has been more or lessstable for a hundred years. A common complaintthat used to be addressed to the MSK scale wasthat there was a missing degree between 6 and7. This subject was investigated by the EuropeanSeismological Commission Working Group (WG)“Macroseismic Scales” (Grünthal 1993, 1998) inthe course of revising the MSK scale to createEMS, with the conclusion that no new degree isnecessary. Nor, for that matter, would such a newdegree be advisable, as it would break up theexisting regularity.

The RF scale is a good example of a scalewith regularity problems, especially in the jumpbetween 9 and 10 RF. This has been noted before,e.g. by Wood and Neumann (1931).

5.6 Reliability

If, as is argued above, all the Cancani family scalesare more or less equivalent in values, it mightbe wondered why any one should be preferableto any other. The key element in preferring onescale over another is the consistency with whichthe scale guides the user to the correct assignmentfor any place. As is well known, macroseismologycontains an element of subjectivity so that giventhe same data, two seismologists may assign dif-ferent intensity values. Usually, these differencesare minor: one worker may assess 4–5 (4 or 5)where another is more certain that the value is4. All the same, anything that reduces subjectivity(or errors) is welcome. Thus, if two workers us-ing scale A produce assessments of intensity thatsometimes differ by up to two degrees but whenusing scale B produce assessments that alwaysagree, then, other things being equal, scale B isbetter.

To some extent, reliability is a function of otherfactors already discussed. So, for instance, a scalethat has consistency problems is likely to causeusers to come to different assignments, as eachuser has to find his way over the obstacle.

Another issue here is that the variability ofearthquake effects in a given place means thata probabilistic approach, especially to damage, isadvantageous. If a scale simply states that for acertain intensity, “chimneys are broken off at theroof line”, users may react differently in caseswhere few or many, but not all chimneys, are soaffected. If a scale defines expected distributionsof effects rather than just the effects in isolation, itbecomes easier to recognise the correct intensityassignment. This was a considerable advance inmacroseismology, which came in really with thefirst version of the MSK scale in 1964.

One of the problems in the past that has led toinconsistent practices has been the fact that mostscales have been published as lists of definitions

426 J Seismol (2010) 14:413–428

without any corresponding guidance as to how theauthors of the scale envisaged it being applied. Apractice which has often been decried (rightly) inthe past is that of assigning intensity on the basisof the highest effect reported. Whilst this is un-doubtedly wrong, one must have some sympathyfor the observatory assistant who has been givena single sheet of paper with the Rossi–Forel scalewritten on it and told to start assigning intensities.For this reason, one of the first decisions takenby the ESC WG in revising the MSK scale wasthat the new scale should be accompanied by clearinstructions as to exactly how the scale should beinterpreted and applied. For this reason, EMS-98is not simply a sheet of definitions, but includesguidelines intended to reduce misapplication.

This discussion is predicated, of course, ontraditional assessment of intensity from data byhuman judgement. This is bypassed when auto-matic algorithms are used to assess data usingcompletely formulaic methods (e.g. Wald et al.1999; Musson 2006), but the traditional applica-tion of intensity scales will always be an issue inthe assessment of historical data sets.

6 Conclusions

In an ideal world, one should not attempt to makeconversions between different intensity scales butrather reassess data using the scale that one wishesto have the data expressed in. In practice, thismay not always be possible, and conversion tablesneed to be applied. For practical purposes, theseismologist needs to know if he has somethingassessed as 6 in scale A what he should write downfor the value in scale B.

Matters are complicated by the fact that exceptwhere scales are radically different in organisationor number of degrees, differences in practice be-tween seismologists tend to outweigh differencesbetween scales. Thus, a value of 6 in scale Amay have one value in scale B if it was origi-nally assigned by one worker and another valueif originally assigned by someone else. Hence theadvantage in going back to the original data.

To improve the usability of past data, we haveincluded here a table intended to give guidance

in converting data sets to EMS-98, but becauseof the way that any existing data set depends onworking practices as well as the scale used, Table 2should be used with care and checks made if at allpossible. Conversion in this manner should onlybe used where absolutely necessary; it would notbe good practice to regard this table as a futurepanacea such that old scales could continue to beused and then converted as required. This tableis essentially the authors’ assessments, in EMS-98, of imaginary press reports exactly expressingthe descriptions of the various intensities in otherscales. It is not, and cannot be, based on empiri-cal analysis of data sets because such an analysiswould only be valid with respect to the agencieswhich produced those particular data sets.

Not all scales are equal; some are better thanothers. In particular, those that reduce disagree-ments between workers are better scales foreveryday use. A good scale should guide everyoneto making the correct judgement and should min-imise the degree of subjectivity. This means mak-ing sure that the definitions of each intensitydegree are internally consistent in describingthings that do go together with the same degreeof shaking. Whilst it is an advantage for a scaleto deal with as many diagnostics as possible, it isnot an advantage to include a diagnostic that isnot reliably associated with any intensity degree.There is thus a trade-off between a scale thatcan be applied in very many cases to arrive ata number (but often the wrong one) and a scalewhich is less easy to apply in some cases (but givesmore dependable answers).

A strong way to reduce disparate intensity as-signments between workers is to have a scalethat not only contains definitions of the variousintensity degrees, but gives specific guidance as tohow these definitions should be interpreted andused. The chief example here is the EuropeanMacroseismic Scale, which is the first macroseis-mic scale to include extensive guidelines as part ofthe scale itself.

Acknowledgment The contribution of RMW Musson tothis paper is included with the permission of the Directorof the British Geological Survey NERC.

J Seismol (2010) 14:413–428 427

References

Ambraseys NN, Melville CP (1982) A history ofPersian earthquakes. Cambridge University Press,Cambridge

Ambraseys NN, Banda E, Irving J, Mallard D, Melville CP,Morse T, Muir Wood R, Munoz D, Serva L, ShilstonD, Surinach E, Vogt J (1983) Notes on historical seis-micity. Bull Seismol Soc Am 73:1917–1920

Ballore FdMd (1916) Earthquake intensity scales. BullSeismol Soc Am 6:227–231

Brazee RJ (1978) Reevaluation of Modified Mercalli inten-sity scale for earthquakes using distance as determi-nant. Technical Memorandum. NOAA, NOAA

Burton PW, Musson RMW, Neilson G (1984) Studies ofhistorical British earthquakes. BGS Global Seismol-ogy report no. 237, Edinburgh

Cancani A (1904) Sur l’emploi d’une double echelle sis-mique des intensitès, empirique et absolue. GerlandsBeitr Geophys 2:281–283

Davison C (1900) Scales of seismic intensity. Philos Mag50:44–53

Davison C (1921) On scales of seismic intensity and on theconstruction of isoseismal lines. Bull Seismol Soc Am11:95–129

Davison C (1933) Scales of seismic intensity: supplemen-tary paper. Bull Seismol Soc Am 23:158–166

de Rossi MS (1874) Bibliografia con annotazione. Bull Vul-canismo Ital 1:46–56

de Rossi MS (1883) Programma dell’osservatorio edarchivio centrale geodinamico presso il R. ComitatoGeologico d’Italia. Bull Vulcanismo Ital 10:3–128

Dengler L, McPherson R (1993) The 17 August 1991Honeydew earthquake, North Coast California: a casefor revising the Modified Mercalli scale in sparselypopulated areas. Bull Seismol Soc Am 83:1081–1094

Dowrick DJ (1996) The Modified Mercalli earthquake in-tensity scale—revisions arising from recent studies ofNew Zealand earthquakes. Bull N Z Soc Earthq Eng29:92–106

Egen PNC (1828) Über das Erdbeben in den Rhein-und Niederlanden vom 23. Februar. 1828. Ann Phys13:153–163

Forel FA (1881) Intensity scale. Arch Sci Phys Nat 6:465–466

Grünthal G (ed) (1989) Thoughts and proposals for theupdating of the MSK intensity scale. Central Institutefor the Physics of the Earth, Potsdam

Grünthal G (ed) (1993) European Macroseismic Scale1992 (up-dated MSK-scale). Cahiers du CentreEuropèen de Gèodynamique et de Seismologie. Con-seil de l’Europe, Conseil de l’Europe

Grünthal G (ed) (1998) European Macroseismic Scale1998. Cahiers du Centre Europèen de Gèodynamiqueet de Seismologie. Conseil de l’Europe, Conseil del’Europe

Guidoboni E, Ebel JE (2009) Earthquakes and tsunamis inthe past. Cambridge University Press, Cambridge

Japanese Meteorological Agency (1996) Explanationtable of JMA seismic intensity scale. http://www.jma.go.jp/jma/kishou/know/shindo/explane.html

Kaka S-L, Atkinson GM (2004) Relationships betweeninstrumental ground-motion parameters and Mod-ified Mercalli intensity in Eastern North Amer-ica. Bull Seismol Soc Am 94:1728–1736. doi:10.1785/012003228

Klügel J-U (2005) Problems in the application of theSSHAC probability method for assessing earthquakehazards at Swiss nuclear power plants. Eng Geol78:285–307. doi:10.1016/j.enggeo.2005.01.007

Krinitzsky EL, Chang FK (1988) Intensity-related earth-quake ground motions. Bull Assoc Eng Geol 25:425–435

Medvedev S, Sponheuer W, Karník V (1964) Neue seis-mische Skala Intensity scale of earthquakes, 7. Tagungder Europäischen Seismologischen Kommission vom24.9. bis 30.9.1962. In: Jena, Veröff. Institut fürBodendynamik und Erdbebenforschung in Jena,vol 77. Deutsche Akademie der Wissenschaften zuBerlin, pp 69–76

Mercalli G (1883) Vulcani e fenomeni vulcanici in Italia.In: Negri G, Stoppani A, Mercalli G (eds) Geologiad’Italia. Vallardi, pp 217–218

Mercalli G (1902) Sulle modificazioni proposte alla scalasismica De Rossi–Forel. Boll Soc Sismol Ital 8:184–191

Milne D (1842) Report of the committee appointed at themeeting of the British Association held at Plymouth,in 1841, for registering shocks of earthquakes in GreatBritain. Report of the British Association of the Ad-vancement of Science, pp 92–98

Murphy JR, O’Brien LJ (1977) The correlation of peakground acceleration amplitude with seismic intensityand other physical parameters. Bull Seismol Soc Am67:877–915

Musson RMW (1991) The use of the MSK intensity scalein the study of British earthquakes. In: 3rd interna-tional symposium on historical earthquakes in Europe,Liblice, pp 5–12

Musson RMW (2006) Automatic assessment of EMS-98 in-tensities. British Geological Survey, Technical ReportIR/06/048

Musson RMW, Cecic I (2002) Macroseismology. In: LeeWHK, Kanamori H, Jennings PC, Kisslinger C (eds)International handbook of earthquake and engineer-ing seismology. Academic, San Diego, pp 807–822

Musson RMW, Grünthal G, Stucchi M (1995) Comment on“the 17 August 1991 Honeydew earthquake: a case forrevising the Modified Mercalli scale in sparsely popu-lated areas” by Dengler and McPherson. Bull SeismolSoc Am 85:1266–1267

Neilson G, Musson RMW, Burton PW (1984) The“London” earthquake of 1580 April 6. Eng Geol20:113–142. doi:10.1016/0013-7952(84)90048-6

Omori F (1900) Intensity scale. Earthquake investigationcommittee publication in foreign languages, vol 4,pp 137–141

428 J Seismol (2010) 14:413–428

Principia Mechanica Ltd (1982) British earthquakes.Principia Mechanica Ltd, Cambridge

Reiter L (1990) Earthquake hazard analysis. ColumbiaUniveristy Press, New York

Richter CF (1958) Elementary seismology. Freeman, SanFrancisco

Serva L (1994) Ground effects in intensity scales. Terra Nova6:414–416. doi:10.1111/j.1365-3121.1994.tb00515.x

Sieberg A (1912) Über die makroseismische Bestimmung derErdbebenstärke. Gerlands Beitr Geophys 11:227–239

Sieberg A (1923) Geologische, physikalische und ange-wandte Erdbebenkunde. G. Fischer, Jena

Stover CW, Coffman JL (1993) Seismicity of the UnitedStates, 1568–1989 (revised). United States Govern-ment Printing Office, Washington

Vogt J, Musson RMW, Stucchi M (1994) Seismologi-cal and hydrological criteria for the new European

Macroseismic Scale (MSK-92). Nat Hazards 10:1–6.doi:10.1007/BF00643437

Wald DJ, Quitoriano V, Dengler LA, Dewey JW (1999)Utilization of the Internet for rapid community inten-sity maps. Seismol Res Lett 70:680–697

Willmore PL (ed) (1979) Manual of seismological ob-servatory practice. World Data Centre A for SolidEarth Geophysics, US Department of Commerce,NOAA, EDIS, World Data Centre A for Solid EarthGeophysics, US Department of Commerce, NOAA,EDIS, 165

Wood HO, Neumann F (1931) Modified Mercalli intensityscale of 1931. Bull Seismol Soc Am 21:277–283

Yamazaki F, Noda S, Meguro K (1998) Developmentsof early earthquake damage assessment systems inJapan. In: 7th international conference on structuralsafety and reliability, pp 1573–1580