Regression analysis of MCS intensity and ground motion parameters in Italy and its application in ShakeMap

The use of intensity scales is historically important because no instrumentation is necessary, and useful measurements on the level of shaking can be made by an unequipped observer (e.g. Musson 2002). To some extent, the mid-years of the 20th century saw a decline in interest of macroseismic investigations, since large improvements were made in instrumental monitoring. However, since the mid-1970s there has been a resurgence in the subject since macroseismic data are essential for revision of historical seismicity and are of great importance in seismic hazard assessments. It follows that macroseismic studies of modern earthquakes are still crucial for (i) assessing the size of historical earthquakes; (ii) studying local ground-motion attenuation and (iii) investigations of vulnerability, seismic hazard and seismic risk. Since the late 1990s, the software package ShakeMap (Wald et al. 1999b) which seeks to estimate rapidly (few minutes) the level of ground shaking resulting from an earthquake has been proposed and implemented in several parts of the world (e.g. USA, Canada, Iceland, Italy and at local scales, for the city of Seattle). ShakeMap is a seismologically based interpolation algorithm that combines observed data and seismological knowledge to produce maps of peak ground motion (PGM). The shaking is represented through maps of peak ground acceleration (PGA), peak-ground velocity (PGV), response spectral acceleration (SA), and ground-motion shaking intensity. The ‘instrumental intensity’ values are derived from the conversion of PGM into intensity values (e.g. Wald et al. 1999a). These maps have become adopted worldwide to provide quantitative, first order assessments of the level of shaking and of the extent of potential earthquake damage. In particular, intensities have been found informative by non-expert audiences unfamiliar with instrumental ground motion parameters. The intensity values are derived from the ground-motion recorded values, using a correlation relationship. For the USGS-ShakeMap standard distribution this calibration has been performed using California earthquakes ground-motion data and the Mercalli Modified (MM) intensity scale (e.g. Wald et al. 1999a). In Italy, the software ShakeMap has been operational at the ‘Istituto Nazionale di Geofisica e Vulcanologia’ since 2006 (Michelini et al. 2008) and the intensity maps of peak ground motion shaking adopt the California relationship of Wald et al. (1999a). In Italy, however, the analysis of historical seismicity through the use of the macroseismic studies has a long tradition. The Mercalli–Cancani–Sieberg (MCS) Scale (Sieberg 1930), is the scale adopted in Italy. MCS combines an earlier ten-degree scale proposed by Mercalli (1902), the evolution of this scale with additional two-degree introduced for dealing with very strong earthquakes by Cancani (1904) and the successive remodulation by Sieberg (1912). To this regard, Musson et al. (2009) provide a thorough assessment of the various scales and of their evolution through time. There are two main reasons that have lead us to re-calibrate the conversion scale between peak ground motion and the reported MCS intensity data. The first follows from the fact that the MM instrumental intensity adopted within the implementation of ShakeMap (Michelini et al. 2008) can be misleading as the MCS representation is customary in Italy. Consequently, differences between the two scales can cause confusion. The second follows from the large number of macroseismic data available for past events in Italy (i.e. Stucchi et al. 2007). These data have been also used to generate scenarios for seismic hazard analysis. The aim of this work is to develop a new correlation relationship between recorded peak ground motions and reported MCS intensities for Italy. The derived MCS instrumental intensity relation is intended to be introduced for the calculation of shakemaps in Italy. To this regard, the intensity maps are the most viewed output from non-specialist audience when consulting, for example, the INGV ShakeMap portal (Michelini et al. 2008). For this reason, it is important to maintain the same intensity scale between the shakemaps and the other products that represent intensities throughout the Italian territory (i.e. the Italian database of macroseismic information, DBMI (Stucchi et al. 2007), and the ‘Did You Feel It’ maps (). In addition, access to a relation that allows conversion between MCS intensities and PGM's allows for the calculation of PGM's ground estimates for historical events which can be of high relevance when attempting to reconstruct past ground motion scenarios. The problem of the correlation between the reported intensity and the ground motion parameters has been debated at length in the literature. Although it is largely accepted that there is a ‘relation’ between intensity and the logarithm of the peak ground motions, either in PGA, or in PGV (e.g. Cancani 1904; Gutenberg & Richter 1942; Kawasumi 1951; Hershberger 1956; Ambraseys 1975; Margottini et al. 1992; Wald et al. 1999a; Faccioli & Cauzzi 2006; Gómez Capera et al. 2007, and see references therein), it has not yet been proposed a physical relation capable to represent it, and the empirical regressions proposed are mainly statistical. We also note that, being the intensity scale based on observations and not on instrumental values, there is no guarantee that a logarithmic relation is effectively applicable. This has been long recognized by several authors (e.g. Hershberger 1956; Ambraseys 1975) who recommended much caution in using these relations. Among all the works available in literature, it seems that the principal differences consist in the selection of the data base. Recently, a good overview of this topic at the global scale, and for Italy in particular, has been prepared by Gómez Capera et al. (2007). In general, the relations are obtained at regional scales, with the exception of the studies by Ambraseys (1975) who proposes a single for and the and et al. who adopt a for Italy, and This that work on regional data base. from some & that for the & and & that and for and that the all the regressions adopt the same between intensity and the logarithm of the peak ground The in the of the data. In general, some works of the use the of the recorded ground motion for intensity (e.g. Hershberger 1956; & 1975; & Wald et al. mainly (e.g. & Margottini et al. 1992; Faccioli & Cauzzi 2006; Gómez Capera et al. have not to the peak values for intensity We note that by using data into intensity the of the large of the peak ground motion data for intensity intensity a single of peak ground motion is through the and in the the that data have on the and with the exception of Gómez Capera et al. all adopted regressions the of the and this be at the of some in the resulting a that the between the and the of a follows from the use of macroseismic scales throughout the world (i.e. the MM for USA, the and MCS for and the for analysis by the studies performed on Italian data by Margottini et al. Faccioli & Cauzzi and Gómez Capera et al. (2007). Margottini et al. obtained first an empirical correlation between and intensity for The two works used and the data earlier by Margottini et al. Faccioli & Cauzzi a relation for intensity and PGV using Gómez Capera et al. used data and adopted the et al. can be as a of the of shaking at an in of observed on and The fact that it is a a physical to some on these is being a scale, and caution is to (i.e. ground and a (i.e. Margottini et al. are the first to provide a data that peak ground motions and MCS intensities for the Italian the earlier by & to earthquakes in Italy and it is not of the Italian In Margottini et al. the intensities were by the authors the data of the instrumental Italian earthquakes since The intensities were into and the (i.e. to the of the of from the the (i.e. are to the of the or to the of data from earthquakes the Margottini et al. data base. revision and of this data performed by Faccioli & Cauzzi who the with and it with other for a of earthquakes and data Although the adopted to instrumental and intensity data not by Faccioli & Cauzzi (i.e. between the and the intensity this data is the most and available for intensities in Italy. Recently, the of the Italian been made available et al. 2008). by earthquakes with a of the from to The to and the data the in Italy by Italian Nazionale Italian e Italian and and the Italian for additional in Italy, there is a large and macroseismic intensity data DBMI data (Stucchi et al. at with a (i.e. This database is a of all the macroseismic analysis for the Italian It a of observations from earthquakes at Although it is that local can the of the have made no to the to the since the intensity values reported in represent The reported intensities the MCS scale in by intensity (e.g. The to access and these two of data us the to a database of intensity and peak ground motion values To this have all the intensity data which are within from the that recorded the data. This performed for all the events within the distribution of the events and the of the earthquakes in the and intensity MCS have been for a of the distribution of the data the from the to the the database is note that there are intensity data at for intensity values (i.e. in the MCS This follows from the data being for events (i.e. earthquakes macroseismic the data not provide at intensity this is an of the data and to some extent it to the intensity values in of observed PGM the of the events and of the used to the data of the intensity data MCS and The is using the of the of this for large the some differences in the it is to the analysis The use of the is by the PGM data distribution the and logarithmic as in The distribution are also for and it is that the from the are not by a For and PGV the the are to the of the the great of the In the using the logarithmic with the distribution To the of the distribution have performed the We can the of a distribution for the and PGV with an the for and PGV with an to and This all that the data to be and be as in the are very to by & PGM data data and of the logarithm in and For intensity the data is to values, and standard To the of the distribution are also as In of intensity and PGM data are used to using The PGM values are using the The intensity standard have been to the of the PGM is from the standard The is the most for data to and for parameters of the The and of this is based on the important that the be to a much the It follows that this can be that the of can not be to the of We the data using using a and a the have obtained and with a standard of the of The however, to some between and high intensity values as in Wald et al. also & 2007), the data is into two and intensities or to The resulting from of using the are data of to this and for the data with intensity the parameters are The standard of the is MCS PGM for (i.e. the PGM data data and standard and The standard are on The and the are for is for PGA, for The of the of the standard with the when to that of the it a with two the standard to estimates for the are large to the that when attempting to with a the available data This be using The for has been also to The for the using data are and with a standard of the as The of the of the are and for the data with intensity the parameters are The standard of the is The values of the standard between and and the values of the of the the to be to the data The for and PGV in the some on which of the be have on the values of the standard to the To this regard, have the analysis using the standard of the in as and found in to in and studies to on data (e.g. Wald et al. 1999a; & an in at intensity data not to the same The for this however, to the differences of the MCS scale when to the MM other in the of intensities between and (e.g. Margottini et al. to of of the data To this have used data to the of of the observed data We the analysis to can be from In have two data of for a and a data The data consist of for intensity level of has been to the standard of the intensity The between and at The values of the for the in are and The data using the values for the in and These values are all to of the observed data. We to these data all the values (i.e. values as the data We first the of the estimates using the data have data using the values and the is to the of the parameters estimates using the data The the estimates to be and as for the estimates of the and In the obtained with the data that with a large data the same of data it be to estimate the in and the of the data distribution of the on data for the The values of and for the a and are as The the for are found also for however, to the of the estimates obtained with the observed data. the analysis using of the and data in of number of data for intensity that of the observed data We to these data (i.e. data as the data In an example, for of the data of the regressions for the and the data The has been for the data and for of the data from the data see that the for the data very used to the In this is not the when the data for of the for the data. The data at intensity data and data The data an of data that of the observed data (i.e. The the values from the adopted for on the and To the of estimates obtained with the observed to of the data to some To this the for on of the data To provide a of the to the the of the using In the distribution of the and of the data We note that the values the of the However, there is a for the distribution of the of for the and the The distribution is very the (i.e. the of the in the the of the regressions a much This is for the which on a very number of data at the intensity values and the of the estimates in the the distribution of the The the same as the for the In the distribution of the of the data for the the in the same of distribution for the data with the the has been to the distribution of the and standard We to the of the of the standard found when the intensity values using the to the observed data (i.e. when to that of the (i.e. of these data in fact lead to the that the observed values of the are To this have the of the standard from the data The are in We see first that the standard from the data through a not from that obtained from the data also through a second in The values obtained from the are very to found from the observed data. the standard obtained from through a the and the data also very values see in In this however, the values obtained from the analysis to some extent from the observed the still within the standard In not of that the observed standard is when using the and data not to between and Since this all follows also from the of the data it seems that of additional of in the (e.g. or most the of the analysis. between the intensity PGM regressions obtained using the and the and PGV in and For the regressions of Gómez Capera et al. Faccioli & Cauzzi Margottini et al. and Wald et al. are also and with the regressions obtained by Margottini et al. Faccioli & Cauzzi and Gómez Capera et al. for Italy, and the of Wald et al. in use in the of maps of shaking in Italy (Michelini et al. 2008). The as to are also in also the for PGV and the of the the with of Faccioli & Cauzzi and Wald et al. (1999a). In and in the of values that a between that found by Faccioli & Cauzzi (i.e. and obtained by Margottini et al. and Gómez Capera et al. (i.e. at intensities between and the in this the of Faccioli & Cauzzi and at intensities between and values to of Gómez Capera et al. (2007). For values between of Faccioli & Cauzzi that to the values of at and of Margottini et al. that, to the level of intensity at The observed differences can be by the data the of intensity values, the adopted to the intensity values with the recorded ground motion and by the For example, Faccioli & Cauzzi not the used and not on the and intensity scales and their data in this adopt a which into the in and and that the data Faccioli & Cauzzi in the of values and it is not to the obtained regressions because Faccioli & Cauzzi and Gómez Capera et al. their analysis to intensities of the main that this the of a MCS intensity scale which can be adopted in the USGS-ShakeMap (Wald et al. 1999b) for the Italian territory (Michelini et al. 2008) to provide MCS intensity maps In addition, calibration of the intensity conversion the to generate maps of PGM parameters and the very large intensity database for past earthquakes available in Italy (Stucchi et al. 2007). This is important when are made to provide estimates of the ground shaking of historical earthquakes on and or the of earthquake scenarios that use peak ground motion attenuation relations by observed data. In the conversion have Wald et al. first the instrumental intensity the and the instrumental intensity is adopt the derived intensity from This follows from the that strong are by and PGV to be a of intensity for strong shaking (Wald et al. To the of the regressions in this have and to the data of all the earthquakes with at instrumental used in this For the shakemaps that adopt the observed PGM data are to obtained conversion from to In the provide all the shakemaps in of MCS intensity and of and PGV for the earthquakes In the two and in and the shakemaps that are of the of These two earthquakes have been to to earthquakes of the seismicity in Italy. In earthquakes and are much earthquakes a century in and large number of In the intensity for the We see a between the strong motion data and the intensity derived maps of MCS intensity. The between the two maps in the level of local that on the number of The standard that on PGM data has been using data in the of and this in a much local shaking distribution when to that obtained using the much number of intensity data in the of In and the PGM data shakemaps with obtained the MCS intensities into PGM using the relations of this note a in the and PGV shakemaps obtained from the data and from the intensity to PGM This that the regressions found in this can be adopted to provide first maps of peak ground motion in these the level of local is by the of observations when using the PGM data in the standard ShakeMap of the earthquake in the in Italy. We have the derived from the and the Italian ground motions et al. with standard and shakemaps in of MCS shakemaps in of shakemaps in of PGV PGM and MCS intensity derived shakemaps are in the and The are the and intensity used as in the analysis. In the obtained for the main This earthquake very and The PGM and intensity derived shakemaps are there seems to be some of intensities with the PGM data derived in of the two maps are in of PGV the data derived has PGV values that using the of this that, to first the PGV obtained from the MCS intensities within the by a relationship using earthquakes throughout all Italy, a representation of the level of shaking in the These are by the maps in which an between the PGA, and PGV maps based either on instrumental or on macroseismic data. of the main in Italy. We have the derived from the and the and PGM see et al. with standard and as and in order to the differences between the shakemaps using recorded data and derived from the macroseismic using the relations found have the differences for all the shakemaps in and in and The used to the differences the of USGS-ShakeMap (e.g. Wald et al. et al. 2008) within a of from the for a of the for and The that the differences for all parameters are In particular, that of the intensity values are within For and that and of the values, within between MCS and PGV values from instrumental data and from macroseismic (e.g. using data are used in standard and values are for for PGA, and for have a correlation of the with and in analysis. To this have of the distribution as of and for all the data The in not to the of the of the data used in the analysis. of the of the intensities and The that no of the and for data In this have performed analysis between MCS intensities and recorded peak ground motion data in of and The data has been for earthquakes that have in Italy in the The work has been by the to represent intensities using the MCS scale within the implementation of ShakeMap for the Italian This between the shakemaps obtained from of the USGS-ShakeMap (Wald et al. et al. 2008) using observed PGM and the of ground motion shaking that on either ‘Did You Feel It’ analysis macroseismic data in (Stucchi et al. 2007). the intensity and the PGM data are by have adopted the which into the in and In order to the have to the data using the This is by the PGM data to a The data used in the analysis has been from two data database of the Italian strong motion et al. 2008) and the of Italy (Stucchi et al. 2007). of the data in data which are two to in studies for Italy. The that with the data available a is to the data two that for and high intensities of the the has been using for data the observed data. have the relations by in the USGS-ShakeMap in use at INGV (Michelini et al. 2008) to (i) the derived MCS intensity maps the reported macroseismic data and maps and (ii) the relations can be used to PGM maps which have found to be with from observed instrumental data. The analysis made on the shakemaps in this work to the of for intensity and PGM intensity. In addition, have that the found regressions not on either or that the obtained from of the regressions in this provide an representation of the level of ground shaking in of the adopted MCS intensity scale in Italy the regressions can be used to ground motions from intensity data This has been by the and by the Italian The authors are to and Stucchi for the of a of the We to and for the and We are also to and for their useful and for the analysis of the macroseismic data. be found in the of this used for the The is in and for the events have been derived from the and the Italian ground motions et al. 2008). For the shakemaps are in of MCS and PGV PGM and MCS intensity data derived shakemaps are in the and The represent the strong motion and the intensity used as for the analysis. are not for the or of by the be to the for the In this some that be into when data of data at In the the topic of and on some that have found of great interest when the analysis of this In particular, have found of importance (i) the for a (i.e. between the two of data is (ii) the of the for and (iii) the data the data. We by the In general, are used in since are used to a in of a or of a and are a relationship can be used to the the of the for a that there is and that the of a is based on & The values have For there is a distribution of is for The distribution of for has the same that that the the is the same to the of The values of for on a the first it is that this is not that have which is not the This the use of a allows for the of in the the analysis and (e.g. & the analysis are by as correlation The of a correlation and For of there is a distribution of and for of there is a distribution of The have the same the have the same The distribution of and is the The these The to the importance of data the analysis. We note that in the it be found a on a standard to to a data For example, some authors the since it of (e.g. use (e.g. To these on the have performed a an data the same of data (i.e. with and of data The data to a with values at the and for For of the are for a of For the were at the and values of and are the data are the values for of the is the is the in the is the with much in the in the is the with much in the in the the of the analysis when no data is (i.e. data data is on the and In the of the work the of the of the of the of using the data the of the of the of the of using the data The and is to the to The differences observed between the and the used to generate the data are to be to the the data of the have been to the and The of to the and the data The regressions are the in in the with much for the in the in the with much in the in the in the aim of this is to (i) the of in (ii) the of the to in the (iii) the and of the on the data the of analysis are to the data have found that all provide to the data of the of the and and regressions are that the in is the of is The when the data (i.e. is used for the the is used as the some on and the in the not when the is used as the the in This is not since the the the is not have found that caution be in the of the to the when the is to the data the of the of the in and the analysis some in and In and the by analysis (i.e. a and by data (i.e. a of and the of this that the to be adopted of using the data and the

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.