Exploring representativeness and reliability for late medieval earthquakes in Europe
Seismic catalogues of past earthquakes have compiled a substantial amount of information about historical seismicity for Europe and the Mediterranean. Using two of the most recent European seismic databases (AHEAD and EMEC), this paper employs GIS spatial analysis (kernel density estimation) to explore the representativeness and reliability of data captured for late medieval earthquakes. We identify those regions where the occurrence of earthquakes is significantly higher or lower than expected values and investigate possible reasons for these discrepancies. The nature of the seismic events themselves, the methodology employed during catalogue compilation and the availability of medieval written records are all briefly explored.
KeywordsEarthquakes Historical seismicity Late medieval Europe GIS Kernel density estimation Risk
A key initiative in historical seismology in recent years has been the collection of earthquake data at a continental scale, especially for Europe. AHEAD (Archive of Historical Earthquake Data; Locati et al. 2014; http://www.emidius.eu/ahead/main/) and SHEEC (SHARE European Earthquake Catalogue 1000–1899; Stucchi et al. 2013; http://www.emidius.eu/SHEEC/sheec_1000_1899.html) have developed systematic catalogues of past seismic events between AD 1000 and 1899, generating and publishing a robust archive of macroseismic information. A third project, EMEC (the European–Mediterranean Earthquake Catalogue; Grünthal and Wahlström 2012), consists of a unified catalogue of earthquakes with an Mw higher than 3.5 in Europe, Mediterranean Africa, Turkey and Cyprus up to 2006. In contrast to the other two catalogues, EMEC is mainly based on instrumental recording of recent seismic events.
These metadata archives are primarily intended as inputs into the seismic hazard assessments which remain fundamental to the development of strategies for earthquake risk reduction (through, in the case of the SHARE project, Seismic Hazard Harmonization in Europe). The valuable contribution of historical seismology to national and international earthquake catalogues has long been recognised (Ambraseys 1971; Gürpinar 1989; Vogt 1991; Caputo and Helly 2008), but these new larger databases now open up fresh possibilities for research. Not only do they add a remarkable volume of data which has been standardised according to published criteria, but both AHEAD (http://www.emidius.eu/ahead/main/) and EMEC (http://emec.gfz-potsdam.de) also operate on open-access online platforms and embed useful tools for geographical and chronological interrogation.
For more than a generation, research tools have been available to seismologists to help evaluate the completeness of historical earthquake catalogues as time series data (e.g. Stepp 1972; Weichert 1980; Woessner and Wiemer 2005; Hakimhashemi and Grünthal 2012; Alamilla et al. 2014), but it is now possible to supply a spatial as well as a chronological assessment of past events. In this paper, therefore we explore the use of kernel density estimation (KDE) to investigate the representativeness of the historical seismic activity in Europe in the late Middle Ages (here defined as AD 1000–1550) from a geographical perspective. We identify those European regions where our knowledge of medieval seismicity is especially weak, and we ask whether medieval seismicity is sometimes overestimated.
2 The earthquake record over time
The reasons for this disparity are also well rehearsed (Guidoboni and Ebel 2009, for example). They include the comprehensiveness and reliability of any individual account of a historical seismic event, the preservation and transmission of that record (which may be one of a number which provide evidence of a single event) and the capacity of modern compilers and analysts to recognise and catalogue the event. When researchers claim that ‘libraries may hide hundreds of treasures that are mostly unknown to seismologists’(Vogt 1991), they concede the degree to which research intensity varies across European regions. In short, it cannot be assumed that current catalogues, vital though they may be, are homogeneous in their representation of past seismicity. The key question to ask is precisely where the strengths and weaknesses of the data might lie.
3 KDE analysis
In order to obtain a mean to compare earthquake distributions over time, KDE analysis was undertaken for selected datasets showing medieval earthquakes (AD 1000–1550), post-medieval or early modern historical earthquakes (AD 1551–1899), and twentieth-century (AD 1900–1999) earthquakes. Calculations were applied both to the entire number of the recorded earthquakes collected within each dataset and for to earthquakes with Mw ≥ 5, introducing a threshold which excludes events that cause little damage. The KDE maps are then displayed using a coloured key which defines density trends. To avoid redundancies during comparison, the density values in each case were homogenised to a range of values ranging from 0 to 100. The mean values of density were extracted from the maps using a zonal statistic analysis and assigned to a shape file displaying the provinces (1248 in total) of all European countries. This allows differences in mean value density to be calculated and then displayed.
By comparing several datasets, Fig. 8 highlights the extent to which the recorded distribution of twentieth-century seismic events differs from those that occurred in the Middle Ages. Once more, it is possible to identify where higher and lower than anticipated levels of activity are located. Negative values (in blue on Fig. 8), which indicate lower than anticipated levels of medieval seismicity, are focused on two areas: eastern Europe and the eastern Mediterranean, including the Balkans, Romania, Greece and Crete, and Iceland. The most under-represented areas lie in the south of Albania, around the Gulf of Corinth and Crete. By contrast, those regions showing a higher-than-expected level of medieval seismicity (in red on Fig. 8), when compared to contemporary seismicity, can be found in Western Europe, especially in Andalusia, the eastern Pyrenees, Switzerland, the Aachen region, northern and central Italy, the Strait of Messina, Slovenia and Dalmatia. Peaks in positive values centre on northern and central Italy.
The higher-than-expected spatial density of earthquakes may in part be explained by the nature of the seismic event itself. While it could be argued that, over a period of a century, the spatial and temporal distribution of small to medium earthquakes on a continental scale might be approximately constant, the recurrence interval of seismic events scales with time, such that the largest earthquakes occur the least frequently; the recurrence period of a large earthquake on a given active fault might be typically in the order of a century to a few millennia. The largest earthquakes, because they have much longer return periods, introduce greater temporal and spatial variability. Once more, there may be a high occurrence of aftershocks after a very large seismic event and, where a large earthquake has occurred, it would be expected that a number of small to medium earthquakes might also strike in the same region.
It is also the case that two earthquakes of the same magnitude may not have the same consequences for above-ground structures because of the nature of local geology and geomorphology. For example, variations in rupture speed may affect the frequency of the shaking experienced at ground level, changing the damage potential of the earthquake. In addition, different continental areas have different attenuation characteristics which affect the distribution of ground shaking. In central Greece, for example, strong earthquakes have been described with a Mw between 6.5 and 7.2 but with only very localised impacts (Ambraseys and Jackson 1990; Stiros and Pytharouli 2014). Another important influence on our mapping is the method by which events have been recorded by catalogue compilers. The observed peaks of post-medieval earthquakes in Switzerland and Slovenia, for example, are probably due to the recording of a large number of low-impact aftershocks as independent earthquake events in these regions and also to some extent the comprehensive research which has been undertaken by the Swiss Seismological Service (Fäh et al. 2011; Živčić 2009, as reported in Stucchi et al. 2013: 533).
This paper highlights some of the strengths and weakness of current historic earthquake meta-datasets. While seismologists have long been aware of the incompleteness of their catalogues, we offer this KDE comparison as another tool in the toolbox, one that provides better geographical definition. The results immediately suggests an agenda for further investigation, particularly across eastern Europe and the eastern Mediterranean where our methodology suggests that there were more and more powerful seismic events during the Middle Ages than have hitherto been recorded. For some of these areas, archaeoseismological and palaeosesimological projects might shed new light on historical seismic events; otherwise, a more detailed assessment is required of the information gap resulting from a scarcity of written documents. Finally, we also highlight here the issue of over-recording, something which may be explained by the nature of the seismic event and the density of human settlement combined with regional cultural and social factors, including the more sophisticated development of risk-sensitive tactics.
This analysis falls within the ArMedEa project (archaeology of medieval earthquakes in Europe; Forlin, Gerrard, and Petley 2015), which aims to build upon the dataset published by AHEAD and survey archaeological and palaeoenvironmental evidence related to late medieval seismicity in Europe. ArMedEa was developed in the Department of Archaeology and the Institute of Hazard, Risk and Resilience of Durham University and supported by a Marie Curie Intra European Fellowship within the 7th European Community Framework Programme. Petley’s contribution was funded by the NERC/ESRC Earthquakes Without Frontiers project, grant reference NE/J01995X/1. The authors would like to thank the anonymous reviewers for their constructive comments.
- Fäh D, Giardini D, Kästli P, Deichmann N et al (2011) ECOS-09 Earthquake Catalogue of Switzerland Release 2011 Report and Database. Public catalogue, 17. 4. 2011. Swiss Seismological Service ETH Zurich, Report SED/RISK/R/001/20110417Google Scholar
- Figliuolo B (1988) Il terremoto del 1456. Studi storici meridionali, Altavilla SilentinaGoogle Scholar
- Figliuolo B (2002) Il fenomeno sismico nel bacino del Mediterraneo in età rinascimentale. Stud Stor 43:881–919Google Scholar
- Guidoboni E, Comastri A (2005) Catalogue of earthquakes and tsunamis in the Mediterranean area from the 11th century to the 15th century, BolognaGoogle Scholar
- Guidoboni E, Ebel JE (2009) Earthquakes and tsunamis in the past. A guide to techniques in historical seismology. Cambridge University Press, CambridgeGoogle Scholar
- Gürpinar A (1989) Importance of historical earthquakes in research in the sitting of nuclear facilities. In: Workshop “Historical earthquakes in the Ibero-Maghrebian region. Methodological approach and case studies”, Lisbon, 15–17 November 1989, Proceedings. pp 3–7Google Scholar
- Illian J, Penttinen A, Stoyan H, Stoyan D (2008) Statistical analysis and modelling of spatial point patterns. John Wiley & Sons, ChichesterGoogle Scholar
- Jotischky A, Hull C (2005) The Penguin historical atlas of the medieval world. Penguin, LondonGoogle Scholar
- Meletti C, Patacca E, Scandone P, Figliuolo B (1988) Il terremoto del 1456 e la sua interpretazione nel quadro sismotettonico dell’Appennino meridionale. In: Figliuolo B (ed) Il terremoto del 1456. Studi storici meridionali, Altavilla Silentina, pp 71–108Google Scholar
- Stepp JC (1972) Analysis of completeness of the earthquake sample in the Puget sound area and its effect on statistical estimates of earthquake hazard, first international conference on microzonation, vol 2. Washington University, Seattle, pp 897–909Google Scholar
- Tsougarakis D, Angelomatis-Tsougarakis H (2012) The archaeologist and the historian: methodological problems faced by historians participating in archaeological surveys. In: Bintliff J, Stöger H (eds) Medieval and post-medieval Greece. The Corfu papers. Archeopress, Oxford, pp 67–72Google Scholar
- University of Thessaloniki (2003) Macroseismic Data used for the compilation of Papazachos and Papazachou (2003) catalogue. In: Online Macroseismic Data of Southern Balkan area, http://www.itsak.gr/en/db/data/macroseimic_data
- Weichert DH (1980) Estimation of the earthquake recurrence parameters for unequal observation periods for different magnitudes. Bull Seismol Soc Am 70:1337–1346Google Scholar
- Živčić M (2009) Earthquake catalogue of Slovenia. http://gis.arso.gov.si/atlaskolja/profile.aspx?
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.