Determination of geomechanical site effects in France from macroseismic intensities and reliability of macroseismic magnitude of historical events

Results from a detailed analysis of the French macroseismic database SIRENE 96 show that geomechanical site effects (i.e. site effects due to superfici al geology) can be statistically detected and quantified from macroseismic data. Our results are used to derive a predictive map of potential site eff ects in France and to check if site effects induce a significant overestima tion of macroseismic magnitude of historical events. The key hypothesis of our work is that geological formations of similar age produce similar site effects because one expects the younger sediments to be less compacted, and therefore more prone to cause site effects. A geolog ical formation is associated with each intensity using a Geographic Informa tion System (GIS) and the digital geological map of France. Two approaches are developed in order to characterise the intensity increase associated wi th geological formations having the same age. The first approach is based o n the intensities that differ significantly from the average pattern, inten sities which are labelled as anomalies. In the second approach, a theoretic al intensity attenuation law is computed for each selected event only from observations located on formations older than 2 My. According to the result s of the first approach, these formations are unlikely to cause site effect s, and for the sake of simplicity they are labelled as rocks (younger forma tions are defined as sediments). The derived attenuation laws are then used to compute the residuals for each observation. Residuals are classified as a function of the age of the formations on which they are located. The res ults of the two approaches are in good agreement. No intensity amplificatio n is detected for formations older than 1 My. For younger formations, the y ounger the formations, the larger the average intensity increase. The avera ge intensity increase reaches 0.6+/-0.2 degree on the MSK scale for 0 My se diments. The results are better expressed in terms of probability: for 0 My sediments, there is an estimated 83% probability of observing an intensity increase; for 0.01 My and 0.75 My sediments, this probability is about 52% . The magnitude of the intensity increase is variable, but it is generally small: it has a 26% probability to reach I degree or above on the MSK scale for 0 My sediments and this probability falls to about 5% for 0.01 and 0.7 5 My formations. A predictive map of geomechanical site effects is produced by extrapolating the characteristics of intensity increases to all geologic formations of s imilar age in France. Glacial and fluvial valleys, already known as being s ubject to site effects, clearly appear on this map, as well as less expecte d areas such as the South Atlantic coast. Lastly, new estimations of macros eismic magnitude that exclude intensities potentially affected by geomechan ical site effects are realised. A magnitude overestimation due to site effe cts can possibly exist, but it remains limited (<0.2). This result tends to demonstrate that geomechanical site effects in France do not significantly distort our knowledge of historical seismicity. (C) 2000 Elsevier Science B.V. All rights reserved.