UNCORRECTED PROOF Seismic hazard in regions of present day low seismic activity: uncertainties in the paleoseismic investigations along the Bree Fault Scarp (Roer Graben, Belgium) K. Atakan a, * , V. Midzi a , B. Moreno Toiran a , K. Vanneste b , T. Camelbeeck b , M. Meghraoui c a Institute of Solid Earth Physics, University of Bergen, Alle Âgt.41, N-5007 Bergen, Norway b Royal Observatory of Belgium, Avenue Circulaire 3, B-1180 Bruxelles, Belgium c EOST, Institut de Physique du Globe, 5, rue Rene  Descartes, F-67084 Strasbourg cedex, France Abstract Earthquake hazard assessment in stable continental regions, such as northern Europe, has traditionally been evaluated on the basis of the instrumentally and historically recorded seismicity, which indicates relatively low hazard levels. Reliability of such estimates is a matter of debate as the long-term potential of large earthquakes usually cannot be determined based on short observational periods generally less than a few hundred years. A signi®cant improvement to this lack of knowledge can be achieved by extending the past observations into the geological time scale. Paleoseismic investigations can provide valuable information to bridge this gap, where the potential for large earth- quakes can be quanti®ed both in magnitude and recurrence period, based on the observation of prehistoric earthquakes (paleoearthquakes) in the geological record (particularly in the last 20,000 years). However, using these records in seismic hazard analysis requires systematic treatment of uncertainties. Usually uncertainties are inherent to the interpretation of geological record, which leads, in the end, to the identi®cation of paleoearthquakes. Field observations used in the analysis may satisfy several alternative interpretations. Such interpretations become useless when alternative solutions exist but not documented in detail, and especially when the relative reliability of the favored interpretation with respect to the alternative interpretations is not known. The recently introduced method using logic-tree formalism, which is based on qualitative description of the uncertainties related to the paleoseismic data and especially in its interpretation, is applied in the paleoseismic investigations performed on the Bree Fault Scarp, along the Feldbiss Fault (Roer Graben, Belgium). The cumulative uncer- tainties associated with the different stages of the study are computed as the combination of the preferred alternative branches in the logic- tree presentation. The ®nal uncertainty and its relative importance in seismic hazard analysis is expressed as the paleoseismic quality factor (PQF), which indicate 0.76. This value can directly be used in seismic hazard analysis. q 2000 Elsevier Science Ltd. All rights reserved. 1. Introduction Paleoseismology is a rapidly growing ®eld since its recognition as a distinct discipline in late 1960s and early 1970s (e.g. [11,24,25,28]). Recent developments in paleo- seismology (e.g. [13,17,19±21,23,26]) and the importance of the data provided by paleoseismological studies in seis- mic hazard analysis increase the need for systematic treat- ment of uncertainties. Uncertainties are inherent in the interpretation of geological phenomena, where ®eld obser- vations may satisfy several alternatives. Quanti®cation of uncertainties related to paleoseismological data in seismic hazard analysis is dif®cult. Unless documented in detail by the scientist providing such data, using it in seismic hazard analysis may lead to a misinterpretation of the true seismic hazard of an area of interest. The relative reliability of a favored interpretation with respect to the alternatives is rarely quanti®ed. In this paper, a qualitative method of describing uncertainties related to interpretation of paleo- seismological data is proposed. The method is illustrated through a simple example using logic-tree formalism applied to the paleoseismological data interpretation process. 2. Stages of interpretation The main objective of the paleoseismological analysis is important to specify before the treatment of uncertainties. Once this is done, the consecutive stages of analysis can be identi®ed. These different stages may then be integrated into a logic-tree as different nodes with alternative branches. At each node, different alternatives can be described with their associated uncertainties. These uncertainties can be expressed in terms of probabilities assigned to each branch Soil Dynamics and Earthquake Engineering 00 (2000) 000±000 SDEE2498 0267-7261/00/$ - see front matter q 2000 Elsevier Science Ltd. All rights reserved. PII: S0267-7261(00)00081-6 www.elsevier.com/locate/soildyn * Corresponding author. Tel.: 147-55-58-3420; fax: 147-55-58-9669. E-mail address: atakan@ifjf.uib.no (K. Atakan). Soil Dynamics and Earthquake Engineering ± Model 5 ± Ref style 3 ± AUTOPAGINATION 2 29-11-2000 14:16 Sd6009 PH A lden