Since publication, we have received a number of questions about AFG which we answer below (also on DOI:10.5281/zenodo.21596364). If you have and further questions, please contact us.
Fault traces are mapped in AFG from attributed geomorphological indicators at 1:25,000 scale, using accurate digital elevation models (vertical error ≤ 2m; Hellenic Cadastre, 2021), together with the mapping workflow presented in ‘Methods’. Positional uncertainty of mapped traces is directly tied to their geomorphic expression: 'sharp' and 'moderate' traces (62% of the active-fault population) are located to within 100 m of the fault itself, while 'rounded' and 'poor' traces have an estimated positional uncertainty of 100–1000 m (Begg et al. 2025, Table 2).
Each of the seven geomorphic criteria stated in the Data Descriptor’s Methods and used with high resolution DEMs, demonstrate tectonic origin. AFG's methodology explicitly addresses the risk that a mapped lineament reflects a stratigraphic or erosional origin, rather than tectonic activity, through criteria (e) and (f) (section ‘Fault trace mapping’). AFG traces are typically supported by more than one criterion.
The entry of Begg et al. (2025) on each trace in the AFG attribute table indicates, as explained in the Data Descriptor, that the trace has been remapped and may not be co-located with previously mapped traces that represent the same feature. Hence, for all previously existing fault traces which are mapped here, attribution includes citation of previous work (even in cases where faults are located proximal to remapped traces), but responsibility for changes in the expression/location of these faults is recognized by the attribution 'Begg et al. (2025)'. Further, new fault traces mapped here for the first time (n>2000), are also attributed with ‘Begg et al. (2025)’.
Within its attribute table, AFG cites all published peer-reviewed and easily accessible literature associated with each fault trace. Differences in fault locality, length, activity, or (in a few cases) dip direction from original publications, are due to remapping of these faults (i.e. and not from literature ‘misuse’). A list of references used in compiling the attribute table is available in the supplementary material of Begg et al. (2025).
Begg et al. (2025) explain that there are no nationwide stratigraphic markers within the Quaternary across onshore Greece, making conventional definitions of active faults (often formed in the last 125 thousand years) impossible. Their definition of “active” faults differs from that of other authors and countries, in defining faults with traces across Quaternary sediments (2.58 million years) as active. This definition must be kept in mind in using the database.
AFG estimate fault lengths from attributed geomorphic features, accurately located on DEMs. Where AFG fault lengths are greater than previously estimated, we believe that this primarily reflects the high resolution of the DEM used here (relative to previous studies). Nevertheless, mapped AFG lengths should be treated as minima (i.e. higher resolution imagery may be able to extend them in the future).
AFG is not suitable for direct input to SHA. AFG provides to date the most complete and homogeneous assessment of Greece's onshore active faults (at the time of publication). The AFG can be used to inform the location of earthquake source faults and, therefore, provides a foundation for national and regional SHA. Seismic hazard modelers should evaluate the hazard after careful assessment and further quantification of the faults’ earthquake parameters.
AFG maps fault traces, individual faults and fault systems recorded on Greece’s landscape from repeated earthquake-slip, however, we do not assign earthquake magnitudes from these structures. Careful assessment of historical and paleoearthquake rupture records are required to determine the relationships between fault lengths and earthquake magnitudes.
AFG reproduces the traces of all historical earthquakes that ruptured the ground surface in onshore Greece from georeferenced published maps which are tested against the geomorphological record available from DEMs. In most cases, DEMs locate more accurately surface traces than geological maps, although the length of historical surface ruptures is best constrained by published information.
Historical earthquakes demonstrate that individual earthquakes can rupture more than one fault, fault section and/or fault segment. Therefore, individual short traces (i.e., <5 km) do not necessarily provide information about earthquake magnitudes. Discontinuous fault rupture traces are documented in at least ten historical Greek earthquakes (i.e. Palyvos et al. 2010; Mouslopoulou et al. 2022), in which multiple traces shorter than 1 km ruptured in a single event, and in large-magnitude earthquakes internationally (i.e. the ~M6-6.5 2016 earthquakes in the Apennines, the M7.8 2016 Kaikōura Earthquake; the M7.6 2023 Kahramanmaraş Earthquake), where numerous traces under 5 km were mapped as part of a single multi-fault rupture (e.g. Litchfield et al. 2018; Improta et al. 2019; Jia et al. 2023). Often, these surface ruptures were also characterized by disparate strikes. Hence, small-sized individual fault traces recorded in AFG are consistent with available empirical earthquake records and well-documented surface ruptures globally.
Fault traces mapped in AFG represent earthquake-related ground surface displacements. Not all traces assigned to a fault need rupture in any one earthquake, and traces assigned to different faults may rupture during multi-fault earthquakes.
The AFG presents a digital database of onshore active faults in Greece (and of their historic ruptures) based on analysis and quantification of DEMs, NOAFaults is a digital active fault repository containing the geometry, slip rates, and historical seismicity of onshore and offshore active faults in Greece and GreDaSS is a Greek Database of Seismogenic Sources. While these three databases are all slightly different representations of active faulting in Greece, they depict a broadly consistent national tectonic framework of active faulting. Once offshore structures are removed, the majority (80-90%) of faults and seismic-sources, respectively, included in NOAFaults and GreDaSS appear to have a geomorphological counterpart in AFG. Many of the additional fault traces represented in AFG (45% and 65% more faults than NOAFaults and GreDaSS, respectively), reflect increased mapping resolution, increased segmentation and new interpretation of geomorphological evidence, rather than a departure from the first-order structural framework represented in NOAFaults and GreDaSS. Hence, the principal differences between the datasets lie primarily in spatial detail and completeness of mapped fault traces, rather than in fundamentally different tectonic interpretations.
These terms are conventions that aim to discriminate between scarps generated along a fault due to its most recent activity as a result of one or more (i.e. <10) large-magnitude earthquakes, and larger scarps that have resulted from repeated slip due to many (e.g. 50, 1000 or 10,000) earthquakes over an extended period of time. Based on empirical observations, the former scarps will range from centimeters to a few tens of meters (e.g. 10, 20 or 30 m), while the latter will be larger (100 m to kilometers). These differences are defined in Begg et al. (2025). While there is no strict threshold assigned to this conceptual approach, their discrimination may help focus future paleoseismic studies.
As the most generalized expression of fault hierarchy in AFG, the faults grouped within fault systems are believed to reflect structures which share geometric and/or kinematic characteristics. The authors make no suggestion that fault systems rupture in any single earthquake. Caution should be applied where any AFG data is used for derivative purposes.
Yes, all faults in AFG have geomorphic expression. This is the basis on which they were mapped (see Figures 5 & 7 in Begg et al. 2025). However, we did not assign a ‘trace_character’ on the ‘probably active’ faults (as we did for active and historically active faults), because they are less likely to be the focus of future paleoseismic work or SHA. The main reason for this attribution is to help future workers focus their work on the traces which are best represented in the landscape (and which are more likely to have ruptured in the recent past).
Yes. Complex fault geometries and fault intersections are mapped in AFG across Greece. Fault intersections may be acute, obtuse or even orthogonal. Orthogonal or obtuse AFG cross-cutting fault relationships are consistent with independently documented orthogonal simultaneously active faulting worldwide (i.e. Mouslopoulou et al. 2007; AIST, 2012; Langridge et al. 2016; Brozzetti et al. 2017; Feng et al. 2020; Plesch et al. 2026) and with globally documented multi-fault ruptures on non-optimally oriented intersecting faults (e.g. the 2016 M7.8 Kaikōura Earthquake, the M6-6.5 Norcia Earthquake, the M7.1 2010 Darfield Earthquake, the 2019 Ridgecrest Earthquakes).
AFG contains only onshore faults that exhibit geomorphic expression. About 50% of AFG faults are mapped here for the first time; hence, they represent faults which were not previously recognized. All new faults are mapped in AFG from attributed geomorphological indicators and their strike/dip-direction may or may not differ from neighbouring, previously recognized, faults. AFG does not intentionally include faults that are not active. However, one of the categories of ‘activity’ in the attribute table is “uncertain” (n=64), and some of these faults may not be active. It is also possible that some faults attributed “probably active” may not be active, though this is less likely. In contrast, we anticipate that future investigations of individual fault traces designated as “probably active” may change their attribution to “active”. No major faults that penetrate to the surface are missing from the AFG database. However, there may be cases, particularly in the Pindos ranges where faulting strikes subparallel to bedding/foliation, that some faults that displace the ground surface may not be represented.
AFG maps and characterizes fault displacements visible at the earth’s surface only. Subsurface faults are not represented in AFG with the single exception of a concealed fault that ruptured historically in Peloponnese (2008) and was identified through its aftershock sequence (Fault ID 453).
AFG is fully peer-reviewed. Both the Data Descriptor (including Methodology, Data Records and Technical Validation), as well as the database itself, have been peer-reviewed at the highest academic standards and published in Nature Scientific Data. No other available national active fault database has been peer-reviewed.
Yes, as with all large databases, AFG may contain errors. For example, we noticed that two traces of the Kalamata surface rupture (ID=2460 and ID=2102) are attributed to dip the wrong way. We also noticed that some traces have been wrongly attributed to ‘Basili et al. 2013’ instead of ‘Zelenin et al. 2022’ (due to a blanket GIS replacement in an effort to cite the earlier database). Such mistakes, analogous to typographical or transcription errors, will be corrected in future releases. If you identify any errors in the AFG attribute table, please contact us!
AFG provides the most accurate location to date (at 1:25,000 scale) of faults that have ruptured the ground surface in Greece over the last 2.58 million years. Hence, AFG provides reliable and conservative data with which to proceed. The question that remains for end users is whether it is important to investigate these features to determine whether they represent a credible risk. Risk is properly addressed by the hazard represented by the earthquake recurrence interval (RI) on a fault multiplied by the cumulative value of the assets involved. To assess seismic hazard, paleoseismological investigations on active faults are amongst the most useful techniques.
AIST - Active Fault Database of Japan, 2012 version (last update 2016). Research Information Database DB095, National Institute of Advanced Industrial Science and Technology (AIST). https://gbank.gsj.jp/activefault/index_e_gmap.html.
Basili, R., Kastelic, V., Petricca, P., Tarabusi, G., Tiberti, M., & Valensise, G. (2013). The European Database of Seismogenic Faults (http://diss.rm.ingv.it/share-edsf/). On-line database https://doi.org/10.6092/INGV.IT-SHARE-EDSF.
Begg, J.G., Mouslopoulou, V., Heron, D., Nicol, A. (2025). AFG - Active Faults Greece: a comprehensive geomorphology-based 1:25,000 fault database. Scientific Data 12, 1853. https://doi.org/10.1038/s41597-025-06283-z.
Brozzetti F, et al. (2017). Newly identified active faults in the Pollino seismic gap, southern Italy, and their seismotectonic significance. Journal of Structural Geololgy 94, 13-31.
Feng, W., Samsonov, S., Qiu, Q., Wang, Y., Zhang, P., Li, T., & Zheng, W. (2020). Orthogonal fault rupture and rapid postseismic deformation following 2019 Ridgecrest, California, earthquake sequence revealed from geodetic observations. Geophysical Research Letters, 47, e2019GL086888. https://doi.org/10.1029/2019GL086888.
Hellenic Cadastre (Ktimatologio S.A.), 2021. Digital Elevation Model – DEM – LSO25 [Data set]. Hellenic Cadastre. Retrieved from https://data.ktimatologio.gr/sites/default/files/2021-10/lso25.xml.
Improta, L., Latorre, D., Margheriti, L., Nardi, A., Marchetti, A., Lombardi, A. M., et al. (2019). Multi-segment rupture of the 2016 Amatrice-Visso-Norcia seismic sequence (central Italy) constrained by the first high-quality catalog of Early Aftershocks. Scientific Reports, 9(1), 6921. https://doi.org/10.1038/s41598-019-43393-2.
Jia, Z. et al. (2023). The complex dynamics of the 2023 Kahramanmaraş, Turkey, Mw 7.8–7.7 earthquake doublet. Science 381, 985–990.
Langridge R, Ries W, Litchfield N, Villamor P, Van Dissen R, Barrell D, Rattenbury M, Heron D, Haubrock S, Townsend D, et al. (2016). The New Zealand active faults database. New Zealand Journal of Geology and Geophysics 59, 86– 96. https://doi.org/10.1080/00288306.2015.1112818.
Litchfield, N.J., Villamor, P., Van Dissen, R.J., Nicol, A., Barnes, P.M., et al. (2018). Surface Rupture of Multiple Crustal Faults in the 2016 Mw 7.8 Kaikōura, New Zealand. Bulletin of the Seismological Society of America 108(3B), 1496-1520.
Mouslopoulou, V., Nicol, A., Little, T.A., Walsh, J.J., 2007. Displacement transfer between intersecting regional strike-slip and extensional fault systems. Journal of Structural Geology 29, 100-116, https://doi.org/10.1016/j.jsg.2006.08.002.
Mouslopoulou, V., Sudhaus, H., Konstantinou, K. I., Begg, J., Saltogianni, V., Männel, B., Antinisari R., Oncken, O., 2022. A deeper look into the 2021 Tyrnavos Earthquake Sequence (TES) reveals coseismic breaching of an unrecognized large-scale fault relay zone in continental Greece. Tectonics, 41, e2022TC007453. https://doi.org/10.1029/2022TC007453.
Palyvos, N., K. Pavlopoulos, E. Froussou, H. Kranis, K. Pustovoytov, S. L. Forman, and D. Minos‐Minopoulos (2010). Paleoseismological investigation of the oblique‐normal Ekkara ground rupture zone accompanying the M 6.7–7.0 earthquake on 30 April 1954 in Thessaly, Greece: Archaeological and geochronological constraints on ground rupture recurrence, Journal of Geophysical Research 115, B06301, https://doi.org/10.1029/2009JB006374.
Plesch, A., Marshall, S.T., Shaw, J.H. (2026). The Community Fault Model (v. 6.1) for Southern California. Bulletin of the Seismological Society of America, 116 (3): 1103–1124. doi: https://doi.org/10.1785/0120250247.
Zelenin, E.A., Bachmanov, D.M., Garipova, S.T., Trifonov, V.G., & Kozhurin, A.I. 2022. The Active Faults of Eurasia Database (AFEAD): the ontology and design behind the continental-scale dataset. Earth System Science Data, 14, 4489–4503.