the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 05 Mar 2026
Sequential approach to seismotectonic zonation for South-East France
Abstract. The south-east of France, encompassing the western Alps, the Jura Mountain range, the Rhône valley and the Provence region, is the most seismically active region in metropolitan France and, consequently, one of the most extensively surveyed and studied. However, seismicity remains low to moderate (less than 10 Mw 5 and 1 Mw 6 events per century since 1300 CE) and geodetic deformation rates appear relatively low (< 20 nanostrain yr-1). The resulting low-signal-to-noise ratios, together with the complex, dense fault networks inherited from a polyphased tectonic history, make this region particularly challenging for seismic hazard assessment. The geophysical and geological data available are extensive, yet inhomogeneous and insufficient to confidently characterize active faults, quantify on-fault deformation and associate seismic rates. Therefore, seismic source characterization through seismogenic area models is commonly adopted. These models are however highly sensitive to the data used to describe seismotectonic behavior. Our objective is to consider newly available geophysical data as complementary constraints to geological observations to further refine seismotectonic zonation models.
We present an innovative sequenced zoning methodology that disaggregates seismotectonic behavior into three components (namely – crustal structure, observed seismicity and surface deformation) each analyzing several key features. We thus derive three novel, independent seismotectonic zonation models, each representing a different perspective on the seismogenic process. Additionally, we associate confidence levels with zone limits to each subsequent zonation model, by assessing feature homogeneity among neighboring zones. Afterwards, we propose a synthetic model which integrates all seismotectonic features by merging the most recurrent and highest confidence zone limits from the three independent zonation models. This approach intends to minimize zone mapping uncertainties by quantitatively assessing seismotectonic observations, and to yield reproducible and updatable models representative of the current state of seismotectonic knowledge. We subsequently compare the resulting zonation models and discuss their implications for seismic source characterization.
- Preprint
(4686 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-623', Francesco Iezzi, 13 Apr 2026
-
RC2: 'Comment on egusphere-2026-623', Francesco Visini, 30 Sep 2026
This manuscript presents a comprehensive analysis of seismotectonic zonation in south-eastern France. The authors combine geological, structural, seismic and geodetic observations to develop three independent zonation models, based respectively on crustal structure, observed seismicity and surface deformation. These models are subsequently integrated into a unified seismotectonic zonation model.
The topic is highly relevant to the scope of NHESS, particularly because seismotectonic zonation is a fundamental component of seismic source characterization and probabilistic seismic hazard assessment.
The manuscript addresses a challenging low-to-moderate seismicity region, where the available observations are heterogeneous, the observation windows are relatively short, and the active-fault framework remains incompletely constrained. A particularly valuable aspect of the study is the attempt to make the construction and hierarchization of seismotectonic zones more objective, transparent and reproducible.
The proposed confidence levels for zone limits represent an important effort to formalize expert judgement and to quantify the contrast between neighbouring zones. The separation of structural, seismicity-based and deformation-based information is also useful, as it allows the authors to preserve the different physical meanings and temporal scales of the available datasets. In this respect, the manuscript provides a potentially valuable framework for future seismic source models and logic-tree applications.
The manuscript is well structured and contains a large amount of relevant information. Some methodological choices could perhaps benefit from further explanation, and the robustness and limitations of the proposed confidence levels could be discussed more explicitly. I would therefore be inclined to recommend publication after minor to moderate revision.
I hope the authors will find the following comments helpful in further strengthening the impact of their work.
Main comments.
- Justification of dataset weights: The weights assigned to the different datasets in Table 1 are central to the construction of the seismicity-based and deformation-based models. For example, different weights are assigned to the FCAT–BCSF and ESHM20 catalogues, to the available focal-mechanism datasets, and to the different GNSS solutions. The manuscript acknowledges that weighting schemes represent an important source of expert-opinion uncertainty, particularly in Sect. 5.3. However, in my opinion, it would be useful to clarify whether the weights reflect data quality, spatial coverage, temporal duration, resolution, independence between datasets, or the authors’ assessment of relative reliability. The authors may wish to consider the following points:
- provide a specific justification for each group of weights, even if they derive from expert judgement;
- clarify whether the weights were defined before or after inspecting the resulting spatial patterns;
- discuss whether the results might be sensitive to reasonable variations in the weights;
- consider including a simple sensitivity analysis showing how the main zone boundaries change when the weights are modified. I appreciate that this last point could require substantial additional work and might perhaps be more appropriately addressed in a dedicated future study.
- Definition and justification of the confidence-level thresholds: The confidence levels of the zone limits are based on the normalized sum of feature differences, ∑ΔFi. However, the thresholds separating low, intermediate and high confidence are different for the three models:
- structure-based model: 0.15 and 0.40;
- seismicity-based model: 0.30 and 0.60;
- deformation-based model: 0.15 and 0.35.
The use of different thresholds may be justified because the three models are based on different types of observations and different feature distributions. Nevertheless, I think it could be useful if the manuscript explained more clearly how these values were selected. At present, the statement that the thresholds were chosen to obtain an “equitable distribution” of the three confidence levels appears somewhat arbitrary and does not fully establish their physical or statistical meaning. This part could perhaps be strengthened by explaining whether the thresholds were based on, for example: the distribution of ∑ΔFi values; quantiles or clustering of the data; expert-defined levels of feature contrast; or another objective criterion. It might also be helpful to specify how values exactly equal to a threshold are treated. For clarity, the intervals could be defined without ambiguity, for example using ≤ and > symbols.
- Comparability of confidence levels among the three models: The manuscript appropriately recognizes that the direct merging of the three models is debatable because the procedures used to derive the confidence levels differ among the models. This is an important and insightful observation. However, the unified model subsequently combines the confidence information from the three models and gives preferential weight to models presenting a larger number of intermediate- and high-confidence limits. This raises a question concerning the comparability of the confidence levels. A high-confidence limit in the structure-based model may not be directly equivalent to a high-confidence limit in the seismicity-based or deformation-based model, particularly because the thresholds and underlying observables differ. I think it could be valuable if the authors clarified whether the confidence levels are intended to have the same meaning in all three models. If feasible, the authors might also consider testing alternative merging strategies and comparing them with the current unified model.
- Need for independent validation: The proposed models are compared with each other and with previous zonation schemes, which is useful for assessing their internal consistency and geological plausibility. I think that a brief discussion could be added to acknowledge the importance of future independent quantitative validation. I recognize that validation is difficult in a low-seismicity region and that the available catalogues are not fully independent. Nevertheless, even a limited validation exercise could (perhaps in a future work) further strengthen the contribution of this work. For example, the authors might distinguish between data used to construct the models and a more recent subset used only for testing or could consider a spatial cross-validation using withheld events. If a formal validation is not currently possible, the authors could perhaps state explicitly that the models are presently evaluated through internal consistency, comparison with previous models and expert geological interpretation, rather than through predictive testing.
- Interpretation of weak or absent seismic and geodetic signals: A potentially important issue concerns the interpretation of areas characterized by limited seismicity or weak geodetic deformation. In such a low-seismicity setting, the absence of observed earthquakes or measurable deformation does not necessarily imply the absence of active tectonic processes. I think it could also reflect:
- the short duration of instrumental and geodetic observations;
- heterogeneous station coverage;
- deformation below the detection threshold or • aseismic deformation;
- or long recurrence intervals.
- This issue is particularly relevant where major geological or fault structures are not clearly expressed in the seismicity- or deformation-based models. The manuscript already acknowledges that some major structural features are not reflected by seismic or geodetic observations, and that zone delineation remains questionable where data are sparse. These aspects deserve an explicit discussion because they affect both the interpretation of the confidence levels and the construction of the unified model. It might be useful if the authors clarified how the distinction between “absence of evidence” and “evidence of absence” is considered in the zonation procedure. It could also be helpful to indicate whether low seismicity or low deformation is treated as a genuine physical signal, as an uncertain observation, or simply as a lack of constraint.
- Clarification of the methodological hierarchy: The manuscript refers to models, features, observables and datasets, but I found that the hierarchy among these terms is not always completely clear. Each model is described as incorporating three key features, while several features contain multiple observables and datasets with different weights. I have interpreted the chain as the following relationship: model → feature → observable → dataset → weighting → zone attribute → confidence level. Probably, something similar would make the workflow easier to reproduce and would help readers understand how the different types of information contribute to the final boundaries.
- Clarification of the role of the unified model in future PSHA :The manuscript states that the four models may be used as alternative models in a future PSHA logic tree. This is a promising application, but the implications should be discussed more carefully. I understood that the three thematic models do not represent alternative interpretations of the same information; rather, they emphasize different physical aspects of the seismogenic process. The unified model integrates all the three, and, for what I can see, could be the “final” model for PSHA. However, the authors should therefore clarify whether the models are intended to be mutually exclusive alternative source models (branches representing epistemic uncertainty in source characterization), or complementary models to be combined.
Minor comments
- The manuscript alternates between “structure-based” and “structural-based” SZM.
- The terminology related to deformation should also be standardized. The manuscript uses “deformation-based”, “surface deformation”, “crustal deformation” and “geodetic deformation” in partly overlapping ways.
- The hierarchy between “feature”, “observable” and “dataset” could be defined when Table 1 is first introduced.
- The confidence intervals should be written using non-overlapping inequalities. For example, the current formulation “0.15 < ∑ΔFi < 0.4” does not specify how values exactly equal to 0.15 or 0.4 are treated.
- “Expert-opinion related uncertainties” could be “uncertainties related to expert judgement”.
- “Evoking transparently the methodology” could be revised to “describing the methodology transparently”.
- “Our strategy apports objectivity” should be corrected, as some subjectivity still exists, for example to “Our strategy provides a more objective basis for assessing zone-boundary confidence”.
Best regards,
Francesco Visini
Citation: https://doi.org/10.5194/egusphere-2026-623-RC2 - Justification of dataset weights: The weights assigned to the different datasets in Table 1 are central to the construction of the seismicity-based and deformation-based models. For example, different weights are assigned to the FCAT–BCSF and ESHM20 catalogues, to the available focal-mechanism datasets, and to the different GNSS solutions. The manuscript acknowledges that weighting schemes represent an important source of expert-opinion uncertainty, particularly in Sect. 5.3. However, in my opinion, it would be useful to clarify whether the weights reflect data quality, spatial coverage, temporal duration, resolution, independence between datasets, or the authors’ assessment of relative reliability. The authors may wish to consider the following points:
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 706 | 335 | 77 | 1,118 | 64 | 100 |
- HTML: 706
- PDF: 335
- XML: 77
- Total: 1,118
- BibTeX: 64
- EndNote: 100
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The paper presents a new seismotectonic methodology for South-East France, the most seismically active region in metropolitan France. The study is based on the disaggregation of seismotectonic behaviour into three main components: the crustal/geological component, the seismological component, and the geodetic component. These three individual models are then integrated to produce a unified model that incorporates all the different datasets.
The manuscript has excellent scientific significance and overall quality, as well as a clear presentation of data, results, and conclusions. The amount of data used is very large and reflects the considerable effort carried out by the authors. The work represents an improvement over current approaches for mitigating seismic hazard in a critical region of France and in neighbouring areas such as north-west Italy, thanks to the integration of more recent datasets and the compilation of information that had not previously been considered.
In its current form, the manuscript would benefit from some clarifications, mainly listed below. I recommend publication once these points have been addressed:
-Clarification of the criteria adopted to build the unified model should be provided, and included also in the Methods section. It is not fully clear to me if the zone limit with the highest confidence level was selected to define the geometry of the unified zone. In general, I found it difficult to understand the role of the confidence levels associated with the zone boundaries, and how these were used later on. Perhaps more clarification on this matter should be provided.
- Some information should be provided on how the qualitative analysis of features of the structural model were quantified to calculate the normalized sum of differences.
- On a similar note, fracture numbers are defined as low, mean, high. What are these adjectives referring to? Which sources were used?
- The crustal model was built adopting one velocity, yet it is mentioned that two velocities for the Adriatic and European Moho are needed. Some justification on why you use only one velocity should be provided.
- I did not find an explanation of how fine and coarse analysis are build. In some areas, multiple signals of one type appear to be present, with variable predominance, yet in the coarse analysis a different dominant color is shown.
- To build the seismogenic thickness, different sources with different uncertainties are used and kept distinct. How were the overlapping zones between the different models handled?
- The ranges of confidence levels for the three models differ considerably. How does this impact the building of the unified model? Two of them appear rather uneven; does this affect the unified model?
Other minor comments are in the attached file.