This manuscript explores how the length of the observation time window of earthquake catalogues influences probabilistic seismic hazard assessment (PSHA) results, using a 1-Myr physics-based synthetic catalogue for the Eastern Betic Shear Zone (EBSZ). The topic is timely and relevant for slow-deforming fault systems where historical and instrumental records span only a fraction of the seismic cycle. However, several key methodological aspects require clarification and stronger justification, particularly regarding the characterization and validation of the synthetic seismicity and the transparency of fault parameters used in the simulations, detailed below.

Fault parameters and slip rates

The manuscript relies heavily on fault-based modelling, yet essential fault parameters are not explicitly reported. It would be important to:

• Provide slip rates for each active fault included in the EBSZ model, rather than only a system-averaged value.
• Summarize key fault parameters (e.g., fault length, segmentation, coupling assumptions) in a table.

this information would be helpful for readers to validate the physical plausibility of the synthetic seismicity and the resulting hazard estimates.

Validation of the synthetic seismicity

The manuscript states that the synthetic catalogue reproduces the main features of observed seismicity, but quantitative validation is limited. How is the credibility of the RSQSim outcomes validated beyond qualitative agreement? Are synthetic and historical catalogues statistically comparable in terms of rates, magnitude distributions, and moment release? The cumulative seismic moment analysis is useful but should be more clearly integrated into the validation framework rather than presented mainly as a discussion element.

Maximum magnitude

• The procedure used to calculate Mmax (average between geological and catalogue-based values) should be explained more clearly and justified.
• The frequent occurrence of M ≥ 7.2–7.4 events in the synthetic catalogues appears high. How does this compare with the observed catalogue (i.e., how many such events are historically documented)?
• Table 1 suggests that only one adjacent seismogenic area allows M ≥ 7.3 even considering uncertainty, which appears inconsistent with the synthetic results.

Accessibility of key references

Several critical methodological choices (e.g., ZESIS parameters, Mmax estimation, declustering rules) rely on references written only in Spanish.

• Given that these references underpin essential tables and assumptions, key information should be summarized explicitly in the manuscript for an international audience.
• This is particularly important for Table 1, which contains crucial parameters for the hazard assessment.

Interpretation of higher hazard from synthetic catalogues

The manuscript concludes that synthetic catalogues lead to systematically higher hazard than historical or ZESIS-based assessments. This interpretation should be refined. That is, historical catalogues do not necessarily always underestimate hazard; sampling a rare large event from a long-recurrence fault could also lead to overestimation. Alternative explanations should be discussed more explicitly, including catalogue incompleteness, overestimated slip rates or coupling in the synthetic model, and the absence of secondary faults or aseismic deformation.

Epistemic uncertainty

The manuscript states that epistemic uncertainties are common to all synthetic sub-catalogues, but fixing epistemic uncertainty does not eliminate it. The authors should clarify how epistemic uncertainties related to fault geometry, slip rates, coupling, and GMPE selection are treated, and how they differ from variability associated with the observation time window.

Thus, I suggest this manuscript can be published after a major revision.

Chung-Han Chan, National Central University, Taiwan, January, 2026.

The manuscript is well written and supported with informative figures. The methods employed are applicable, appropriately applied and adequately described. I have only some minor/moderate suggestions prior to publication.

General comment on the use of synthetic sub-catalogues

The manuscript states that 100 sub-catalogues of 1 kyr duration are randomly extracted from a 1 Myr synthetic catalogue and used for PSHA. However, I would like to suggest clarifying more explicitly that each sub-catalogue represents only one possible realization of earthquake occurrence over a limited time window, and not the long-term “true” seismic hazard of the system. As currently written, the text may give the misleading impression that each sub-catalogue is equally representative of the long-term hazard, whereas they should instead be interpreted as stochastic realizations sampling different portions of the seismic cycle.

Slip-rate input and data origin.

The mean slip rate of the fault system (1.5 ± 0.3 mm/yr; Echeverría et al., 2013) is a key input of the synthetic model and therefore strongly controls the long-term seismic moment release and earthquake productivity. However, its origin is not discussed. I would like to suggest clarifying how these slip rates were derived (e.g., paleoseismological trenching, geomorphology, geodesy) and over which temporal window they are averaged. This information would be important for assessing the consistency between the geological input and the simulated seismicity.

Magnitude–frequency distributions of sub-catalogues

It may be useful to include a figure showing the variability of the magnitude–frequency distributions (MFDs) of the individual sub-catalogues compared with those derived from the historical/instrumental catalogue and from the ZESIS database. This would help the reader to visually assess how representative the synthetic sub-catalogues are with respect to observed seismicity and to the long-term model.

Spatial variability of hazard curves

In Figures 8 and 9, the hazard curves and return periods differ significantly between cities. It is not clear why such variability is expected if all sites belong to the same tectonic domain (same seismogenic area). If the cities are located within the same source zone, the observed differences may simply reflect source discretization effects (e.g., edge vs. central locations). I would like to suggest discussing this point more explicitly. Alternatively, if some cities are located outside the main study area, it may be worth questioning whether including peripheral sites is necessary.

Interpretation of epistemic vs aleatory uncertainty

The manuscript states that the use of multiple sub-catalogues allows the evaluation of epistemic uncertainty. However, since all sub-catalogues are generated from the same synthetic model with identical fault geometry and slip-rate inputs, epistemic uncertainty is common to all of them. I would like to suggest that the variability observed among sub-catalogues mainly reflects aleatory variability associated with earthquake occurrence. In practice, the study is quantifying the dispersion of hazard estimates around the long-term mean due to temporal sampling effects.

Complex ruptures and maximum magnitudes

The manuscript suggests that complex ruptures are more common than previously thought and associates the largest magnitudes (Mw ≈ 7.2) with such events. However, it is not entirely clear from the figures whether “complex” refers to multi-segment/multi-fault ruptures or simply to large slip values on individual faults. I would like to suggest clarifying this distinction. Moreover, it may be useful to explore whether the large magnitudes are controlled primarily by rupture geometry or by slip distribution, and whether this can be verified within the simulation framework.

Interpretation of Table 3 and justification of the ensemble size

The justification of the use of 100 sub-catalogues based on the similarity of the Gutenberg–Richter parameters in Table 3 is not entirely clear. I would like to suggest explaining more explicitly how this comparison demonstrates that the ensemble provides a statistically stable estimate of the long-term seismicity model.

Synthetic vs observed hazard levels

The manuscript concludes that hazard derived from historical and instrumental catalogues is less conservative than that obtained from synthetic seismicity and attributes this to the limited sampling of large earthquakes. While this is a plausible interpretation, it implicitly assumes that the synthetic model represents the “true” long-term hazard. I would like to suggest treating this assumption more cautiously. Historical catalogues do not necessarily always underestimate hazard: the sampling of a rare large event on a long-recurrence fault could also lead to an overestimation (assuming the average recurrence represents the “true” hazard). Alternative explanations could be discussed, including catalogue incompleteness, uncertainties in slip-rate estimates or fault coupling in the synthetic model, and the possible absence of secondary faults or aseismic deformation.

Role of very large magnitudes in hazard

The manuscript highlights a significant contribution of magnitudes larger than Mw 7.2. However, in my experience (mainly in Italy, so this may not be fully transferable), such events are not expected to dominate hazard at return periods of 500–2000 years and may be more relevant for risk analysis than for classical PSHA. I would like to suggest further discussing this point.

Validation/comparison with observations of the synthetic catalogue

The manuscript states that the synthetic catalogue reproduces the main features of observed seismicity, but quantitative validation is limited. I would like to suggest discussing whether synthetic and historical catalogues are comparable in terms of rates, magnitude distributions, and cumulative moment release. The cumulative seismic moment analysis is useful, but I would encourage integrating it with a direct MFD comparison.

Best Regards