the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Aug 2026
Monitoring-Informed Scenario Forecasting of Spatial Seismic Demand for Proactive Resilience in the İzmir Basin
Abstract. Recorded seismic observations from monitoring networks provide a direct empirical basis for updating regional representations of earthquake demand, yet their use in scenario forecasting and proactive resilience planning remains limited. This study proposes a monitoring-informed sense–forecast–adapt workflow for forecasting spatial seismic demand in the İzmir Basin, western Türkiye. The framework transforms accumulated strong-motion records into a regional residual correction layer, embeds this layer within active-fault-based earthquake scenarios, and translates the resulting demand fields into adaptation-oriented spatial indicators.
A transparent reference ground-motion representation is first fitted using event magnitude, source-to-site distance, and site-condition information. Event–station residuals are then aggregated at the station level and interpolated to construct monitoring-informed residual and residual-variability fields. These updating layers are embedded within five representative active-fault scenarios around the İzmir Basin to generate baseline and monitoring-updated PGA fields. The resulting demand fields are evaluated through persistence, scenario sensitivity, monitoring/inspection priority, rank variability, and spatial concentration of high-demand domains.
The workflow is evaluated through residual-field validation, Gaussian-process benchmarking, robustness analyses, cumulative model-version tracking, and Monte Carlo uncertainty propagation. The results identify the Tuzla Fault scenario as the dominant demand case and the Yenifoça Fault scenario as the main secondary case. High-demand zones organize into spatially coherent regions, several locations remain repeatedly critical across multiple source scenarios, and the virtual regional demand representation stabilizes as the monitoring archive expands from 1977–2010 to 1977–2025.
The main contribution is an updateable forecasting workflow that integrates monitoring-derived residual updating, scenario-based spatial demand simulation, uncertainty-aware exceedance mapping, and resilience-oriented decision indicators. By transforming accumulated strong-motion observations into an evolving regional demand representation, the proposed framework moves beyond pointwise ground-motion estimation and provides a digital-twin-oriented basis for proactive seismic resilience planning in monitored basin environments.
- Preprint
(2499 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2026-4292', Anonymous Referee #1, 15 Aug 2026
-
AC1: 'Reply on RC1', Şahin çağlar Tuna, 25 Oct 2026
We sincerely thank the Referee for the positive assessment of the revised manuscript and for the constructive comments provided during the review process. We greatly appreciate the Referee’s time and consideration. The subsequent revisions undertaken during the review process have further strengthened the manuscript, particularly in terms of methodological validation, uncertainty treatment, reproducibility, and clarity.
Citation: https://doi.org/10.5194/egusphere-2026-4292-AC1
-
AC1: 'Reply on RC1', Şahin çağlar Tuna, 25 Oct 2026
-
RC2: 'Comment on egusphere-2026-4292', Anonymous Referee #2, 26 Sep 2026
Reviewer Comments
Monitoring-Informed Scenario Forecasting of Spatial Seismic Demand for Proactive Resilience in the İzmir Basin
General assessment
The manuscript presents a relevant and potentially useful monitoring-informed framework for spatial seismic-demand forecasting in the Izmir Basin. The integration of strong-motion observations, residual-based spatial updating, active-fault scenarios, uncertainty propagation, and resilience-oriented indicators is coherent, and the long-term monitoring database provides a valuable basis for the proposed application. The manuscript is generally well organized and the digital-twin-oriented interpretation is appropriately qualified. However, several methodological aspects require further clarification and, in some cases, additional validation before the main conclusions can be fully supported. In particular, the interpretation of station-level residual corrections, predictive validation across stations/events/time, consistency of source-distance definitions, treatment of uncertainty, and reproducibility of the reference and spatial models should be strengthened. The comments below are intended to improve the methodological robustness, transparency, and reproducibility of the study while preserving its overall framework and practical relevance.
1.Separation of event effects from station-level residual tendencies: The current updating layer is constructed by averaging event–station residuals at each station. However, the residuals may contain event-specific effects in addition to persistent site/path effects. Please clarify how event effects are prevented from being absorbed into the station-level correction field, and provide a sensitivity analysis using event-centered residuals or an equivalent mixed-effects formulation.
2.Benchmark against a no-correction / simple-correction baseline: The leave-one-station-out results show only modest rank association for both IDW and Gaussian-process residual prediction. Please demonstrate whether the spatial residual layer adds predictive value relative to simpler alternatives such as no correction, a global mean residual, or a nearest-station correction.
3.Event-blocked and/or temporal out-of-time validation: The stated objective is forward scenario forecasting, yet the current validation focuses mainly on spatial interpolation at omitted stations. Please add a validation that more directly tests generalization to earthquakes not represented in the calibration data, preferably through grouped-event validation and/or an out-of-time experiment.
4.Consistency of the source-to-site distance definition between calibration and scenario forecasting: The reference model is calibrated using an epicentral-distance-based effective distance, whereas the scenario calculations use the shortest horizontal distance to the mapped fault trace combined with an assumed depth. Please discuss the transferability of the fitted distance term between these two distance definitions and, if possible, provide a sensitivity check or clearer limitation statement.
5.Distinction between aleatory residual variability and uncertainty in the correction field: The Monte Carlo analysis uses the interpolated station residual standard deviation as the local dispersion term. Please clarify whether this quantity is intended to represent event-to-event aleatory variability, epistemic uncertainty in the estimated station correction, or both. These concepts should be separated conceptually, and the terminology should be revised accordingly.
- Magnitude-range representativeness and possible small-event dominance: The monitoring database spans approximately M 2.2–7.6, whereas the scenario forecasts considered in the manuscript correspond to Mw 6.5–7.0 events. Please clarify the magnitude distribution of the calibration dataset and discuss whether the fitted reference model may be disproportionately influenced by the large number of small- and moderate-magnitude events. A magnitude-distribution summary and, if feasible, a sensitivity analysis using a moderate-to-large-magnitude subset would strengthen confidence in the applicability of the reference model to the scenario range.
- Spatial-block validation of the residual correction field: The current leave-one-station-out validation assesses interpolation performance at omitted individual stations; however, this procedure may still benefit from nearby monitoring stations and may therefore provide an optimistic representation of performance in sparsely instrumented areas. Please consider adding a spatial-block cross-validation or an equivalent analysis in which groups of neighboring stations are withheld simultaneously. Alternatively, prediction error could be evaluated as a function of distance to the nearest available monitoring station. This would provide a more realistic assessment of the spatial transferability of the residual correction field.
- Sensitivity of station-level residual corrections to the number of records per station: The station-level correction term is calculated as the mean residual over all usable records at each station. However, the statistical reliability of this quantity may vary substantially among stations depending on the number of available recordings. Please report the distribution of records per station and discuss how differences in sample size influence the estimated station-level residual tendencies. A sensitivity analysis using alternative minimum-record thresholds, or an uncertainty measure such as the standard error of the station mean residual, would help demonstrate the robustness of the spatial correction field.
- Separation of archive growth from monitoring-network expansion in the model-version analysis: The cumulative model-version experiment shows increasing stability as the archive expands from 1977–2010 to 1977–2025. However, the number of stations and spatial-update stations also increases substantially between model versions. Please clarify to what extent the observed stabilization reflects accumulation of additional recordings versus expansion of the monitoring network itself. If feasible, a supplementary analysis using a common set of stations across model versions would help distinguish temporal learning from changes in spatial monitoring coverage.
10.Explicit form of the reference ground-motion model: Please provide the explicit functional form of the reference ground-motion model.
11.Difference between the initial and usable number of records: Please clarify the difference between the initial number of records and the final number of usable records.
12.Minimum-record criterion for station residuals: The minimum-record criterion used for station-level residual updating should be specified.
13.Details of the IDW interpolation: Please provide additional details on the spatial interpolation procedure.
14.Gaussian-process benchmark and the choice of IDW: The Gaussian-process benchmark should be described in greater detail, and the reason for retaining IDW should be clarified.
15.Normalization used in MPI: Please clarify the normalization procedure used in the monitoring-priority indicator.
- ‘Monitoring / inspection priority’ terminology: The terminology ‘monitoring / inspection priority’ could be clarified because actual inspection priority would normally require exposure and vulnerability information.
17.Connectivity definition in the connected-component analysis: Please clarify the connectivity definition used in the connected-component analysis.
18.Interpretation of mean persistence: Please clarify the interpretation of mean persistence because the threshold is based on the pooled 80th percentile.
19.Rationale for scenario magnitudes: The rationale for the assigned scenario magnitudes should be stated more explicitly.
20.Monte Carlo spatial dependence: Please clarify whether the Monte Carlo perturbations are generated independently at individual grid points or whether spatial covariance is incorporated.
21.Consistent use of ‘digital-twin-oriented’: Please maintain consistent use of the term ‘digital-twin-oriented’ throughout the manuscript.
22.Figure readability: Figures 3–7 could be improved slightly for readability.
23.Terminology and language check: Please perform a final terminology and language check. Expressions such as ‘dissolved scenario sources’ may be confusing.
Citation: https://doi.org/10.5194/egusphere-2026-4292-RC2 -
AC2: 'Reply on RC2', Şahin çağlar Tuna, 25 Oct 2026
We sincerely thank the Referee for the detailed and constructive comments. The manuscript has been substantially strengthened in response to these suggestions, particularly with respect to methodological validation, robustness analyses, uncertainty treatment, reproducibility, and terminology.
A detailed point-by-point response to all comments is provided in the attached response letter. The corresponding revisions have been incorporated into the revised manuscript.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 199 | 95 | 35 | 329 | 30 | 27 |
- HTML: 199
- PDF: 95
- XML: 35
- Total: 329
- BibTeX: 30
- EndNote: 27
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The revised article, in its final form, is accepted for publication in the journal.