the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Apr 2026
Operational damage misclassification in scenario-based ShakeMaps: evidence from station-updated ground-motion fields in a deep-basin urban environment
Abstract. Near-real-time ShakeMap-based damage assessments are widely used to support rapid post-earthquake decision-making. However, their operational reliability depends on how accurately ground-motion fields represent local site and basin effects. This study investigates the potential for operational damage misclassification arising from purely scenario-based ShakeMap representations in deep-basin urban environments.
The 30 October 2020 Samos earthquake was analysed for the Mansuroğlu Neighborhood (Bayraklı District, Izmir, Türkiye) using a two-stage framework: (i) a scenario-based rapid damage estimation and (ii) a station-updated near-real-time configuration incorporating strong-motion recordings from Disaster and Emergency Management Authority of Türkiye (AFAD) stations located within a 3 km radius.
Results show that the scenario-based configuration systematically underestimates intermediate-period spectral demand (T = 0.6–1.0 s), which governs the response of the predominantly mid-rise reinforced concrete building stock. These discrepancies propagate into cumulative damage exceedance probability estimates. While the scenario-based approach largely confines the {Moderate + Extensive + Collapse} exceedance probability to the 0–10 % range, station-based updating increases this range to approximately 15–30 % in critical zones.
This shift represents a transition across an operationally meaningful threshold with direct implications for response categorization and resource prioritization during the early post-earthquake phase. The findings demonstrate that misclassification risk in rapid damage assessment arises not only from modelling uncertainty but also from threshold-sensitive distortions in exceedance estimation.
Even a limited number of spatially proximal strong-motion stations can substantially enhance the robustness of ground-motion representation. The study therefore highlights ShakeMap calibration as a governance-relevant intervention in seismic risk management rather than merely a technical refinement.
- Preprint
(2287 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 22 May 2026)
- RC1: 'Comment on egusphere-2026-1523', Gordon Woo, 02 May 2026 reply
-
RC2: 'Comment on egusphere-2026-1523', Abhineet Gupta, 10 May 2026
reply
Kara has used GMPEs to generate a map of ground shaking from the 2020 Samos earthquake and compared the median ground motion estimates obtained from GMPEs with the observed ground motions at 4 stations. Based on their analysis, they have concluded that the median ground motion estimates may not be representative of actual ground shaking, and therefore may underestimate damage.
By itself, the value of this study is unclear to me. It is well known that the median ground motion estimates from GMPEs are not representative of actual observations, and USGS ShakeMaps and PAGER already incorporate observed ground motions to update the estimates. The author has only analyzed one event, which has not been compared against the published USGS ShakeMap for the event, despite being in the title.
One of the key contributions of the manuscript: calibration of shaking maps using observed ground motions, has not been described anywhere besides mentioning that the ground motions were calibrated using bias correction.
While their conclusions regarding damage underestimation and basin effects for the one event are supported, they claim broad conclusions generally applicable, for which they have not provided any analyses.
I also found a general lack of references for the statements included throughout the manuscript.
Finally, while I fully support the use of AI tools for language improvement, the manuscript has not been throughly reviewed after their use, and contains AI slop and repetitions, endemic to AI tools.
I have included additional major and minor comments below:
- ShakeMap refers to a specific product created and published by USGS. It appears that the author has used ShakeMap to refer to a map of ground shaking from an earthquake event. Please only use ShakeMap when referring to the product, and use another terminology, e.g., shaking maps, when referring to a generic map of ground shaking.
- Introduction through line 84: The author has not introduced their study event in the first part of the Introduction, instead indicating that this issue occurs broadly. However, both ShakeMap and PAGER estimates produced by USGS immediately after an earthquake do in fact account for observed ground motions, and are only constrained by the availability of local station data (see e.g., Wald et al. Challenges in Rapid Ground Motion Estimation for the Prompt Assessment of Global Urban Earthquakes; https://doi.org/10.1785/0120100101). As a result, please clarify whether the author is referencing only the Samos earthquake as missing observational data in the produced ShakeMaps, and evidence that observational data were not included.
- Line 110: Suggest adding map of earthquake epicenter, faulting regime, and damaged areas to help orient readers.
- Line 113: Suggest adding more information about “structural deficiencies”, including the type of structures commonly in the area, type of structures impacted and the most, common deficiencies that resulted in damage.
- Line 115: Please add reference for how the effect of “structural deficiencies” was dissociated from the “influence of local site response effects”.
- Section 3: Please describe how the earthquake source was characterized for generating the shaking maps: e.g., point source, planar source, 3D geometry, etc. Additionally, please provide details about the values of source parameters used in each of the 4 GMPEs.
- Line 269: How was the dataset of 962 buildings collected: were they randomly sampled within a selected region, or were they all the damaged buildings within a region, or were they a subsample of the damaged buildings within a selected region? How was the region selected?
- Line 316: How were the grid estimates associated with station locations? Additionally, since the scenario based levels are obtained using GMPEs, why were grid based values used instead of calculating them directly at the station locations?
- Figure 5: How were H1 and H2 components defined at the stations? Did they align with geographic N and E, or with respect to the direction of earthquake rupture?
- Eq 3, 4: What was the reason for choosing to use absolute difference instead of the ratio? Typically it’s the log of spectral accelerations that are generated by GMPEs, and the standard deviation is also given for log(Sa). similarly the fragility functions are generally defined on the log scale (using lognormal distribution) on Sa. As a result, ratios of spectral acceleration are more representative of the differences that absolute values.
- Line 341: What is the process of calibration? Paragraph 29: Please include references for the statements in the paragraph.
- Line 46: What is “multi-level building damage classification framework”?
- Line 47: Sentence intent is unclear: … enable response strategies to be formulated on a rational basis …
- Line 49: Please add a reference. Perhaps it has the same reference as Li and Xu, 2025; Li et al., 2023, in which case, suggest adding it again to clarify.
- Line 50: What is meant by time gain of 6-12 hours?
- Line 54: What is “contemporary disaster management”?
- Line 54: What are “Such platforms”?
- Line 55: What are “spatially explicit estimates”?
- Line 56: I do not believe that HAZUS has capability for near-real-time damage estimation system. Please verify for all listed systems.
- Line 59: What is meant by “reliability”? Does it refer to accuracy, or the real-time availability and uptime of the products?
- Line 61: Please provide a reference.
- Line 62: Please define the terms: Vs30, Z1.0, and Z2.5.
- Line 64: What is meant by “inevitably introduces both epistemic and aleatory uncertainties”. Please clarify and add reference.
- Line 64: What is meant by “persistent methodological gap”?
- Line 67: What is meant by “distort intermediate-period demand”?
- Line 67: What are the “complications” to the balance?
- Line 69: Please further describe “system performance in terms of statistical prediction accuracy and bias reduction”
- Line 71: Please add references for “previous studies”.
- Line 74: What is meant by “optimistic bias”?
- Line 75: Please provide reference.
- Line 76: It is unclear what the sentence is intended to say. It looks repetitive.
- Line 79: what is “response category”?
- Line 80: What are “operational response thresholds”, and is the author referring specifically to Turkey or globally?
- Line 101: Are these Mw the ones published immediately after the event, or the Mw as of 2026 by the different agencies? If the latter, why would there be epistemic uncertainty of 0.4 points 5 years after the event?
- Line 115: The sentence structure is circular as the author mentions: although structural deficiency contributed, one primary factor was local site effects + “structural vulnerability”.
- Line 124: What is meant by “seismic hazard assessment based solely on standardized code-based approaches”, especially in the context of author’s focus on ShakeMaps?
- Line 125: Please explain “detailed microzonation studies”.
- Line 139: The previous statement does not support the inclusion of “in combination with local site amplification effects”.
- Line 142: The author has described the structural deficiencies, but then point to the relevance of “basin-related amplification characteristics”. It is unclear how the structural deficiencies reason given by the author instead points to site amplification as the reason for excessive damage.
- Line 146: Since the author has only analyzed one event, suggest changing the statement to “… this deep-basin urban environment…”
- Line 164: Please define M, E, C.
- Line 165: Please provide supporting reference. This seems counter-intuitive as emergency responders would rather prefer to have absolute damage counts, but it is in fact the alleatory uncertainty from modeling limitations that restricts providing counts, and instead probabilities are provided.
- Line 166: What is meant by “theoretical damage distributions”?
- Line 211-216: The author has listed three sources of Vs30. It is unclear which source was ultimately used, and if multiple sources were used, how they were combined.
- Line 230: Please provide a reference.
- Line 233: How is “intermediate-period” defined?
- Line 234: Please provide reference.
- Eq 1: Why is velocity represented by m/sn instead of m/s? What does “n” refer to?
- Line 241: What does it mean by “characterized by predominant site periods in the range of 1.0–3.0 s”?
- Line 245: Which “observations” is the author referring to?
- Line 248: What does it mean by “was adopted through bias correction, not as an optional enhancement, but as a methodological requirement”
- Figure 3: The panel on the left is difficult to read. Usage of green color both for the background and the damage state makes it difficult to distinguish them. The resolution appears low to differentiate among buildings.
- Line 261: What does “harmonization” refer to?
- Line 264: Please provide reference.
- Line 278: Please provide a reference.
- Line 291: Suggest including the fragility curves for all selected building types to help orient readers about their relative performance.
- Line 301: Please describe “intensity-based fragility functions”.
- Line 319: What is meant by “local representational differences”?
- Line 320: How was the radius established: was it based on the outer extents of the neighborhood, or based on a weighted centroid?
- Figure 5: Can the spectral acceleration plots be enlarged to improve clarity?
- Figure 5: What is “Intensity” in the table?
- Table 1: Please define the terminology used in the table in the table caption, e.g., S_T, SHA, etc.
- Table1 : Which directional component at each station location is listed in the table?
Citation: https://doi.org/10.5194/egusphere-2026-1523-RC2 -
RC3: 'Comment on egusphere-2026-1523', Anonymous Referee #3, 11 May 2026
reply
The study evaluated the ground-motion characteristics and observed damage distribution in an area of Turkey, during the Samos earthquake, using 2 stage-operational framework and investigated the potential operational damage misclassification which arises from scenario based on ShakeMap representation in deep-basin environments. Starting from one single example, as the author mentioned at Conclusions, the extrapolation has to be done with caution for other tectonic and urban environments. So, it opens future directions of research and I hope that the author will expand his research in future. I suggest where the author refers to the ' literature' to put in the parenthesis some examples which probably are already at Bibliography.
Citation: https://doi.org/10.5194/egusphere-2026-1523-RC3
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 149 | 35 | 16 | 200 | 13 | 15 |
- HTML: 149
- PDF: 35
- XML: 16
- Total: 200
- BibTeX: 13
- EndNote: 15
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The Samos earthquake, which occurred offshore of Samos Island, Greece, on 30 October 2020 represents a key reference event for earthquake engineering studies in Türkiye, and merits the ground motion study undertaken by the author. The limitations of GMPEs are well-known, so the difference between scenario-based and observation-supported analyses will be large in deep-basin urban environments.
The author states that near-real-time ShakeMap-based damage assessments are widely used to support rapid post-earthquake decision
making. It is pointed out that in densely built urban districts such as Mansuroğlu, cumulative damage exceedance levels approaching 15–30% may increase the likelihood of localized road obstruction due to debris accumulation, thereby challenging uniform accessibility
assumptions embedded in emergency response planning. Under such conditions, the author states that response strategies may require area
specific prioritization.
Given the time elapsed of almost six years since the Samos earthquake, the author should address some key issues in post-earthquake decision making. In what specific ways can updated shaking maps improve post-earthquake response? How could updated shaking maps have contributed operationally to the rapid response on 30 October 2020, beyond immediate first-responder engagement?
Another key risk mitigation issue raised by this paper is the local level of code compliance. What data on code compliance is readily available to first responders?