the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Jun 2026
Lithosphere–Coversphere–Atmosphere–Ionosphere Coupling with the 2023 Turkey Earthquake Doublet: A Deviation–Time–Space–Frequency Analysis
Abstract. The prerequisite of extracting reliable earthquake precursors from multi-parameter observations is still a challenge because of high false anomalies in the traditional analysis techniques. The given limitation is partly mitigated by a novel criterion of Deviation–Time–Space–Frequency (DTSF) anomaly detection on the 2023 Turkey earthquake doublet (Mw 7.8, Mw 7.5). The DTSF criterion builds on Deviation-Time-Space (DTS) analysis, where an anomaly must meet four rigorous criteria: statistical significance of greater than 15–day baselines of greater than ±1.4σ, quasi-synchronous activation, spatial association with geosphere-specific manifestation zones as well as frequency-domain validation in the form of band-specific power enhancement, cross-layer coherence of ≥0.5, and physically consistent phase relationships. Microwave brightness temperature (MBT), surface latent heat flux (SLHF), outgoing longwave radiation (OLR), total electron content (GPS-TEC, GIM TEC), and electron density/temperature (Ne, Te) under geomagnetically quiet conditions are analyzed. Results demonstrate the significance of the enhanced DTSF anomaly extraction approach in two aspects. First, the systematic vertical coupling sequence between the lithospheric stress and ionospheric perturbations through four temporal stages by rough estimations of the wavelet coherence analysis and phase–lag evaluations. SLHF precedes TEC by 2.5±0.3 days (C=0.71) and OLR by 1.2 ± 0.2 days (C=0.61) during pre-seismic phases; co-seismic coupling exhibits same-day MBT-TEC coherence (C=0.70–0.85), distinguishing impulsive seismic forcing from gradual processes. Second, the frequency criterion is a high-pass filter of credible anomalies that rejects noise in meteorological and space weather applications better than the DTS analysis that would falsely identify a precursor. For this case study, DTSF criterion achieved a detection rate of 89 % with 8 % false positives, 85 % reduction in false anomalies compared to conventional criteria. This case-specific study demonstrates the potential of the DTSF approach for validating Lithosphere–Coversphere–Atmosphere–Ionosphere (LCAI) coupling chain, which may promote the development of synergistically multi–parametric identification of earthquake precursors, while validation remains pending across additional earthquake cases.
- Preprint
(8040 KB) - Metadata XML
-
Supplement
(1059 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-2310', Anonymous Referee #1, 01 Aug 2026
-
RC2: 'Comment on egusphere-2026-2310', Anonymous Referee #2, 28 Aug 2026
The manuscript presents an interesting refinement of the DTS methodology by adding a frequency-domain (wavelet coherence, phase-lag, spectral non-stationarity) criterion, tested on the 2023 Turkey Earthquake doublet. The core idea, that requiring cross-layer coherence and physically consistent phase lags reduces false positives relative to amplitude-only deviation detection is a methodological contribution. However, several aspects of the presentation, validation design, and internal consistency need tightening before the claims are fully supportable.
Comments:
1. No sensitivity test for the hard coherence threshold (C̄XY ≥ 0.5)
A sensitivity analysis is reported for the composite F-score weights (±0.10 perturbation → <0.05 F-score change), which is good practice, but the hard 0.5 coherence cutoff embedded in criterion F2 is not similarly stress-tested — despite several reported coherence values sitting close to this boundary (e.g., 0.48–0.62, 0.35–0.45). Extend the existing sensitivity framework to the coherence threshold itself (e.g., report detection rate/false-positive rate at 0.45, 0.50, 0.55) so readers can judge how threshold-dependent the headline results are.
2. Disturbance-radius formulas (ρ_LC, ρ_A, ρ_I) — derivation not restated
The manuscript cites a companion paper (Rasheed et al., 2025) for the generic disturbance-radius formulas but doesn't briefly restate how these were derived or validated independently of the present case. A two-sentence summary would help readers assess circularity risk, since the same author group derived both the radii and this validation.
3. GIM-TEC spatial grid resolution vs. the localized precursor zone is not discussed
GIM-TEC data are used at 2°×2° spatial resolution (per Table 1), which corresponds to a grid cell roughly 200×200 km at these latitudes — potentially spanning or exceeding the estimated ionospheric disturbance radius ρ_I. The manuscript does not discuss whether the epicentre and the reporting GPS stations (e.g., ISTA) fall within the same GIM cell, or how grid-cell averaging might dilute or smear a localized precursor signal. Add a sentence clarifying the spatial relationship between the GIM-TEC grid cell(s) used, the epicentre, and ρ_I, and comment on whether coarse grid resolution could bias the reported GIM-TEC vs. GPS-station (point) comparisons.
4. Table 2 shading rule ('DTSF criterion across ≥3 parameters') not illustrated
The rule that shaded rows in Table 2 satisfy the DTSF criterion across three or more parameters is stated only in the caption and never illustrated with a worked example in the main text. Walking through one example row (e.g., showing which parameters trigger the shading and how the F-score threshold interacts with the ≥3-parameter rule) would aid readability.
5. Stage boundaries (Sect. 4.1) do not fully align with Table 2's DAE range
The Discussion describes four temporal stages with specific day windows (e.g., Stage 1: T₀-15 to T₀-10), but these boundaries don't map perfectly onto Table 2, which only tabulates data from DAE -10 onward
6. Anomaly attribution between the two doublet earthquakes is not addressed
The 6 February 2023 Turkey sequence comprises two major ruptures (Mw 7.8 and Mw 7.5) roughly nine hours apart. The manuscript treats the co-/post-seismic window as a single block but never discusses how anomalies observed after the first rupture (e.g., co-seismic MBT-TEC coherence) are disentangled from signals potentially driven by, or contaminated by, the second rupture and its own aftershock sequence
7. Stage boundaries (Sect. 4.1) do not fully align with Table 2's DAE range
The Discussion describes four temporal stages with specific day windows (e.g., Stage 1: T₀-15 to T₀-10 days), but these boundaries don't map perfectly onto Table 2, which only tabulates data from DAE -10 onward. Extending Table 2 to cover the full T₀-15 window, or clarifying why earlier days are omitted, would make the stage narrative verifiable against the data.
8. Typesetting/formatting errors in Equations 2–9
Several equations contain garbled symbols and dropped subscripts (e.g., the scale parameter 'a' appears inconsistently in Eqs. 7–9; the exponential decay term in Eq. 3 is hard to parse). These should be corrected in the final typeset version so the underlying math can be verified.
9. Reference list name-order inconsistency
The Rasheed et al. (2025) reference lists the corresponding author as 'Lixin, W.' rather than 'Wu, L.' as used consistently elsewhere in the reference list. Please correct for consistency.
10. 'TID' acronym used before being defined
'TID' (traveling ionospheric disturbance) appears in Sect. 2.4 ('0.2–5 MHz (TID band)') before being spelled out; it is only expanded implicitly later. Please define at first use.
11. GPS station identification in Table 1 / Fig. 1
Station codes (ISTA, TUBI, NICO, ZECK) are used throughout the text and figures, but full station names/coordinates are given only in running text for some stations. Consolidating full names and coordinates in Table 1 or the Fig. 1 caption would aid readability.
Citation: https://doi.org/10.5194/egusphere-2026-2310-RC2 -
RC3: 'Comment on egusphere-2026-2310', Anonymous Referee #3, 13 Sep 2026
Report on Lithosphere–Coversphere–Atmosphere–Ionosphere Coupling with the 2023 Turkey Earthquake Doublet: A Deviation–Time–Space–Frequency Analysis by Rasheed, Chen, Ding, Wang, Mahmood, Wu
The authors propose the Deviation Time Space Frequency (DTSF) method to detect earthquake related anomalies across the lithosphere, atmosphere, and ionosphere.
They applied the method to the 6 February 2023 Turkey earthquake doublet using Microwave Brightness Temperature( MBT), Surface Latent Heat Flux(SLHF), Outgoing Longwave Radiation (OLR), Total Electron content(TEC), and Swarm plasma data.
Wavelet analysis identifies multi-day pre-seismic relationships and strong same-day co-seismic coherence among selected parameters.The authors report that the DTSF method detects anomalies before and during the earthquakes while producing fewer false alarms than the conventional Deviation Time Space (DTS) method.
However, the present analysis does not demonstrate that the reported anomalies are specifically related to earthquake preparation.
General Comments
The method is tested retrospectively on a single earthquake sequence, without non earthquake controls, so its false alarm rate and predictive performance cannot be established. Moreover, the analysis combines measurements with different temporal resolutions and spatial coverage, often reducing them to daily anomaly labels. Same-day occurrence does not demonstrate simultaneity, and without explicit spatiotemporal collocation, the datasets cannot be considered mutually confirming evidence.
Specific comments
The study combines brief Swarm passes, AMSR-2 overpasses, hourly TEC observations, high-rate GPS data, and daily SLHF and OLR averages. These measurements are frequently reduced to a binary daily anomaly label. Occurrence on the same day does not demonstrate simultaneity. The exact UTC interval and duration of each anomaly should be reported. Without this information, the proposed temporal ordering and phase relationships cannot be verified.
The observations refer to different locations and spatial scales. Surface parameters are averaged over regional grids, TEC is derived from maps and geographically separated GPS stations, and Swarm data are selected within a broad area around the epicentres.
The authors do not demonstrate that compared measurements sample the same atmospheric or ionospheric region. In particular, local Swarm electron density and column-integrated TEC cannot be considered mutually confirming measurements without explicit spatiotemporal collocation.
The equations given in the paper produce approximate radii of 972 km for the lithosphere/coversphere, 3,780 km for the atmosphere, and 8,265 km for the ionosphere for an Mw 7.8 earthquake. Such large regions provide limited spatial discrimination and substantially increase the probability of chance associations.
The criterion Kp < 4 includes Kp = 3, 3+, and 4−, which cannot be regarded as strictly quiet conditions. Figure 2 appears to show Kp values near or above 3 during parts of the analysed interval. These conditions may influence TEC and plasma parameters. A daily summary is insufficient; geomagnetic indices should be aligned with the exact anomaly intervals.
The background definitions used for the different datasets are not clearly described.
A complete earthquake/no-earthquake confusion matrix is not provided. Therefore, the reported false alarm rate, detection rate, and predictive performance cannot be independently assessed.
Terms such as validation, mechanistic confirmation, causal relationship, and credible precursor are not supported by a retrospective single-event analysis. The observations show, at most, associations compatible with the proposed LCAI hypothesis.
Citation: https://doi.org/10.5194/egusphere-2026-2310-RC3
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 184 | 83 | 31 | 298 | 46 | 21 | 16 |
- HTML: 184
- PDF: 83
- XML: 31
- Total: 298
- Supplement: 46
- BibTeX: 21
- EndNote: 16
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Please see the attached pdf file.