China&ndash;Korea Linked Typhoons: A transboundary sequential hazard framework for East Asian tropical cyclones

Na, Hana; Jung, Woo-Sik

doi:10.5194/egusphere-2026-3205

Preprints

https://doi.org/10.5194/egusphere-2026-3205

Preprints

21 Jul 2026

| 21 Jul 2026

Status: this preprint is open for discussion and under review for Natural Hazards and Earth System Sciences (NHESS).

China–Korea Linked Typhoons: A transboundary sequential hazard framework for East Asian tropical cyclones

Hana Na and Woo-Sik Jung

Abstract. Tropical cyclones over the western North Pacific (WNP) are among the most damaging natural hazards in East Asia, yet the transboundary sequential process by which a single typhoon affects China and then Korea has not been quantified. This study proposes the China–Korea Linked Typhoon (CKLT) framework and applies it to 70 years of official records (1949–2018) from the China Meteorological Administration (CMA) and the Korea Meteorological Administration (KMA). A CKLT is defined by three criteria: (i) the same WNP typhoon number appears in both CMA and KMA records; (ii) the China impact onset precedes or coincides with the Korea impact onset; and (iii) the Forecast Window (FW), defined as the interval between the two onsets, does not exceed 14 d. The FW is an observational temporal linkage metric, not a forecast skill measure. The 14 d threshold is jointly supported by Gaussian mixture model decomposition, kernel density estimation, and physical transit-time considerations, with the resulting case count stable across thresholds of 12–19 d. Mann–Kendall analysis revealed asymmetric long-term changes: China-affecting typhoon frequency in the CMA W8 station record increased (+ 0.570 events decade⁻¹, p = 0.006; constant-network subset + 0.45 events decade⁻¹, p = 0.032) while station-based mean wind and gust decreased; Korea-affecting typhoons in the KMA advisory record showed no frequency trend but exhibited intensification in advisory-based extreme gusts (+ 1.41 m s⁻¹ decade⁻¹, p = 0.022) and typhoon-conditional precipitation (+ 24.7 mm decade⁻¹, p < 0.001). Of the 227 Korea-affecting typhoons, 75 shared a common WNP typhoon number with CMA records and satisfied the sequential onset criterion; applying the 14 d Forecast Window threshold identified 36 (15.9 %) as CKLTs, with the FW distribution adequately represented by a parsimonious gamma model (α = 1.20, β = 5.21; KS p = 0.258) and a 6 d median (95 % bootstrap CI [2.95, 6.78] d; B = 5000). These results support the CKLT framework as an observation-based, population-level timing indicator for transboundary typhoon hazard analysis and bilateral disaster-preparedness research, rather than as a deterministic forecast of individual storm impacts.

Received: 03 Jun 2026 – Discussion started: 21 Jul 2026

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Hana Na and Woo-Sik Jung

Status: open (until 22 Sep 2026)

Post a comment Subscribe to comment alert

RC1:
'Comment on egusphere-2026-3205', Anonymous Referee #1, 19 Aug 2026 reply

This manuscript proposes a China–Korea Linked Typhoon (CKLT) framework, which quantifies the phenomenon whereby a single WNP typhoon successively affects China and then Korea as a transboundary sequential hazard process, and applies it to the official operational records of the CMA and KMA for 1949–2018. The topic is of practical relevance and is well aligned with current international trends in multi-hazard early warning and compound-event research. My major concerns are related to the definition and physical meaning of CKLT, the justification of the 14-day threshold, the interpretation of the Forecast Window, etc. Specific comments are listed below.
Major comments:
1. The conceptual definition of “linkage” lacks physical support. The identification of a CKLT relies solely on a common typhoon number, the temporal ordering of impacts, and a 14-day window. In essence, this amounts to two samples drawn from the life cycle of a single typhoon, rather than two separable events connected by physical transmission. The authors should rigorously distinguish between “a continuous track / life-cycle continuation of the same system” and “a genuine cascading / compound event.” I recommend adding an analysis of physical characteristics, such as track, translation speed, and transit distance between the China and Korea impact onsets, to demonstrate that CKLTs constitute a physically distinguishable, independent class of events.
2. The authors should provide a clearer framework to explaining why is 14 days preferred over 12, 15, 16, or 19 days? Why is the KDE minimum given greater practical importance than the GMM crossing point? Was the threshold specified before the analysis, or selected after examining the data? The manuscript currently reports that the identified case count is relatively stable between approximately 12 and 19 days, but this stability alone does not establish 14 days as the uniquely appropriate threshold.
3. The CMA uses a station-based measured wind-speed threshold (Bft ≥ 8), whereas the KMA uses an administrative/geographic threshold The two criteria are entirely different, so the “weakening over China vs. intensifying over Korea” asymmetric trend reported in the manuscript is very likely an artifact of the differences between the two recording protocols rather than a genuine physical asymmetry.
4. Only 36 CKLT cases are identified, which is a limited sample size. The gamma fit (KS p = 0.258 indicates only “not rejected”), the discrepancy between the bootstrap median (4.70 d) and the point estimate (6.0 d), and the treatment of FW = 0 (same-day impacts) all reflect uncertainty in the statistical inference.

Minor Comments:
1. Although the authors acknowledge that FW is not a measure of numerical forecast skill and does not use forecast-model output or advisory-issuance times, the term “Forecast Window” may still imply predictive capability or operational lead time. This becomes particularly problematic when the manuscript connects FW to early warning and preparedness. I recommend replacing the term, such as “Sequential Impact Interval”, and adding clarifying statements throughout the text to eliminate ambiguity.
2. The statement that “FW is not a forecast quantity” appears repeatedly, such as in the abstract, Sects. 2.5, 3.3, 4.3, and the conclusions, which could be consolidated.
3. The physical-mechanism explanations remain at the hypothesis stage, and the build-up on SST warming and poleward migration in the introduction is overly long; please condense it.
4. The term “China impact onset” should be defined more precisely. Is this the first calendar day on which any CMA station records Beaufort ≥ 8 winds? Similar for “Korea impact onset”.
5. Since both onset dates are recorded at daily resolution, the uncertainty of FW should explicitly be acknowledged. A one-day difference may partly reflect the arbitrary assignment of observations to calendar days.
6. Are the present results sensitive to the choice of analyzed time period, e.g., 1979-2025?

Reply

Citation: https://doi.org/10.5194/egusphere-2026-3205-RC1
- AC1: 'Reply on RC1', Hana Na, 31 Aug 2026 reply
  
  Major Comment 1
  RC1: The conceptual definition of "linkage" lacks physical support... I recommend adding an analysis of physical characteristics, such as track, translation speed, and transit distance between the China and Korea impact onsets, to demonstrate that CKLTs constitute a physically distinguishable, independent class of events.
  AC Response:
  
  We thank the reviewer for this comment and agree that the physical basis of "linkage" required clearer justification and quantitative support in the original manuscript.
  We agree with the reviewer's distinction. A CKLT is, by construction, a continuous-track life-cycle process, the same storm system that first produces a documented impact in China and subsequently produces a documented impact in Korea, rather than two physically independent events connected by a separate transmission mechanism. Our use of "cascading/compound event" terminology (following Zscheischler et al., 2018, 2020) was intended to describe the transboundary impact structure rather than to claim physical independence between the two impact stages of a single storm's life cycle. We have revised the Introduction and Sect. 2.5 to state this distinction explicitly.
  To support this quantitatively, we computed the transit distance (great-circle distance between the JMA best-track position at the China impact onset and at the Korea impact onset) and implied translation speed for CKLT cases with unambiguous best-track identification at both onsets (n = 21; same-day-onset cases and cases in which the onset ordering could not be resolved at daily temporal resolution were excluded). The median transit distance was 724 km (mean 934 km, range 196–2339 km) and the median translation speed was 14.0 km h⁻¹ (mean 14.5 km h⁻¹, range 2.4–28.8 km h⁻¹). These values are consistent with the a priori transit-time reasoning given in Sect. 2.5 (500–1000 km at 15–25 km h⁻¹), now based on direct track measurement, and support the identification of CKLT cases as physically coherent, continuous-track transit events rather than an artifact of the window definition. These figures represent a preliminary verification based on independently reconstructed best-track matching; final values reported in the revised manuscript will be confirmed against the authors' own case-by-case JMA best-track identification (as used to construct Fig. 3) prior to publication.
  We have added this summary and a supporting table (typhoon ID, SII category, transit distance, translation speed) to Sect. 3.3 of the revised manuscript.
  Major Comment 2
  RC1: The authors should provide a clearer framework to explaining why is 14 days preferred over 12, 15, 16, or 19 days? Why is the KDE minimum given greater practical importance than the GMM crossing point? Was the threshold specified before the analysis, or selected after examining the data?
  AC Response:
  We thank the reviewer for asking us to clarify the logic underlying the 14-day threshold, which we agree was not stated with sufficient precision in the original manuscript.
  Pre-specified vs. post-hoc. We agree the original phrasing ("pre-analysis inclusion criterion... based jointly on three lines of evidence") did not make the logical structure sufficiently clear, and could be read as claiming strict pre-registration, which is not accurate since GMM and KDE are by definition computed from the observed SII distribution. To clarify: the threshold was not selected by observing the SII distribution and choosing whichever value looked most favorable. Rather, it reflects convergence between an independent, theory-driven bound and two independent data-driven estimates. The physical bound (Sect. 2.5) is derived from typhoon translation-speed and transit-distance climatology, which does not depend on the observed SII distribution. The GMM decomposition and KDE are each computed from the SII distribution, but represent two methodologically distinct, independently arrived-at statistical estimates of where the short-SII and long-SII populations separate. We consider agreement among a theory-driven bound and two independent statistical methods to be stronger support for the threshold than strict pre-registration alone would provide, and have revised Sect. 2.5 to describe the logic in these terms rather than using the potentially misleading term "pre-analysis."
  Why 14 d and not 12, 15, 16, or 19 d. Given 500–1000 km transit distance and 15–25 km h⁻¹ translation speed, direct transit requires approximately 1–3 d. Allowing for CMA impact duration (1–5 d) and slower or recurving tracks, now shown by our transit-speed analysis (response to Major Comment 1) to extend toward 10–14 d for long-SII cases, the physically plausible upper bound extends to approximately two weeks. Independently, the KDE local minimum falls in the 14–16 d range. We selected 14 d as the lower, stricter edge of this range, which keeps CKLT identification conservative and consistent with the physical upper bound. As reported in Sect. 3.3.3, the resulting case count is stable across thresholds of 12–19 d (33–37 cases), so the specific choice of 14 among these nearby values does not materially affect the composition of the CKLT sample. 14 d was retained because the result is insensitive to it in this range, not because it is uniquely optimal.
  Why the KDE minimum rather than the GMM crossing point. The two methods answer different questions. The GMM crossing point (26.7 d) identifies where the short-SII and long-SII Gaussian components are equally probable, the point of maximum ambiguity between "linked" and "not linked," and is therefore a permissive, high-recall boundary. The KDE local minimum (14–16 d) identifies where the empirical density itself is lowest, a conservative, high-precision separation. Because the CKLT framework is intended to identify a confidently linked subset rather than to maximize inclusion, we prioritized the KDE-based separation. 14 d additionally has the property of sitting well below the GMM crossing point, so CKLT cases fall unambiguously within the short-SII population under both criteria. We have revised Sect. 2.5 and 3.3.3 to state this prioritization explicitly, rather than presenting the two methods as equally weighted.
  Major Comment 3
  RC1: The CMA uses a station-based measured wind-speed threshold (Bft ≥ 8), whereas the KMA uses an administrative/geographic threshold. The two criteria are entirely different, so the "weakening over China vs. intensifying over Korea" asymmetric trend reported in the manuscript is very likely an artifact of the differences between the two recording protocols rather than a genuine physical asymmetry.
  AC Response:
  We thank the reviewer for raising this important concern. We agree that the CMA and KMA operational definitions differ fundamentally (Sect. 2.2), and that this heterogeneity must be addressed before the asymmetric trend can be interpreted as a physical signal rather than a protocol artifact.
  First, we extended the constant-network robustness check already applied to typhoon frequency in Sect. 4.5 to the intensity variables underlying the asymmetric-trend claim. Restricting the CMA wind and gust analysis to the 22 stations continuously operational since 1949 to 1955, the China-side weakening trend remains essentially unchanged for mean maximum wind speed (constant-network: −0.49 m/s/decade, p<0.001; full network: −0.51, p=0.001) and remains significant for mean maximum gust (constant-network: −0.38 m/s/decade, p=0.032; full network: −0.63, p=0.001). This test cannot be explained by network expansion within the CMA record.
  Second, we note that both sides of the reported asymmetry are independently corroborated in the literature using observational datasets unrelated to either CMA's or KMA's operational criteria. On the China side, Jiang, Luo, and Zhao (2013, J. Meteor. Res., 27(1), 63 to 74) analyzed nationwide maximum wind speed across 535 CMA stations from 1956 to 2004, not restricted to typhoons, and found a declining trend of similar magnitude (−1.46 m/s/decade), explicitly ruling out instrument renewal, station relocation, and urbanization as causes, and attributing part of the decline along the southeast coast to declining typhoon frequency and intensity. On the Korea side, Basconcillo and Moon (2023, Scientific Reports 13, 5097) used the IBTrACS international best-track archive, independent of the KMA advisory criterion, and found a significant increasing trend in the peak intensity of typhoons passing through the Korean Peninsula from 1981 to 2020, with an abrupt increase since 2003. Together, these independent studies indicate that both the China-side weakening and the Korea-side intensification are documented in observational records that do not share either agency's specific operational definition, which substantially reduces the likelihood that the reported asymmetry is solely an artifact of the CMA/KMA protocol difference. We have added these citations and a brief discussion to Sect. 4.1 and 4.5.
  Third, we note that the China-side weakening trend is concentrated in the post-1984 sub-period (1984 to 2018: wind −0.55 m/s/decade; 1949 to 1983: +0.48, not significant) rather than a uniform linear decline, and we state this explicitly in the revised manuscript.
  Finally, we have revised the manuscript to state that the CMA-side and KMA-side trends should each be interpreted as robust within their own respective observational protocols, and the asymmetry as a documented divergence between two independently robust records rather than a claim of directly comparable absolute magnitudes, consistent with our existing statement in Sect. 2.2.
  Major Comment 4
  RC1: Only 36 CKLT cases are identified, which is a limited sample size. The gamma fit (KS p = 0.258 indicates only "not rejected"), the discrepancy between the bootstrap median (4.70 d) and the point estimate (6.0 d), and the treatment of FW = 0 (same-day impacts) all reflect uncertainty in the statistical inference.
  AC Response:
  We thank the reviewer for this comment and agree that the limited sample size (n=36) warrants explicit, upfront acknowledgment rather than being addressed only in scattered statements throughout the manuscript.
  Sample size and gamma fit. We agree that KS p=0.258 indicates the gamma model is not rejected by the data, not that it is confirmed as the uniquely correct distributional form. This is already noted in Sect. 3.3.4 and 4.5, where we state that the gamma form is adopted as a parsimonious working model consistent with the non-negative, right-skewed support of the Sequential Impact Interval, rather than as a uniquely identified generating distribution, and that other right-skewed distributions could provide comparable fits and should be re-examined as the CKLT sample expands. We have moved this caveat earlier, into Sect. 3.3.4 immediately after the KS test is reported.
  Bootstrap median vs. point estimate. The discrepancy between the bootstrap mean of the resampled median (4.70 d) and the original-sample point estimate (6.0 d) is already explained in Sect. 3.3.4: with n=36 and a right-skewed distribution, the sampling distribution of the median is itself discrete and skewed, so the bootstrap mean of resampled medians is expected to differ somewhat from the original point estimate. The original-sample point estimate (6.0 d) remains our central inference; the bootstrap interval (Table 5) characterizes sampling uncertainty around it rather than replacing it. We have clarified this distinction explicitly at first mention.
  Treatment of SII=0 (formerly FW=0). As stated in Sect. 2.6, SII is recorded at daily resolution, so SII=0 cases represent same-calendar-day impacts rather than exact zero-duration physical lags; the gamma model is used as a descriptive approximation to the empirical daily distribution, not as a literal continuous-time generating process at sub-daily resolution. We retained SII=0 cases in the fit because the fitted shape parameter (α=1.20) yields a finite density at the origin, avoiding an arbitrary exclusion rule. We have added a sentence to Sect. 2.6 stating this treatment explicitly as a limitation tied to daily-resolution source data, consistent with our response to Minor Comment 5.
  We have consolidated these three points into a single, upfront paragraph in Sect. 4.5 (Limitations), rather than leaving them distributed across Sects. 2.6, 3.3.4, and 4.5 as in the original submission.
  Minor Comment 1
  RC1: Although the authors acknowledge that FW is not a measure of numerical forecast skill... I recommend replacing the term, such as "Sequential Impact Interval", and adding clarifying statements throughout the text to eliminate ambiguity.
  AC Response:
  We thank the reviewer for this suggestion and agree. We have replaced "Forecast Window (FW)" with "Sequential Impact Interval (SII)" throughout the manuscript, including the abstract, Sect. 2.5 (definition, Eq. 1), Sect. 3.3 (results), Sect. 4.3 (operational implications), the conclusions, and all associated table and figure labels (Tables 3 to 5, Figs. 3 to 5). The definition itself is unchanged: SII = KorStart − ChnStart. We consider this renaming to directly resolve the ambiguity raised, since "Sequential Impact Interval" does not carry the predictive-skill connotation of "Forecast Window" while preserving the operational-planning meaning the term was intended to convey.
  Minor Comment 2
  RC1: The statement that "FW is not a forecast quantity" appears repeatedly, such as in the abstract, Sects. 2.5, 3.3, 4.3, and the conclusions, which could be consolidated.
  AC Response:
  We thank the reviewer for this suggestion and agree the repetition was unnecessary. The statement (now referring to the Sequential Impact Interval, SII, following our response to Minor Comment 1) appeared in similar form in the abstract, twice in Sect. 2.5, once in Sect. 3.3, once in Sect. 4.3, and once in the conclusions.
  We have retained one full statement of the point, in Sect. 2.5 immediately following the definition of SII (Eq. 1), where we explain that SII is computed entirely from post-event observational records and does not involve forecast-model output, advisory-issuance timing, or a predicted variable. In the abstract, Sect. 3.3, Sect. 4.3, and the conclusions, we have shortened each instance to a brief cross-reference, for example "SII is a population-level observational metric, not a forecast quantity (Sect. 2.5)," rather than restating the full explanation each time.
  Minor Comment 3
  RC1: The physical-mechanism explanations remain at the hypothesis stage, and the build-up on SST warming and poleward migration in the introduction is overly long; please condense it.
  AC Response:
  We thank the reviewer for this suggestion. We agree the paragraph was longer than necessary for its function in the Introduction and have condensed it from approximately 130 to 60 words.
  Original: "Long-term changes in WNP typhoon activity further amplify the urgency of a transboundary perspective. The poleward migration of maximum typhoon intensity (Kossin et al., 2014) has progressively increased the exposure of mid-latitude regions such as Korea to intense typhoons, while recent studies have reported increases in rapid intensification frequency and typhoon-related precipitation (Liu and Chan, 2022; Stansfield and Reed, 2023). In addition, accelerating sea surface temperature (SST) warming in the East China Sea and the seas surrounding the Korean Peninsula (Han et al., 2023; Kim et al., 2024) provides a thermodynamic environment that may modulate typhoon intensity, moisture supply, and precipitation structure as storms approach Korea. Nevertheless, most existing studies treat China and Korea as separate typhoon impact domains, leaving insufficiently characterised the spatiotemporal linkage structure by which hazard characteristics evolve as a single typhoon sequentially affects both countries along a continuous transboundary storm track. This single-country analytical limitation is itself a barrier to realising the cross-border early warning goals articulated in the Sendai Framework."
  Revised: "Long-term changes in WNP typhoon activity further amplify the urgency of a transboundary perspective: poleward migration of typhoon intensity (Kossin et al., 2014) and warming seas around the Korean Peninsula (Han et al., 2023; Kim et al., 2024) together increase Korea's exposure to intense, moisture-laden storms. Yet most existing studies treat China and Korea as separate impact domains, leaving the spatiotemporal linkage structure between them uncharacterised, a limitation that itself impedes the cross-border early warning goals of the Sendai Framework."
  We retained the key citations and the logical link to the single-country analytical gap, and removed supporting detail that was not essential to motivating the CKLT framework; Liu and Chan (2022) and Stansfield and Reed (2023) remain cited later in the Discussion (Sect. 4.4), so no citation is lost from the manuscript overall.
  Minor Comment 4
  RC1: The term "China impact onset" should be defined more precisely. Is this the first calendar day on which any CMA station records Beaufort ≥ 8 winds? Similar for "Korea impact onset".
  AC Response:
  We thank the reviewer for this comment and agree the definitions should be stated explicitly at first use rather than left implicit.
  China impact onset (ChnStart) is the first calendar day on which any of the 184 CMA W8 stations records sustained wind of Beaufort force 8 or greater (≥17.2 m/s) for the typhoon in question, taken as the minimum station-level onset date across all stations reporting Bft ≥ 8 for that typhoon.
  Korea impact onset (KorStart) is the first calendar day on which the typhoon is recorded as entering the KMA emergency zone (north of 28°N, west of 128°E) while a typhoon watch or warning advisory is in effect.
  We have added these two explicit definitions to Sect. 2.5 immediately before Eq. 1, replacing the current implicit phrasing, and added a cross-reference to Sect. 2.1 and 2.2 where the underlying CMA and KMA criteria are described.
  Minor Comment 5
  RC1: Since both onset dates are recorded at daily resolution, the uncertainty of FW should explicitly be acknowledged. A one-day difference may partly reflect the arbitrary assignment of observations to calendar days.
  AC Response:
  We thank the reviewer for this comment and agree this uncertainty should be stated explicitly rather than left implicit.
  Both ChnStart and KorStart are recorded at daily resolution in the source records (CMA W8 station dates, KMA advisory dates), so the Sequential Impact Interval (SII, formerly Forecast Window; see response to Minor Comment 1) is itself only resolved to the nearest calendar day. A reported SII of 0 or 1 d can therefore reflect true same-day or near-simultaneous impacts, or it can reflect two impacts separated by a few hours that happen to fall on adjacent calendar days depending on when within the day each onset was recorded, an artifact of calendar-day binning rather than a physically meaningful one-day gap. This uncertainty applies broadly across the SII distribution but is most consequential for the shortest-interval cases.
  We have added an explicit statement of this daily-resolution uncertainty to Sect. 2.5, immediately after Eq. 1, and added a cross-reference to it from Sect. 2.6 where SII=0 cases are discussed, and from our response to Major Comment 4.
  Minor Comment 6
  RC1: Are the present results sensitive to the choice of analyzed time period, e.g., 1979-2025?
  AC Response:
  We thank the reviewer for this suggestion. Extending the record to 2025 is not possible, as the CMA W8 and KMA impact records used in this study were obtained under data-use agreements covering 1949 to 2018 only (Sect. 2.2). We instead tested sensitivity by restricting the record to the satellite era (1979 to 2018), which additionally addresses the position-uncertainty caveat already noted in Sect. 4.5 for the pre-1977 JMA record.
  The results are robust to this restriction. For the China side, both CMA trends remain significant with similar or larger magnitude: mean maximum wind speed −0.76 m/s per decade (p=0.003; full period −0.51, p=0.001) and mean maximum gust −0.57 m/s per decade (p=0.032; full period −0.63, p=0.001). For the Korea side, the estimated slope for extreme gust is essentially unchanged (+1.33 m/s per decade versus +1.41 for the full period), confirming that the magnitude and direction of the trend do not depend on the choice of start year. The associated p-value is higher under this restriction (p=0.458 with n=38, versus p=0.022 with the full n=68), which follows directly from the roughly 45% reduction in sample size for a variable with substantial year-to-year variability, rather than from any change in the underlying trend. We consider the near-identical slope estimates across both period choices, for both China- and Korea-side variables, to be evidence that our results are not an artifact of the particular 1949 to 2018 study window. We have added this sensitivity analysis to Sect. 4.5.
  
  Reply
  
  Citation: https://doi.org/10.5194/egusphere-2026-3205-AC1

Hana Na and Woo-Sik Jung

Data sets

CKLT processed case list and figure plot data (1949–2018) H. Na and W.-S. Jung https://doi.org/10.5281/zenodo.20509810

Model code and software

CKLT analysis code (Python) for China–Korea Linked Typhoon identification and statistical analysis H. Na and W.-S. Jung https://doi.org/10.5281/zenodo.20509869

Hana Na and Woo-Sik Jung

Viewed

Total article views: 216 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
174	23	19	216	17	16

HTML: 174
PDF: 23
XML: 19
Total: 216
BibTeX: 17
EndNote: 16

Views and downloads (calculated since 21 Jul 2026)

Month	HTML	PDF	XML	Total
Jul 2026	114	15	13	142
Aug 2026	60	8	6	74

Cumulative views and downloads (calculated since 21 Jul 2026)

Month	HTML	PDF	XML	Total
Jul 2026	114	15	13	142
Aug 2026	60	8	6	74

Viewed (geographical distribution)

Total article views: 188 (including HTML, PDF, and XML) Thereof 188 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 01 Sep 2026

Short summary

Some typhoons hit China and then strike Korea a few days later. Using seventy years of weather records from China, Korea, and Japan, we identified these linked storms and measured the gap between the two impacts. On average, Korea was affected about six days after China, offering a useful window for emergency planning. Storms reaching Korea have also brought stronger gusts and heavier rain over time. These findings can help neighbouring countries warn each other and prepare earlier.


Total:	0
HTML:	0
PDF:	0
XML:	0