the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Sustained Late-Nineteenth-Century Monsoon Drought in Seoul and the Indian Ocean–North Atlantic Forcing of Its Circulation Weakening
Abstract. The Cheugugi rain-gauge record of Seoul, Republic of Korea (1777–1907), is among the world's longest instrumental daily precipitation records and a rare quantitative gauge of the pre-industrial East Asian monsoon. Here, we reconcile it against the original chronicle and use it to document, and attribute, a severe sustained rainy-season drought from 1881 to 1907: rainy-season (June–August, JJA) precipitation fell ~27 % and annual totals 28 % (with a Pettitt change point in 1880, p ≈ 0.01), with the contribution of days >100 mm falling by ~43 %. Three distinct record channels – the gauge, a separate documentary reconstruction, and qualitative rain reports in the Seungjeongwon Ilgi – registered the same shift, arguing against any decay in observational practice. In the Twentieth Century Reanalysis, the unconstrained precipitation field did not resolve the drought, whereas its pressure-constrained circulation registered a weakened monsoon low. Pacemaker experiments (10 members each) were conducted to identify the oceanic driver, prescribing the observed sea-surface-temperature anomalies over the tropical Pacific, North Atlantic, and Indian Ocean individually and globally. The historical forcing produced a significant weakening of the low-level monsoon circulation over Korea: a nearly 8 m depression of the JJA 850 hPa geopotential height (p < 0.001, the most robust signal), weakened southerly inflow, and significant reduction in the vertically integrated moisture supply. The geopotential-height weakening was driven principally by the Indian Ocean, the largest single-basin contributor, together with the North Atlantic, and not by the tropical Pacific despite its El Niño-like state. The reconciled record supplies a pre-industrial monsoon baseline and, with the experiments, attributes the monsoon-circulation weakening that accompanied the historical Korean drought to Indian Ocean and North Atlantic cooling. Future work should disentangle the tropical-cyclone contribution to the heavy-rainfall deficit and confirm the attribution with additional models.
- Preprint
(1458 KB) - Metadata XML
-
Supplement
(601 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2026-4812', Anonymous Referee #1, 16 Sep 2026
-
AC1: 'Reply on RC1', Byung-Kwon Moon, 12 Nov 2026
Response to Anonymous Referee #1 (RC1)
We would like to express our gratitude to the referee for the meticulous and constructive review. The point-by-point response is attached as a PDF document (14 pages). The redrawn Figs. 3–6, the new Fig. S5 and Tables S6–S7, and the cross-model figures from the response to Referee #2 are placed in the text adjacent to the corresponding paragraphs.
In summary: (1) The discrepancy in basin attribution between the height and wind-based fields is now explained (low signal-to-noise ratio of the wind-based fields; the meridional wind follows the zonal gradient, not the magnitude, of the height anomaly). The basin attribution is restricted to the T42/ERSSTv5 Z850 response, expressed as a ranking by point estimate with Holm-adjusted contrasts. (2) The single-model caveat has been superseded by new experiments conducted in response to Referee #2. A second atmospheric model (CAM4) replicates the full-forcing decrease of the Korea-box 850 hPa height, and a second SST reconstruction (COBE-SST2) replicates it at approximately half the amplitude with weaker statistical support. However, neither replicates the basin partition. We therefore propose to remove the basin attribution from the title, and to state in the abstract and Conclusions what holds across models and reconstructions, what does not, and which experiments would resolve the basin partition. (3) Multiplicity is now controlled (Holm adjustment within the primary family, false-discovery-rate q across all box-level tests, field-significant stippling following Wilks, 2016), and Figs. 3–6 are redrawn. (4)–(6) The domain sensitivity, the power of the observational ENSO tests and the block-bootstrap treatment would be summarised in the main text. The three technical corrections are addressed. Two further changes made on our own initiative – an instrumental extension indicating that annual precipitation remained below the pre-drought mean until 1913, and a re-examination of the provisional gauge values (dataset version 2.1), which leaves the main conclusions unchanged – are described at the end of the response. The revised manuscript will be submitted upon the editor’s request.
-
AC1: 'Reply on RC1', Byung-Kwon Moon, 12 Nov 2026
-
RC2: 'Comment on egusphere-2026-4812', Anonymous Referee #2, 21 Oct 2026
This study investigates the prolonged late-nineteenth-century drought in Seoul and its potential oceanic drivers using historical precipitation records, the 20CRv3 reanalysis, and SST-forced numerical experiments. The study is scientifically relevant, and the historical precipitation dataset is particularly valuable. However, several issues concerning model limitations, ocean-basin attribution, and data uncertainties require further clarification and validation. My Comments are listed below:
1. The conclusion that the Indian Ocean and North Atlantic were the primary drivers of monsoon circulation weakening is mainly based on a single T42 atmospheric model and the 850 hPa geopotential height response. However, the relative contributions of different ocean basins vary across circulation and moisture transport variables. Moreover, the model's summer rainband is displaced approximately 7° southward, limiting its ability to reproduce local precipitation. The authors should further assess the robustness of the attribution, clarify its model dependence, and distinguish between the attribution of circulation weakening and that of the observed drought.
2. Figure 5 compares the sum of individual ocean-basin responses with the global SST-forced response to evaluate nonlinear interactions. However, the global experiment also includes SST anomalies outside the three selected basins. Therefore, the difference cannot be attributed entirely to nonlinear interactions. The authors should clarify this limitation, further discuss possible interactions among ocean basins, and consider additional experiments or a more cautious interpretation.
3. Ocean observations during the nineteenth century were sparse, potentially introducing uncertainties into the ERSSTv5 reconstruction and the resulting attribution. Some historical precipitation records also contain provisional values. The authors should further assess how these uncertainties affect the SST-forced responses and estimated drought characteristics, particularly changes in extreme precipitation.
4. The drought analysis relies primarily on the Seoul Cheugugi record. A single station cannot establish whether the drought extended across the Korean Peninsula or the broader East Asian monsoon region.
5. Different reference periods are used for the precipitation analysis, SST composites, and numerical experiments. The authors should clarify the rationale for these choices and examine whether the main results remain robust under alternative reference periods. The statistical treatment of the change-point detection and significance tests should also be clearly documented.
Citation: https://doi.org/10.5194/egusphere-2026-4812-RC2 -
AC2: 'Reply on RC2', Byung-Kwon Moon, 12 Nov 2026
Response to Anonymous Referee #2 (RC2)
We would like to express our gratitude to the referee for a review that targeted the three assumptions on which the attribution depended: a single model, a single SST reconstruction, and a basin decomposition whose residual also contained the SST outside the three basins. Our point-by-point response is attached as a PDF (19 pages), in which every new or revised table and figure – Tables S8–S12, Figs. S4–S8 and the proposed Figs. 4 and 5 – is placed next to the paragraph that discusses it.
New experiments were run during the discussion: the full set repeated with a second atmospheric model (CAM4 physics in CESM 2.2; six experiments × 10 members), an experiment forced only by the SST outside the three basins (REST), and the full-ocean and single-basin experiments repeated with COBE-SST2. They reproduce the decrease of the Korea-box 850 hPa height under the full historical SST in CAM4 (−6.8 m against −7.7 m in the T42 model, p < 0.001 in both) and, at about half the amplitude and with weaker statistical support, under COBE-SST2 (−3.8 m; p = 0.03 as a single test, not significant after adjustment within the four-forcing family), but not the basin partition: no single basin is significant after Holm adjustment in more than one configuration. The REST height response is not statistically resolved in either model, and the complete four-term residual is not significant in either, so neither the absence of basin interactions nor a difference between the models is established. Neither model reproduces the Korean rain band; the attribution is therefore one of the circulation weakening, not of the rainfall deficit. We propose a title that names no basin, and an abstract and Conclusions that state what holds across models and reconstructions and what does not, and that name the experiments which would settle the basin partition (larger ensembles, pairwise-forcing runs, a third model and further reconstructions). The uncertainty of the SST reconstructions, the provisional gauge values, the other early Korean stations and the provincial Cheugugi records, and the reference periods are addressed in comments 3–5 with new Tables S8, S9 and S11. The revised manuscript will be submitted upon the editor's request.
-
AC2: 'Reply on RC2', Byung-Kwon Moon, 12 Nov 2026
Data sets
Cheugugi-Seoul daily precipitation dataset version 2 (1777–1907), analysis code, and SST-forced attribution experiments Byung-Kwon Moon https://doi.org/10.5281/zenodo.21867395
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 607 | 136 | 76 | 819 | 88 | 68 | 60 |
- HTML: 607
- PDF: 136
- XML: 76
- Total: 819
- Supplement: 88
- BibTeX: 68
- EndNote: 60
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript reconciles three independent reconstructions of Seoul's Cheugugi rain-gauge record (1777-1907) into a new daily dataset (Cheugugi-Seoul v2), documents a severe and sustained rainy-season drought from 1881-1907, and uses a set of SST-forced pacemaker experiments to attribute the associated weakening of the low-level monsoon circulation over Korea principally to Indian Ocean and North Atlantic cooling, with a comparatively minor role for the tropical Pacific despite an El Nino-like SST pattern during the drought epoch. The dataset reconciliation is a valuable contribution in its own right, and the study is commendably transparent about its own limitations, in particular the displaced summer rain band of the atmospheric model used for attribution. I recommend publication after the points below are addressed; none of them require new experiments.
General comments
The three-channel argument against observational-practice decay (Sect. 3.2) is a convincing way to rule out the most obvious alternative explanation for the drought signal. The use of 20CRv3 in Sect. 3.3, exploiting the fact that only pressure is assimilated over this period, is a defensible and instructive way to interrogate an early reanalysis. The pacemaker experimental design (Sect. 2.2, 3.5) is sound, and the additivity check (TP+NA+IO vs. GOGA, Fig. 5) is a welcome addition. My main concerns are (i) an inconsistency between diagnostic fields in the basin attribution that is not fully reconciled, (ii) the strength of the title/abstract claim relative to the single-model basis of the attribution, and (iii) some statistical details that are currently confined to the Supplement and would strengthen the main text if summarized there.
Specific comments
1. Basin attribution is inconsistent across diagnostic fields (Sect. 3.5, p. 8). The headline finding that the Indian Ocean and North Atlantic, not the tropical Pacific, drove the circulation weakening is based on the 850-hPa geopotential height field, described as having "the highest signal-to-noise ratio." However, the same section reports that for the 850-hPa meridional wind and the vertically integrated moisture flux, no single basin was more than marginally significant and the tropical Pacific contributed comparably. Since these latter fields most directly govern moisture delivery to Korea, I would ask the author to discuss explicitly why the basin ranking differs across fields, rather than noting the discrepancy in passing.
2. The title and abstract state the attribution in fairly unconditional terms, while the Conclusions appropriately note that the result "rests on a single T42 model whose southward-displaced rain band makes local precipitation unreliable" and that confirmation with additional models is warranted. I suggest bringing this caveat forward into the abstract, or softening the title/abstract phrasing, so readers who do not reach the Conclusions are not left with an overstated impression of certainty.
3. Statistical multiplicity across the many significance tests (Pettitt, block-bootstrap, permutation, Welch t-tests across five experiments and multiple fields, Spearman rank correlation, CSI, Fisher's exact test, superposed epoch analysis) is not discussed. Please clarify whether multiple-comparison correction was applied across the basin x field Welch t-tests, and whether the p<0.05 stippling in Figs. 3, 4, and 6 reflects pointwise or field significance.
4. The attribution diagnostics are computed over a fairly large "Korea box" (34-43N, 124-130E) relative to the single point-station record (Seoul, ~37.5N, 127E) that motivates the study. Some justification for this domain choice, or a sensitivity check with a tighter box, would strengthen the link between the box-mean diagnostics and the Seoul record.
5. The reversal of the conventional El Nino-driven explanation for Late Victorian droughts is central to the paper's novelty, but rests on a short (27-year) epoch (r = -0.07 for Nino3.4-Seoul correlation; p ~ 0.19 for the Fisher's exact test). A brief discussion of statistical power at this sample size would help readers weigh how strongly "not significant" argues against an El Nino role.
6. The block-bootstrap methodology and treatment of serial autocorrelation (Text S2) materially affect the robustness claims for a strongly autocorrelated, multidecadal series. Please summarize the approach briefly in the main text.
Technical corrections
- p. 3 (Sect. 2.2): the claim that the model's summer rain band sits "~7 degrees too far south" has no citation or supporting figure. Please clarify whether this is an established bias from prior work or a diagnostic finding of this study.
- p. 11 (Discussion): please clarify whether the comparison to tree-ring drought atlases (Cook et al., 2010) was checked directly against the atlas data, or is cited from the literature.
- Please confirm that all cross-references to Supplementary Tables/Text (S1-S8, Table S1-S5) are consistent with the final Supplement.