the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 16 Jun 2026
Human Social Sensing for Characterizing Long-Term Water-Related Discourse Dynamics: Implications for Sociohydrological Analysis
Abstract. Sociohydrological models seek to represent co-evolving human and water systems. Yet, static assumptions or poorly constrained parameters may limit their capacity to reproduce long-term transitions in societal discourse and policy priorities that shape hydrologic outcomes. To address this gap, we propose and evaluate a human social sensing (HSS) framework that converts long-term public discourse into quantitative, time-resolved proxy indicators associated with water-related discourse dynamics in sociohydrological systems. HSS integrates large-scale text mining with structured manual coding to capture scalable patterns and interpretable context in water-related narratives. We applied this framework to 61 years (1960–2020) of South Korean newspaper coverage, revealing a clear transition from disaster- and development-centered framing to sustained attention on water quality, environmental management, and ecosystem protection, alongside a tonal reorientation from economic development priorities toward environmental sustainability. Text mining identified structural changes in themes and salient terms, including the rising prominence of water-quality and water-resource themes. Meanwhile, manual coding provided an interpretable contextual benchmark for evaluating computational outputs and for identifying periods in which discourse shifts preceded institutional change. Together, the proposed framework generates high-resolution indicators of evolving water-related societal discourse and demonstrates the complementary strengths of computational analysis and human interpretation. By generating empirically derived discourse-level proxy indicators, HSS offers a structured framework for incorporating long-term societal framing and public discourse dynamics into sociohydrological analysis, while preserving interpretive caution about the indirect relationship between media discourse and public perception.
- Preprint
(1305 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- CC1: 'Comment on egusphere-2026-3395', Nima Zafarmomen, 17 Jul 2026
-
RC1: 'Comment on egusphere-2026-3395', Anonymous Referee #1, 30 Jul 2026
Lee et al.’s study proposes a Human Social Sensing framework that performs text analysis to characterize changes in South Korean newspaper discourse concerning water between 1960 and 2020. The authors identify a shift from disaster response and infrastructure development toward water-quality management, environmental protection, and sustainability. These findings are largely descriptive. They are also reasonably intuitive given South Korea’s broader transition toward stronger environmental regulation. Therefore, the scientific contribution to HESS appears extremely marginal.
More importantly, I struggled to identify what the manuscript teaches us about hydrological systems or human-water interactions in a manner that constitutes a substantive contribution to HESS. HESS publishes multidisciplinary research, but that interdisciplinarity is expected to advance hydrological science (HESS encourages and supports fundamental and applied research that advances the understanding of hydrological systems, their role in providing water for ecosystems and society, and the role of the water cycle in the functioning of the Earth system). In the present manuscript, no hydrological observations, processes, states, fluxes, or model outputs are analyzed. The study does not relate the derived discourse indicators to precipitation, floods, droughts, streamflow, water quality, water use, reservoir operations, or observed management responses. Instead, the connection to sociohydrology is developed primarily as a prospective application in the Discussion, rather than demonstrated within the analysis itself.
If the authors consider the manuscript to fall under # 3 of the HESS scope, they need to demonstrate that connection operationally rather than conceptually. For example, they could use a discourse indicator to estimate social memory decay after a major hydrological event. They could also test whether the indicators explain water-management or behavioral responses. Another option would be to incorporate an indicator into a coupled sociohydrological model. The authors could then test whether it improves the representation of an observed human–water feedback. Merely listing parameter calibration, proxy-variable, and lumped-model approaches without implementing and evaluating at least one of them is insufficient to establish a hydrological contribution.
Overall, I do not consider the manuscript suitable for publication in HESS in its present form and recommend rejection rather than major revision. The work may be better suited to a journal whose central remit includes environmental discourse, communication, policy, or social dimensions of environmental management. Potential journals could be Society and Natural Resources, Environmental Science and Policy, etc.
The following comments may help the authors improve the manuscript for submission elsewhere.
- Introduction: What specific research question is the manuscript trying to answer? The general objective is stated, but the scientific question remains unclear. The authors should also explain why this question matters beyond documenting a historical change in newspaper discourse.
- The authors should clarify the specific gap that HSS fills in hydrology and sociohydrological modelling. The important challenge is not simply improving text-mining methods, but converting latent and usually unobserved social constructs into measurable variables or parameters that can be used in a sociohydrological model. How does the proposed framework accomplish this conversion, and which model components does it help constrain, how it is operationally used?
- Section 4.3.1: The discussion of social-parameter calibration is useful. However, the manuscript does not show how HSS outputs would be used in practice. Please provide a detailed worked example.
- The four week per year sampling strategy requires stronger justification. Could major floods, droughts, pollution events, or policy debates etc have occurred outside these windows? The authors should compare the fixed and random samples separately. They should also test whether the results are sensitive to the chosen weeks.
- The orientation of the two newspapers should be discussed. Do they have known political, ideological, regional, or institutional alignments? Changes in ownership, readership etc may influence those observed trends.
- The categories “water policy,” “water quality,” “water for the environment,” and “natural hazards” may overlap. How were articles with multiple themes handled? The authors should provide clearer decision rules. They should also explain whether multiple classifications were allowed.
- Please include additional references only when they are directly relevant to the present study. Research on assimilating biophysical remote-sensing observations into hydrological models is not a direct methodological precedent for integrating discourse-derived social indicators. Such references should be included only if the authors establish a clear and substantive connection.
Citation: https://doi.org/10.5194/egusphere-2026-3395-RC1 -
RC2: 'Comment on egusphere-2026-3395', Anonymous Referee #2, 31 Jul 2026
General comment
This manuscript presents a Human Social Sensing (HSS) framework combining manual coding and text-mining methods to examine long-term changes in water-related newspaper discourse in South Korea. The interdisciplinary topic is interesting and potentially relevant to sociohydrology.
However, the manuscript currently provides mainly a descriptive analysis of media discourse. The derived indicators are not directly linked to observed hydrological conditions, water-management responses, human behaviour, or practical applications in sociohydrological modelling. The main contribution therefore appears to be the HSS framework itself. Substantial revision is needed to clarify the hydrological contribution, strengthen its connection to sociohydrological modelling, and demonstrate the relevance of the proposed framework to HESS.
Major comments
1. The central concern is that the connection to sociohydrology remains largely conceptual. The use of socioeconomic indicators alone is insufficient to establish a clear contribution to hydrological science.
a. The manuscript mentions sociohydrological models and their limitations several times (Lines 18–20, 40–41, 44–47, and 381–383), but the proposed framework is not implemented within or evaluated alongside such a model. The authors should demonstrate how an HSS indicator could be incorporated into, used to constrain, or evaluated alongside a sociohydrological model. Preferably, this should include a modelling application showing how the identified discourse trends could be linked to specific model variables or parameters.
b. The discourse indicators should also be compared with corresponding historical meteorological, hydrological, and water-management data, such as precipitation, discharge, drought and flood records, water-quality observations, reservoir operations, or policy changes. A more in-depth attribution analysis is needed to determine whether changes in media discourse correspond to actual hydrological events or management responses.
2. The novelty and robustness of the HSS framework require further justification.
a. The manuscript should more clearly distinguish the proposed framework from previous studies combining manual coding, TF–IDF, semantic analysis, and topic modelling. Its methodological novelty should be stated explicitly. The authors should also define the quantitative indicators generated by the framework, explain their uncertainty and temporal resolution, and show how they could be used reproducibly in hydrological or sociohydrological applications.
b. Only four weeks per year were sampled from two newspapers. This strategy may omit important floods, droughts, pollution incidents, and policy debates occurring outside the selected periods. The authors should provide a stronger justification and assess the sensitivity of the results to the sampling strategy, for example by comparing the fixed-period and randomly selected samples.
c. Although two researchers independently coded the articles, no quantitative measure of inter-coder reliability is reported. Statistics such as Cohen’s kappa or Krippendorff’s alpha should be provided. The decision rules for overlapping themes and ambiguous tone categories should also be clarified.
3. The analysed data represent newspaper discourse rather than direct observations of public perception. Changes in media discourse may also be influenced by policy changes, government regulations, or management decisions. Although this limitation is acknowledged in Section 4.4, the distinction should be maintained consistently throughout the manuscript. For example, Section 2.5 refers to the coding categories as measures of “social values,” while Lines 244–246 interpret the results as a “broad societal reorientation.” Similar claims regarding “public attention” and “public priorities” appear in the Results and Discussion. These statements should refer more precisely to changes in newspaper discourse or media framing unless supported by independent survey, behavioural, or institutional evidence.
Citation: https://doi.org/10.5194/egusphere-2026-3395-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 87 | 33 | 7 | 127 | 4 | 5 |
- HTML: 87
- PDF: 33
- XML: 7
- Total: 127
- BibTeX: 4
- EndNote: 5
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The manuscript presents a novel Human Social Sensing (HSS) framework for quantifying long-term water-related societal discourse using historical newspaper archives. By integrating manual coding with computational text-mining techniques—including TF-IDF, semantic analysis, and topic modeling—the authors derive time-resolved indicators of changes in water-related discourse in South Korea from 1960 to 2020. The results indicate a long-term shift from disaster response and infrastructure development toward water quality, environmental management, and sustainability-oriented narratives. The study provides an interesting interdisciplinary contribution by proposing a structured framework that could potentially support sociohydrological modeling through empirically derived discourse-based proxy variables.
egusphere-2026-3395.pdf