the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 17 Apr 2026
From basin-scale variability to regional coherence: emergence of a dominant common mode in reservoir storage dynamics in Catalonia
Abstract. Long-term records of reservoir storage provide an integrated observational view of regional hydrological variability in managed systems and are directly relevant for water-resource management and drought risk assessment at regional scale. Here we analyze daily records of stored water volume percentages for major reservoirs in Catalonia over the period 1960–2026, using exclusively observational data provided by the Catalan Water Agency. After physically consistent preprocessing, we construct a set of annual metrics, including the annual mean stored volume percentage and the fraction of days below defined low-storage thresholds. Pairwise correlation analyses reveal a clear contrast between intra-basin and inter-basin behavior and show a pronounced increase in inter-basin synchronization after the late 1980s, particularly for low-storage and extreme drought metrics that are most relevant for water scarcity conditions. To assess whether this increased synchronization reflects emergent inter-basin coupling or the dominance of a common large-scale signal, we explicitly remove the leading regional mode of variability using a principal-component-based common-mode analysis applied to the annual reservoir metrics. After removal of this dominant mode, inter-basin correlations are strongly reduced in both early and late periods, and the apparent pre/post contrast is substantially weakened. The analysis provides robust observational evidence that inter-basin correlations strengthen after the late 1980s and that this change is largely captured by a single dominant regional common mode (PC1) in the annual reservoir metrics. Residual inter-basin structure beyond this leading mode is comparatively weaker and does not exhibit the same systematic pre/post amplification. These results indicate that reservoir systems that were historically only weakly coordinated now behave increasingly as a single regional system dominated by a common large-scale signal, particularly during drought-relevant conditions. This has direct implications for water management, as it limits the effectiveness of spatial diversification and inter-basin compensation strategies. No causal attribution is attempted; the results instead provide a quantitative observational baseline for future studies combining reservoir data with climatic and operational covariates.
- Preprint
(3277 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2026-384', Anonymous Referee #1, 04 May 2026
- AC1: 'Reply on RC1', Carles Serrat, 13 May 2026
-
RC2: 'Comment on egusphere-2026-384', Anonymous Referee #2, 28 May 2026
This manuscript investigates the emergence of a dominant common mode in reservoir storage dynamics in Catalonia through principal component removal and pairwise correlation analysis. While the findings are of potential interest, I have considerable concerns regarding the academic depth and methodological rigor of the work, which in my view preclude its publication in a high-standard academic journal in its current form.
Although the authors explicitly position this study as purely observational and intentionally circumvent attribution analysis, the statistical framework underpinning the study is itself insufficiently developed. Several methodological issues warrant serious attention. First, the relatively short common period used for the calculation of Pearson correlation coefficients — with a minimum of only 10 years — is not adequately acknowledged or discussed, despite the well-known sensitivity of correlation estimates to sample size. Second, the choice of 1986 as the change point for partitioning the full observational record into pre- and post-1986 periods appears arbitrary and lacks a transparent justification, whether statistical, physical, or historical in nature. Third, and perhaps most critically, the study does not report any assessment of the statistical significance of the results, nor does it provide any quantification of the associated uncertainties. Taken together, these shortcomings substantially undermine the robustness and interpretability of the conclusions drawn.
Beyond the statistical concerns, I would strongly encourage the authors to consider incorporating an attribution analysis into the study. Given the intriguing narrative the manuscript puts forward, a more rigorous investigation into the underlying drivers of the identified common mode would considerably enhance the scientific credibility and broader contribution of the work.
In light of these concerns, I recommend rejection of the manuscript in its current form.
Citation: https://doi.org/10.5194/egusphere-2026-384-RC2 -
AC2: 'Reply on RC2', Carles Serrat, 29 May 2026
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-384/egusphere-2026-384-AC2-supplement.pdf
-
AC3: 'Reply on AC2', Carles Serrat, 29 May 2026
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-384/egusphere-2026-384-AC3-supplement.zip
-
AC4: 'Reply on AC2', Carles Serrat, 29 May 2026
The attached files contain the revised manuscript with modifications highlighted in red together with the detailed response to Referee #2.
Citation: https://doi.org/10.5194/egusphere-2026-384-AC4
-
AC3: 'Reply on AC2', Carles Serrat, 29 May 2026
-
AC2: 'Reply on RC2', Carles Serrat, 29 May 2026
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 196 | 47 | 22 | 265 | 16 | 21 |
- HTML: 196
- PDF: 47
- XML: 22
- Total: 265
- BibTeX: 16
- EndNote: 21
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript provides a well-motivated observational analysis of long-term reservoir storage dynamics across multiple river basins in Catalonia. The central question (whether reservoir systems exhibit increasing inter-basin synchronization over time) is relevant to hydrology, drought risk assessment, and water resources management.
The methodological framework is generally sound, combining annual aggregation, correlation analysis, and principled attempt to separate common mode variability using principal components analysis. A key strength is the explicit attempt to distinguish between genuine intra-basin and inter-basin behavior and shared regional forcing represented by a dominant common mode. The PC1 removal analysis is valuable and strengthens interpretability of the results.
However, several important conceptual and methodological issues require clarification. In particular, the interpretation of the common mode removal, the justification of the temporal breakpoint, and the inference that residual stricture is limited beyond PC1 need further support. Overall, the manuscript is publishable but requires revision, primarily to strengthen interpretation and clarify methodological assumptions.
1. Interpretation of the common mode (PC1)
The most important conceptual issue concerns the interpretation of PC1. The manuscript states that removing PC1 reveals limited residual inter-basin structure, implying that synchronization is largely explained by a single large-scale/regional signal. However, PC1 is defined on annual aggregated reservoir anomalies, not on external forcing. This means that it represents the dominant covariance structure of the dataset, not necessarily a physically unique regional climate mode. The concern is that the manuscript risks overstating the physical interpretation of a purely statistical construct.
It is my recommendation that the authors clarify that PC1 represents the leading empirical mode of co-variability in reservoir storage and not a uniquely identifiable physical driver. Consider tempering conclusions about single dominant regional mode unless linked to independent climatic evidence.
2. Causality vs statistical co-variability
The discussion carefully avoids casual claims, which is appropriate. However, some phrasing (e.g. line 159, explained primarily by a single regional-scale signal shared across reservoirs) implicitly suggests causality. I recommend that the authors distinguish more clearly between statistical common mode (observed covariance structure) and physical forcing mechanics (not directly analyzed here). The existence of a dominant PC1 is well demonstrated, but its physical interpretation remains unsolved, which the authors do indicate is not the goal of this manuscript. The PC1 should not be referred to as a “regional signal” without supporting climatic or operational evidence.
3. Choice of temporal breakpoint (1986)
The split into pre and post 1986 periods is described as exploratory. The results are strongly dependent on this segmentation, and no formal change point detection is provided. The breakpoint may appear somewhat ad hoc despite justification. The authors should either include a simple change point analysis or emphasize that results are robust to nearby split years (if tested).
4. Correlation as the primary metric.
Pearson correlation is appropriate for linear co-variability, but reservoir dynamics are often nonlinear and threshold driven, and correlation can be inflated by shared trends or persistence. Although PC1 removal addresses this, the manuscript would benefit from discussion of robustness to Spearman correlation or rank-based measures.
5. Interpretation of increasing synchronization
The manuscript concludes that increasing synchronization is largely due to a dominant common mode. While the PC1 results support this, the following nuance is important. The “removal of PC1 reduces correlations”, line 214, does not necessarily imply the absence of multi-basin coupling or the absence of higher-order shared structure.
6. Hydrological interpretation of reservoir storage
The manuscript treats reservoir storage percentage as a proxy for hydrological state, which is valid, but this is strongly influenced by operational decisions and may not directly reflect natural hydrological coherence. I recommend adding a brief discussion of management driven synchronization (e.g., policy, allocation rules) vs climate-driven synchronization. This is especially relevant given the strong post 1980 signal.
7. Missing linkage to external forcing
The manuscript explicitly avoids external datasets, which is acceptable. However, the interpretation of a regional mode would be strengthened by at least referencing known climatic shifts that impact Catalonia (e.g., Mediterranean drought regimes). Even without analysis, a short contextual paragraph would improve interpretability.
Additional corrections/suggestions
Technical corrections