the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 20 May 2025
Advances in Monitoring Black Sea Dynamics: A New Multidecadal High-Resolution Reanalysis
Abstract. The Black Sea regional reanalysis serves as an essential tool for understanding the Black Sea’s response to climate variability and advancing regional ocean monitoring efforts. In particular, the Black Sea reanalysis (BLK-REA) is built with high spatial resolution, 1/40° horizontal grid and incorporating 121 vertical levels. The model implementation includes lateral open boundary conditions (LOBC) at the Marmara Sea, allowing more accurate inflow/outflow dynamics through the Bosphorus Strait. BLK-REA assimilates sea level anomaly (SLA) and in-situ observations and applies a heat flux correction via sea surface temperature relaxation. Enhancements in data assimilation (DA) include an improved background error covariance matrix and an observation-based mean dynamic topography for SLA assimilation. When compared to available observations, the numerical results show high accuracy, with the largest temperature errors observed in the upper layers, primarily linked to the formation of the seasonal thermocline during the summer months. The SLA anomaly error is consistently around 0.02 m from the year 2000 onwards, and regions with elevated SLA errors are closely associated with the Rim Current and its mesoscale variability. Furthermore, BLK-REA plays a crucial role in generating Ocean Monitoring Indicators, which are essential for tracking and assessing the impacts of climate change in the Black Sea. For example, temperature data indicate ongoing warming in the 25 to 150 m layer, where the Cold Intermediate Layer is located. In addition, the Black Sea meridional overturning circulation has decreased from 0.1 Sv in 1993 to approximately 0.01 Sv in 2010, highlighting significant changes in the basin's circulation. The system is regularly updated, with the next version expected to improve both the model and DA components. For a future perspective, the next BLK-REA will expand the domain to include the Azov Sea and will feature an enhanced Bosphorus LOBC.
- Preprint
(7620 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-2114', Anonymous Referee #1, 09 Jul 2025
The study introduces a new high-resolution regional ocean reanalysis for the Black Sea (BLK-REA), covering 1993–2022. Key strengths include an enhanced model grid, improved boundary conditions, an updated observation-based mean dynamic topography (MDT), and refined error handling. Validation spans multiple time periods, depth levels, and subregions, and the authors effectively situate their work within the existing literature while making good use of publicly available data and tools. However, the manuscript omits any quantitative comparison with the previous reanalysis. To substantiate the claimed improvements, the authors should present figures and tables that compare error metrics from the old and new reanalyses, and from the model free-run (without assimilation but with SST nudging), using the same validation datasets and circulation diagnostics. A more fundamental concern lies in the title’s promise to “monitor the dynamics” of the Black Sea: by using a sequential 3D-VAR assimilation scheme without dynamically consistent background covariances, it’s unclear whether true ocean dynamical features are being captured or artificially generated. This issue could become especially critical as the authors plan to extend the reanalysis back to 1980, before satellite data were available, as trends or interannual variability could simply reflect changes in data coverage. Finally, the manuscript would benefit from careful copyediting and enhanced figure design to improve readability and visual clarity.
Specific Comments:
- Abstract and Introduction
1- The abstract and Introduction are not well structured. In both the authors discuss their previous reanalysis before describing it.
2- The abstract highlights pronounced interannual variability in the meridional overturning circulation, but the intended message isn’t clear. In particular, it’s uncertain whether data assimilation has modified the overturning cell compared to the model free-run. This point is worth exploring further.
3- Line 47: “models that are constrained”, may be “models that are driven”? also I would remove “best” from “best available observations” unless the authors had a different meaning in mind.
- Methodology
1- The authors state that the current reanalysis uses monthly-averaged LOBCs from a U-TSS simulation covering 2016–2019, but the reanalysis period spans 1993–2022. Could the authors clarify how these boundary conditions are applied outside the U-TSS period, and discuss any implications for consistency over the full timeline? Have the authors tried to use more frequent LOBCs, like weekly?
2- The use of an observation-based mean dynamic topography (MDT) in the current assimilation system represents a significant methodological shift from the previous version described in Lima et al. (2021), which relied on a model-derived MDT (1993–2012). To justify this change, the authors should briefly comment on the impact of this change, andprovide supporting evidence (e.g., comparative diagnostics or sensitivity tests) illustrating how it improved SLA assimilation or the overall model skill.
3- The presentation of the cost function (Equations 1 and 2) should include a description of the terms: what does x include?, the choice of the components of observational errors R, etc. What the authors mean by “use the same values for S and T” in Line 133? And what threshohld did they apply for data rejection in line 135?
4- The choice of specific multi-year periods EOFs (1984–1993, 1994–2003, etc.) should be explained. Why these particular time blocks were selected? Were alternative choices considered? What about the seasonal aspects?
5- The estimation of representation errors for SLA and T/S observations is not fully detailed. Since salinity errors are larger in the upper 200 m despite LSBC being applied below 700 m, could you clarify how these errors are specified and whether dynamic bias correction or adjustments to the prior covariance could further reduce the persistent salinity biases?
- Results
1- The authorsthe analysis error for validating the reanalysis. I would suggest using the forecast error, or at least discussing it.
2- In Section 3.1 and Figure 2 (top), you show that temperature RMSD exceeds 2 °C in the upper 50 m during summer, suggesting a persistent misrepresentation of the seasonal thermocline. Could you provide some discussion of possible causes and comment on whether any sensitivity tests were explored?
2- In Section 3.1, the authors note that the intensified upwelling and resulting SST biases along the Anatolian coast may stem from the bulk formulation, and that refinements are planned in future versions. It would be helpful if you could briefly elaborate on what aspects of the formulation you intend to refine or test, and whether any preliminary sensitivity experiments have been conducted to assess their impact.
3- Figure 2 (bottom) and Figure 4 highlight that salinity RMSD can exceed 1 psu near the surface, particularly in the western Black Sea region. Could you elaborate on whether these discrepancies may be attributed to biases or uncertainties in freshwater forcing?
4- In Section 3.1 and Figure 7, the authors note that reanalysis-derived EKE values are substantially higher than those derived from altimetry, especially in the Batumi and Sevastopol eddies. I cannot follow the authors reasoning that this is due to the coarse resolution of the data. This is more likely to be coming from the model being over-energetic in resolving mesoscale variability, which in this case needs to be better discussed and analyzed in the manuscript.
5- It would be useful to check the quality of the figures and captions. For instance, the font in Figure is too small. The caption of Figure is not clear. The title of Figure 7 “for layer: surface”?
- Sections 2.2 and 3.1: To help readers better understand data coverage and its impact on validation, it would be useful to indicate how many in-situ profiles are typically assimilated per cycle and to quantify observation density over time (e.g., profiles per year or season).
Citation: https://doi.org/10.5194/egusphere-2025-2114-RC1 -
RC2: 'Comment on egusphere-2025-2114', YoungHo Kim, 21 Jul 2025
This manuscript presents a new high-resolution multidecadal reanalysis for the Black Sea (BLK-REA), incorporating key advancements in model resolution, lateral boundary conditions, and data assimilation techniques. The study delivers an extensive validation of the reanalysis and discusses its application to ocean monitoring indicators (OMIs) such as ocean heat content, the Cold Intermediate Layer (CIL), Rim Current variability, and meridional overturning circulation (MOC). These contributions are highly relevant and timely, particularly in the context of climate-related changes in regional semi-enclosed basins.
However, despite its technical strengths, the manuscript requires significant revision before it can be considered for publication. Several methodological and interpretative aspects need clarification, refinement, or deeper discussion. I recommend Major Revisions to address the concerns outlined in the detailed comments. My detailed comments are followed below:
- The manuscript states in line 178 that a “quasi-independent validation” was performed. However, for an objective and rigorous validation, the use of fully independent observational datasets is typically recommended. In particular, the use of observations that were excluded from data assimilation due to background quality control raises concern about possible contamination by erroneous data. For instance, in Figure 2, there are conspicuous, vertically extensive anomalies in temperature bias around 1998 and salinity bias in 1996, which could be indicative of problematic observational data rather than model performance. A discussion on the possibility of observation-induced artifacts and how such risks were mitigated is strongly recommended.
- In Figures 3 and 4, temperature and salinity biases in the Eastern region show a sudden increase below 700 m, which is unexpected. Generally, variability and errors tend to decrease with depth. The authors should examine whether these anomalies arise from insufficient data qulity control, issues in model initialization, or perhaps unresolved physical processes. An explanation or hypothesis for these unusual patterns would improve credibility.
- Although the Eastern region has the highest number of salinity observations (Figure 4, right), the RMSD and bias near the surface are larger than in other regions. The manuscript does not address this, and a discussion is warranted. Potential causes could include persistent local biases in atmospheric forcing, river discharge representation, or misrepresentation of mixing processes.
- The sentence “Bias values are in general small, and in particular in the western region” seems inconsistent with the plotted salinity bias in Figure 4, where large positive biases are evident near the surface in the western region. This discrepancy should be corrected or clarified.
- The description of how SLA is assimilated (e.g., through the dynamic height operator and use of MDT) could be elaborated further. It is unclear how the model handles vertical mapping and whether SLA assimilation is constrained only over deep waters. Furthermore, in Figure 5, the RMSD of SLA decreases after 2000-likely due to increased availability of satellite altimetry data. This hypothesis should be explicitly stated and supported with quantitative or qualitative evidence. The temporary increase in SLA RMSD around 2016 is notable but unexplained. Likewise, elevated RMSD near the coastline (Figure 6) may stem from the fact that SLA assimilation is performed only in areas deeper than 1000 m, thus leaving high-variability coastal regions unconstrained. A discussion of these spatial and temporal variations in SLA errors is necessary to clarify the model’s performance boundaries.
- The recovery of the Black Sea Meridional Overturning Circulation (MOC) after 2010 is intriguing, especially given the concurrent decline or near-disappearance of the Cold Intermediate Layer (CIL). Although the authors note this decoupling, a deeper discussion on alternative dynamics that might drive the MOC increase (e.g., changes in wind-driven circulation, lateral advection, or water mass transformation) would be beneficial. The potential use of age-tracer or Lagrangian particle tracking experiments to explore such processes could be suggested for future work.
- Much of the text after line 419 in the “Conclusions” section reads more like future planning than summarizing key findings. It may be clearer and more structured to rename this section “Discussion and Outlook”, where the manuscript first summarizes the conclusions, then outlines future improvements and data needs.
Citation: https://doi.org/10.5194/egusphere-2025-2114-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 255 | 65 | 14 | 334 | 11 | 25 |
- HTML: 255
- PDF: 65
- XML: 14
- Total: 334
- BibTeX: 11
- EndNote: 25
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1