| 13 Feb 2026
PolyU2025 SLA: A global 0.25°×0.25° monthly sea level anomaly dataset (1993–2024) determined from satellite altimetry for sea-level and climate change research
Abstract. Long-term and spatially consistent sea-level anomaly (SLA) products derived from satellite altimetry are fundamental for sea-level and climate change studies. In this study, we develop the PolyU2025 SLA (The Hong Kong Polytechnic University 2025 sea level anomaly), a new global monthly gridded SLA product generated using a fully independent data-processing framework. This product is provided on a regular 0.25° × 0.25° grid and spans the period from January 1993 to December 2024 and is intended to be updated regularly. The PolyU2025 SLA is evaluated through systematic intercomparisons with the Copernicus Climate Change Service (C3S) gridded SLA product as well as with independent tide-gauge observations. The results demonstrate a high level of consistency between PolyU2025 and C3S at global and basin scales, characterized by near-zero differences in global-mean SLA and statistically indistinguishable estimates of global mean sea level trends and accelerations, confirming that both products are suitable for climate-scale applications. At regional and short time scales, differences between the two products become more pronounced, particularly in dynamically active regions, and are mainly associated with differences in the representation of short-term and mesoscale variability. These differences reflect methodological trade-offs in data processing and spatial mapping rather than systematic biases. Overall, the PolyU2025 SLA provides a stable and consistent characterization of sea-level change from regional to global scales and serves as a complementary dataset to existing gridded SLA products, especially for long-term and climate-oriented sea-level studies and multi-product assessments of regional sea-level variability and uncertainty. The PolyU2025 SLA product is openly available at https://doi.org/10.5281/zenodo.17810525 (Yuan et al., 2025).
The paper describes a new sea-level dataset using observations from satellite altimetry, overall it is well written, and the quality meets the standards of ESSD. The sea-level community would applaud the release of this dataset, especially if it is updated routinely. Therefore, I would recommend it for publications.
Major comments:
Gridded sea level products, especially with global coverage, are essential for investigating sea level rise. The C3S product is a prominent dataset that is widely used by the community; the PolyU SLA data perform comparably to the C3S data, demonstrating its reliability, and potential applications. In particular, I believe that the PolyU data would contribute considerably to understanding the sea level rise along coast where the altimetry observations suffer from larger uncertainties; now we have more for further evaluations against tide gauges or other products.
Major suggestions:
Data processing: The mapping approach relies on the LSC method. Authors mentioned that the spatial distributions are important (lines 272-274). This raises a question from me, what if the distributions are uneven, how does it affect the spatial interpolation? More importantly, I concerned with the choice of parameters used in LSC, the relevant discussions on the effect of parameters on the final SLA are missing. Plus, authors claimed that the LSC is unbiased, what if the covariance matrix are not stable, do you consider regularizations? If so, I doubt the solutions are ‘unbiased’. Finally, I am also curious that how do authors implement the LSC, i.e., you estimate the covariance matrix globally, or regions by regions, the latter would require less computation resources but more time, although I am not sure how this affects the final solutions. So please provide more details on the mapping approach.
Comparison with C3S: Authors stated that the major (methodological) differences include the along-track data processing and crossover adjustments, and mapping approach. For instance, we can observe apparent differences from Figures 5 and 7, prominent regions include Kuroshio Current, Gulf Stream, and other dynamically active regions. But how much would two products differ over these regions, adding relevant time-series for comparison would help to understand.
Comparison with TG: authors compared their data to TG observations, instead of selecting the nearest grid point, they considered the grid point that shows the strongest correlation. This is a good choice. But how do authors define the search radius, it is ambiguous (line 445, is it 1 degree or a specific distance, e.g., 50 km). It is also ambiguous how authors compare SLA to TG in Figure 8. Do you remove the linear trends? If not, I would suggest authors compare them after removal of the trends, or better also remove seasonal changes. Please show us examples for both the highest variance reduction and the lowest (i.e., add time series for SLA and TG to figure 8).
Minor suggestions:
Line 22, what do you mean ‘a stable’ characterization?
Line 36, ‘sea level’ missing a hyphen, please use the consistent typing style;
Line 46, global or near-global coverage?
Line 55, what do you mean climate-scale? This term is cited multiple times, but I think it is ambiguous.
Line 90, with plans for continuous updates, you mean you are committed to updating?
Line 145, do authors consider removal of the TG associated with large trends? Because these TG stations may also reflect local changes that cannot be captured by SA. That’s why I recommend authors to remove the trends before comparison.
Line 158, what do you mean ‘SLA variable’
Line 164, why resampling reduces the measurement noise? In my experiences, use more data (especially their average) would reduce the noise.
Line 330, is there any evidence supporting the millimeter level uncertainty? How do you mean exactly, do the lone-term trends have a millimeter uncertainty?
Line 366, in studies ‘focusing’
Line 375, several studies have suggested that sea level trend estimates should consider color noise (e.g., Mu et al., 2025; https://doi.org/10.1029/2025GL117434). Mu et al. (2025) did consider a AR(1) model, but they used yearly data. The monthly data should consider AR(3) or AR(5). I don’t think this is a series issue, but authors should add more wordings, so readers can learn more.
Line 460, in figure 8, I would suggest authors to add several examples of time series for comparison.
Line 480, what is this ‘nested moving-average approach’, any terminology for this method.
Line 520, in figure 10, I would suggest authors to add histogram.
Line 580, in figure 12, it would be nice to show the average for those stations, may be along with their standard deviations.