the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 30 Sep 2025
A 1 km Hourly High-Resolution 3D Wind Field Dataset over the Yangtze River Delta Incorporating Dynamical Downscaling, Observational Assimilation, and Land Use Updates
Abstract. High-resolution three-dimensional (3D) wind field data are critical for a wide range of applications, including wind energy assessment, low-altitude aviation, air quality modeling, and extreme weather forecasting. Although ERA5 reanalysis remains widely used, its relatively coarse spatial resolution (~31 km) limits its ability to capture local-scale atmospheric processes. To address this, this study develops an hourly 3D dynamic wind field dataset with 1 km horizontal resolution covering the Yangtze River Delta (YRD) region during the summer months (June–August) from 2021 to 2023, namely YRD1km, generated through advanced dynamical downscaling of ERA5 using a customized Weather Research and Forecasting (WRF) model configuration. The methodology integrates multi-source observational nudging with high-resolution land use parameterization to enhance near-surface wind accuracy and terrain-induced flow representation, particularly in urban clusters and mountainous areas. Validation against ground-based observations confirms the superior performance of YRD1km over ERA5 for hourly 10-m wind components, with Mean Absolute Error (MAE) reduced by approximately 22 % for U and 26 % for V, Root Mean Square Error (RMSE) reduced by 18 % for U and 23 % for V, and Nash–Sutcliffe Efficiency (NSE) improved by 33 % and 40 %, respectively. On a daily mean basis, both MAE and RMSE are reduced to below 0.4 m/s, and NSE reaches approximately 0.88. Spatially, YRD1km captures finer spatial wind speed gradients and localized terrain-induced circulations that are not captured by ERA5. Temporally, consistent accuracy improvements with approximately 20 % lower hourly error variability are seen when compared to ERA5. Vertically, 42.2 % accuracy gains are observed in the near-surface layer when compared with radiosonde profiles. Moreover, in a representative convective storm case, YRD1km captures multi-level wind structures that are closely linked to the initiation and continuous development of deep convection, highlighting its diagnostic advantage in high-impact weather events. Overall, the YRD1km 3D wind field dataset and its integrated methodological framework provide a robust foundation for regional meteorological applications, including high-resolution AI-based forecasting, renewable energy planning, and weather risk management in rapidly developing regions such as the YRD. The YRD1km 3D wind field dataset is available at https://doi.org/10.57760/sciencedb.23752 (Zhang et al., 2025).
- Preprint
(19502 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on essd-2025-419', Anonymous Referee #1, 20 Oct 2025
-
AC1: 'Reply on RC1', Yan-An Liu, 24 Dec 2025
Comments:
1. The manuscript evaluates the dataset mainly using MAE / RMSE / NSE, but the verification metrics used in Figures 5 and 6 appear inconsistent with those used elsewhere. Unless there is a justified reason for deviation, please ensure that the verification metrics are consistent throughout the manuscript. If this is not possible, please provide an explanation in the methods section or the figure captions.
Thank you for this thoughtful comment. In our original manuscript, we aimed to maintain consistency in the statistical metrics and initially calculated and plotted MAE, RMSE, and NSE. Upon analyzing their temporal variations, we observed that MAE and RMSE exhibited highly similar trends. To present the results more clearly, we therefore retained only MAE and NSE in the figures. A similar relationship was observed for the vertical validation in Figure 6.
We fully understand that this simplification may have caused confusion, and we sincerely appreciate your suggestion for greater consistency. In response, we have revised Figures 5 and 6 accordingly. Specifically, RMSE has been added to Figure 5, and Bias in Figure 6 has been replaced by MAE to align with the evaluation framework used elsewhere in the manuscript.
Citation: https://doi.org/10.5194/essd-2025-419-AC1 -
AC2: 'Reply on RC1', Yan-An Liu, 25 Dec 2025
We sincerely thank the reviewer for the thoughtful and constructive comments. We have carefully considered each point and revised the manuscript accordingly. Detailed responses to all comments are provided below and can be found in the attached file.
-
AC1: 'Reply on RC1', Yan-An Liu, 24 Dec 2025
-
RC2: 'Comment on essd-2025-419', Anonymous Referee #2, 07 Jan 2026
The authors have created a high-resolution, dynamically downscaled dataset over the Yangtze River Delta called YRD1km, based on the WRF model, dynamically downscaled from the ERA5 reanalysis. The dataset is then further refined using a hybrid nudging approach, combining observational nudging and analysis nudging, to produce a generally skillful 3D wind dataset. The authors verified the dataset against surface observations primarily for June 2022, as well as against radiosonde observations, and for a convective case study event, also in June 2022. In general, the manuscript is well written and presented, but the analysis, methodology, and conclusions drawn from this work (and particularly the verification) are incomplete. I do believe this work is valid and important, but I recommend major revisions and inclusion of additional detail + verification work prior to publication.
Scientific Comments:
- Though the dataset is valid for the summer months of 2021-2023, most of the verification was conducted in June 2022, with the first few figures and tables focusing solely on June 1, 2022. Additionally, the time period for the daily verification statistics for Table 3 is unclear; also, the time series for Figure 5 wasn't clear (are those averaged MAE/NSE across all stations?). Though I don't doubt that 1-km WRF would generally outperform the 0.25-deg ERA5 for winds, the shown verification statistics should be more transparent and comprehensive; otherwise it feels like "picking and choosing".
- The model setup is not particularly clear, e.g. the justification of the physics parameterizations based on sensitivity testing is dubious. Those schemes are the baseline "default" schemes for the WRF model; though I understand that extensive testing of physics schemes would be unfeasible and beyond the scope of this work, many other dynamical downscaling related articles have discussed this and have either used other schemes and/or have provided justification for why they chose the schemes that they did. Additionally, the nesting and initialization setups are unclear, i.e. was two-way nesting used? Did all three domains start at the same time, meaning that the ERA5 initial conditions would be interpolated directly to the 1-km grid? Is a 1 hour spinup really adequate? Are there discontinuities/"jumps" in the data every 6 hours?
- How much of the skillful verification statistics are due to the WRF model/enhanced spatial resolution, rather than an "edit" of grid point values based on observation nudging? If WRF grid points were nudged on the hour, and verified on the hour (and on the same grid points as the nudging? unclear in the manuscript), then there would be a reduction in error metrics via that "edit".
It would have been nice to see a baseline configuration with WRF being run without any nudging at all, to actually determine how much the nudging improved skill.
Additionally, it wasn't clear how the obs nudging was fully conducted, i.e. what variables were nudged (was it just wind?). Would the other variables (e.g. 2-m temperature, precipitation) be compromised because of this nudging, i.e. is this dataset only useful for wind, and not for other variables?
- Are there not other studies/datasets with high-resolution downscaling over the YRD, that could be compared against? Purely from an interpolation/scale standpoint, of course the YRD1km dataset would be better able to resolve the winds than ERA5, especially if ERA5 grid points were interpolated to the same observation points for verification (i.e. multiple obs stations within the same ERA5 grid box would have the same ERA5 wind interpolated to them, but would have different WRF winds from different WRF grid points).
This ties into the utility of the dataset and the claims of such utility within the article. Statements like "The scale-dependent improvements emphasize the application value of YRD1km for both short-term weather monitoring and long-term climate analyses in the YRD region" are not backed up within the article. There just isn't enough verification conducted, over long enough time periods + seasons, against other competitive datasets, to make such claims. In particular, YRD1km isn't useful for "short-term weather monitoring" because it is dynamically downscaled from a reanalysis dataset (i.e. not useful as a forecast or nowcast dataset, only hindcasting), nor is it long enough yet for long-term climate analyses. I would recommend the authors be more specific about what the YRD1km can be used for, e.g. as a future training dataset for AI-based downscaling (with a longer dataset), or as input into high-resolution air-quality dispersion modelling or even sub-kilometer scale downscaling (down to LES scales), for specific applications (these were stated in the conclusion, but could be made more explicit and specific in other parts of the article).
Minor Comments:
- Should the citation/URL of the dataset be listed directly in the abstract? (not sure on this, please double check)
- Statements like "high spatiotemporal resolution is fundamental to modern meteorological services, wind energy development, and the
safe operation of low-altitude economy..." should be backed with citations, even if seemingly self-evident- MODIS2001 ---> did you mean MODIS2010?
- Inconsistencies with abbreviations, e.g. LU-ESA2020 in the text but LUT-ESA2020 in the table
- "Figures 5 presents" --> line 360
Citation: https://doi.org/10.5194/essd-2025-419-RC2 -
AC3: 'Reply on RC2', Yan-An Liu, 17 Jan 2026
We sincerely thank the reviewer for the careful assessment of our manuscript and for recognizing the scientific value and relevance of the YRD1km dataset. We appreciate the constructive comments regarding the completeness of the analysis, methodology, and validation, particularly with respect to the verification strategy.
In response to these concerns, we have substantially revised the manuscript by adding new experiments, expanding the validation period and scope, and clarifying key methodological details. The revised version includes more comprehensive and transparent verification analyses, as well as clearer descriptions of the model configuration and evaluation framework. Below, we provide detailed, point-by-point responses to each comment and describe the corresponding revisions made to the manuscript.
Detailed responses to all comments are provided below and can be found in the attached file.
-
AC3: 'Reply on RC2', Yan-An Liu, 17 Jan 2026
-
EC1: 'Comment on essd-2025-419', Qingxiang Li, 11 Jan 2026
Both reviewers have requested major revisions. The manuscript therefore requires substantial revision in response to the reviewers’ comments, with a point-by-point reply to each comment. After this, it will be sent back to the reviewers for a second round of review.
Citation: https://doi.org/10.5194/essd-2025-419-EC1
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 502 | 242 | 34 | 778 | 31 | 36 |
- HTML: 502
- PDF: 242
- XML: 34
- Total: 778
- BibTeX: 31
- EndNote: 36
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The authors have developed a high-resolution 3D wind field dataset (YRD1km) over the Yangtze River Delta by running WRF driven by ERA5 reanalysis, assimilating observations, and updating land-use information. This dataset addresses a significant lack of high-resolution, 3D wind products in this important region during the summer months. However, the manuscript still has several structural and methodological issues. Therefore, I recommend that it be considered for publication only after major revision.
Comments: