A high spatiotemporal resolution atmospheric motion vector dataset from Fengyun-4B Geostationary High-speed Imager observations for 15 tropical cyclones over the western North Pacific and South China Sea (2022–2025)
Abstract. High spatiotemporal resolution wind observations are essential for resolving the rapidly evolving inner-core circulation, spiral rainbands, upper-level outflow, and asymmetric structural changes of tropical cyclones (TCs), yet these features remain inadequately sampled by conventional observing systems and operational atmospheric motion vector (AMV) products. This study presents a consistently processed and systematically evaluated minute-scale temporal and mesoscale spatial atmospheric motion vector (MAMV) dataset derived from the Geostationary High-speed Imager (GHI) onboard Fengyun-4B (FY-4B) for 15 TCs over the western North Pacific and the South China Sea (SCS) during 2022–2025. The dataset includes daytime visible-band (VIS) MAMVs with a nominal vector spacing of 3 km and day-and-night infrared-band (IRX) MAMVs with a nominal spacing of 24 km. Four stable products are generated at the 1, 2, 6, and 7 min marks of each recurring 10 min observation cycle. The dataset represents the highest combined spatial and temporal resolution currently available for a multi-case geostationary satellite TC wind-field dataset, providing detailed observations of inner-core flow, spiral-rainband circulation, convective-cloud boundaries, peripheral winds, and upper-tropospheric outflow that are difficult to obtain from radiosondes, conventional AMVs, or hourly reanalysis fields. Quality-screened vectors were evaluated against radiosonde observations and ERA5 reanalysis winds. The mean speed bias, mean vector difference, and standard deviation of vector differences were 2.34, 4.05, and 1.77 m s⁻¹ for VIS, 2.21, 4.74, and 2.17 m s⁻¹ for IRX when using radiosondes as reference, they are 1.62, 5.57, and 4.63 m s⁻¹ for VIS, 0.31, 5.95, and 3.66 m s⁻¹ for IRX when using ERA5 as reference. The dataset provides a new observational basis for continuously monitoring the multiscale evolution of tropical cyclones, including changes in inner-core circulation, spiral rainbands, asymmetric wind structures, peripheral flow, and upper-tropospheric outflow, thereby supporting improved analysis of TC development, structural change, and short-term intensity evolution. The FY-4B/GHI IRX MAMV dataset for 2022–2025 is available at doi:10.5281/zenodo.22790846; The FY-4B/GHI VIS datasets are available at doi:10.5281/zenodo.22791421 for 2022–2023, doi:10.5281/zenodo.22791444 for 2024, and doi:10.5281/zenodo.22791469 for 2025 (Xia et al., 2026a, b, c, d).
Review of “A high spatiotemporal resolution atmospheric motion vector dataset from Fengyun-4B Geostationary High-speed Imager observations for 15 tropical cyclones over the western North Pacific and South China Sea (2022–2025)”
This manuscript presents a multi-case dataset of rapid-scan atmospheric motion vectors (MAMVs) derived from the FY-4B Geostationary High-speed Imager (GHI) for 15 tropical cyclones (TCs) over the western North Pacific and South China Sea during 2022–2025. There are two products in this dataset. One is the daytime visible (VIS) product from 250 m pixels with a nominal 3 km vector spacing; the other is the day-and-night 11.4 µm infrared (IRX) product from 2 km pixels with a nominal 24 km spacing, both generated four times per 10-min observation cycle. The vectors are evaluated against radiosondes and ERA5, with case-mean mean vector differences (MVDs) of approximately 4–6 m s⁻¹, and the data are openly available.
The dataset is relevant to ESSD. GHI rapid-scan imagery of TCs is not widely accessible, and a consistently processed, multi-storm AMV archive would be valuable for TC structure studies, model evaluation, and data assimilation research. However, in my assessment, the manuscript and dataset do not yet meet the ESSD criteria for data quality and usability, and I therefore recommend major revision at this stage.
Firstly, the validation does not support the stated accuracy. The radiosonde comparison is dominated by pseudo-replicated pairs, ERA5 is not an independent reference, and nothing is validated as a function of storm-relative radius, although the inner core and rainbands are the headline applications of the dataset. Secondly, the deposited files contain placeholder heights (Height = 0 m with Pressure = 999 hPa), fill speeds of 999.9 m s⁻¹, and speeds exceeding the tracking limit of the search box, some of which pass the quality control recommended by the authors. These values are not documented and may bias the low-level statistics and mislead users.
Furthermore, the “3 km”, “24 km”, and “minute-scale” descriptors describe output sampling rather than effective resolution, and the novelty claim does not acknowledge previous Himawari-8/9 rapid-scan TC AMV studies and datasets (Tsukada and Horinouchi, 2020; Tsujino et al., 2021; Horinouchi et al., 2023; Tsukada et al., 2024; Tsujino, 2026). The change in satellite position from 133°E to 105°E in 2024 is also not documented. Detailed comments and suggestions are listed below.
The validation would be more convincing if it were based on independent samples, and if the number of independent soundings behind each result were transparent to the reader.
An evaluation against references that are independent of ERA5 and that sample the storm environment (e.g., dropsondes, wind profilers, or scatterometer winds), together with an assessment of how the accuracy varies with distance from the storm centre, would be needed to support the claimed applications. A comparison with existing operational AMV products over the same scenes would also help to demonstrate the added value of GHI.
More importantly, an inspection of sample files from the deposit shows that a notable fraction of vectors passing the recommended quality control (QI > 0.65, R > 0.8, and speed ≥ 4 m s⁻¹) have Height = 0 m, many of them with Pressure = 999 hPa exactly (about 3% of the IRX vectors on 26 August 2025, and 7–11% of the VIS vectors on 24 August 2025 and 14 July 2023). These appear to be fallback values rather than retrieved heights. If included in the validation, they would be classified as low-level vectors, which may contribute to the largest low-level bias against ERA5 (2.18 m s⁻¹, Line 583). In addition, although the README states that invalid values are NaN, the IRX files contain fill speeds of 999.9 m s⁻¹.
Furthermore, some QC-passing vectors exceed the maximum speed that can be tracked with the stated search configuration (Line 278–282), i.e., approximately 50 m s⁻¹ for VIS and 93 m s⁻¹ for IRX. For example, VIS speeds reach 78 m s⁻¹, and several IRX vectors over land on 26 August 2025 exceed 100 m s⁻¹ at 566–846 hPa. This points to an inconsistency either in the described configuration or in the time intervals used for the speed calculation. These issues should be clarified, and the treatment of such values should be transparent to users.
The claim that the dataset “represents the highest combined spatial and temporal resolution currently available for a multi-case geostationary satellite TC wind-field dataset” (Line 45–46) also needs to be placed in the context of existing work, none of which is cited. Himawari-8/9 target-area (2.5-min) and 30-s imaging have been used to derive inner-core AMVs (Tsukada and Horinouchi, 2020; Tsujino et al., 2021; Horinouchi et al., 2023), a novel eye-region AMV derivation was validated against dropsondes (Tsukada et al., 2024), inner-core AMVs for Typhoon Neoguri (2025) are publicly available (Tsujino, 2026), and GOES-16/17 mesoscale-sector vortex-scale AMVs have been developed for hurricane applications (Stettner et al., 2019). The novelty claim should be reconsidered in light of this work, and the specific contribution of this multi-storm FY-4B/GHI dataset should be stated more precisely.
Minor comments:
Reference:
Hersbach, H., Bell, B., Berrisford, P., Hirahara, S., Horányi, A., Muñoz-Sabater, J., Nicolas, J., Peubey, C., Radu, R., Schepers, D., Simmons, A., Soci, C., Abdalla, S., Abellan, X., Balsamo, G., Bechtold, P., Biavati, G., Bidlot, J., Bonavita, M., De Chiara, G., Dahlgren, P., Dee, D., Diamantakis, M., Dragani, R., Flemming, J., Forbes, R., Fuentes, M., Geer, A., Haimberger, L., Healy, S., Hogan, R. J., Hólm, E., Janisková, M., Keeley, S., Laloyaux, P., Lopez, P., Lupu, C., Radnoti, G., de Rosnay, P., Rozum, I., Vamborg, F., Villaume, S., and Thépaut, J.-N.: The ERA5 global reanalysis, Q. J. Roy. Meteor. Soc., 146, 1999–2049, https://doi.org/10.1002/qj.3803, 2020.
Horinouchi, T., Tsujino, S., Hayashi, M., Shimada, U., Yanase, W., Wada, A., and Yamada, H.: Stationary and transient asymmetric features in tropical cyclone eye with wavenumber-1 instability: Case study for Typhoon Haishen (2020) with atmospheric motion vectors from 30-second imaging, Mon. Weather Rev., 151, 253–273, https://doi.org/10.1175/MWR-D-22-0179.1, 2023.
Stettner, D., Velden, C., Rabin, R., Wanzong, S., Daniels, J., and Bresky, W.: Development of enhanced vortex-scale atmospheric motion vectors for hurricane applications, Remote Sens., 11, 1981, https://doi.org/10.3390/rs11171981, 2019.
Tsujino, S.: Inner-core atmospheric motion vectors in Typhoon Neoguri (2025), Zenodo [data set], https://doi.org/10.5281/zenodo.18870924, 2026.
Tsujino, S., Horinouchi, T., Tsukada, T., Kuo, H.-C., Yamada, H., and Tsuboki, K.: Inner-core wind field in a concentric eyewall replacement of Typhoon Trami (2018): A quantitative analysis based on the Himawari-8 satellite, J. Geophys. Res.-Atmos., 126, e2020JD034434, https://doi.org/10.1029/2020JD034434, 2021.
Tsukada, T. and Horinouchi, T.: Estimation of the tangential winds and asymmetric structures in typhoon inner core region using Himawari-8, Geophys. Res. Lett., 47, e2020GL087637, https://doi.org/10.1029/2020GL087637, 2020.
Tsukada, T., Horinouchi, T., and Tsujino, S.: Wind distribution in the eye of tropical cyclone revealed by a novel atmospheric motion vector derivation, J. Geophys. Res.-Atmos., 129, e2023JD040585, https://doi.org/10.1029/2023JD040585, 2024.
Wang, M., Zhang, W., Cui, Y., and Li, B.: Evaluation of FY-4B surface shortwave radiation products over China: Performance improvement induced by the orbital drift from 133°E to 105°E, Remote Sens., 18, 1454, https://doi.org/10.3390/rs18101454, 2026.