Data from: NWCSAF high resolution winds (NWCSAF GEO-I HRW) stereo AMVs over the Atlantic Ocean
Data files
Sep 11, 2026 version files 1.88 GB
-
NWCSAF_GEO-I_HRW_Stereo_AMVs_Supplementary_v1.0.zip
1.88 GB
-
README.md
9.59 KB
Abstract
The “stereo height assignment method” developed for NASA and NOAA for GOES-R/ABI Atmospheric Motion Vectors (AMVs), a purely geometric method based on the parallax displacement observed by two different satellites, has been included in the NWCSAF geostationary AMV product (NWCSAF GEO-I HRW, High Resolution Winds) as an extension to its v2025 code, as an additional height assignment option for AMVs in the region jointly observed by MTG-I1 and GOES-East over the Atlantic Ocean.
A comparison has also taken place, based on ECMWF ERA5 reanalysis winds and EarthCARE Mie Attenuated Backscatter (MAB) cloud height, against the alternative height assignment method in the HRW algorithm: “Cross Correlation Contribution (CCC)”, based on the output of several NWCSAF GEO-I Cloud products (Cloud type, Cloud top temperature and height, Cloud microphysics) and with direct or indirect dependence on up to 16 NWP forecast variables.
Considering the EarthCARE MAB, for low-level clouds (< 2 km), both stereo and non-stereo heights appear to conform well with the visually identifiable cloud tops, although there are many places where the result is also ambiguous. For higher clouds, clear differences are certainly seen between stereo and non-stereo heights. Considering ECMWF ERA5 reanalysis winds, the most pronounced distinctions emerge also at high levels (> 6 km), where the mean biases of stereo and non-stereo AMVs remain similar, but the standard deviations generally favor the stereo method, reflecting improved precision in the upper-tropospheric flow where the parallax signal is strongest.
The updated HRW algorithm will be released to NWCSAF users in 2027 inside the following software package version (NWCSAF GEO-I v2027).
Dataset overview
This dataset contains figure products and supporting comparison outputs used to evaluate stereo and non-stereo NWC SAF High Resolution Winds (HRW) atmospheric motion vectors (AMVs) over the Atlantic Ocean. The dataset includes curtain-plot comparisons with ERA5 reanalysis, publication-quality PNG figures, and MATLAB .fig files that preserve the corresponding MATLAB figure objects for further inspection and reuse.
The files support analyses comparing stereo and non-stereo HRW wind retrievals, including comparisons of retrieved wind vectors and assigned heights. Users do not need the associated manuscript to navigate the dataset; the file groups and naming conventions are described below.
Description of the data and file structure
The submission contains 816 files in total. The major file groups visible in the submitted collection are described below.
1. Curtain-plot PNG files
The principal data products are PNG curtain plots with filenames of the form:
Curtain_Plot_MTGI_IR105_DD-Mmm-YYYY_HH-MM-SS_ERA5_latpXX.XX_lonmYY.YY.png
Examples:
Curtain_Plot_MTGI_IR105_03-Jan-2025_03-40-00_ERA5_latm24.25_lonm27.00.pngCurtain_Plot_MTGI_IR105_28-Mar-2025_17-00-00_ERA5_latp20.75_lonm43.50.png
These files provide individual atmospheric curtain-plot comparisons associated with MTG-I infrared-channel HRW retrievals and ERA5 reanalysis. Each filename identifies the observation date and time and the geographic location associated with the comparison.
Filename components:
Curtain_Plot— atmospheric curtain-plot comparison product.MTGI— Meteosat Third Generation Imager.IR105— infrared 10.5 µm channel used for the HRW product.DD-Mmm-YYYY_HH-MM-SS— date and UTC time of the comparison.ERA5— indicates that ERA5 reanalysis is included as the reference/comparison dataset.latpXX.XX/latmXX.XX— latitude, wherepdenotes positive (north) andmdenotes negative (south).lonpYY.YY/lonmYY.YY— longitude, wherepdenotes positive (east) andmdenotes negative (west)..png— publication-quality raster image that can be viewed with standard image software.
The curtain plots are intended to allow users to visually examine the vertical and spatial relationship between the HRW retrievals and the ERA5 atmospheric reference fields at the specified time and location.
2. Manuscript and analysis figures: PNG and MATLAB FIG files
Additional figures are provided in both .png and/or MATLAB .fig format. Examples include:
Fig4a.png/Fig4a.figFig4b.png/Fig4b.figFig5a.png/Fig5a.figFig5b.png/Fig5b.figFig6a.png/Fig6a.figFig6b.png/Fig6b.figFig7a.png/Fig7a.figFig7b.png/Fig7b.figFig8a.png/Fig8a.figFig8b.png/Fig8b.figFig13.figFig14.figFig15a.figFig15b.figtropics_IR3D.png/tropics_IR3D.fig
The numbered Fig* files are figure products associated with the study analyses. PNG versions are static, publication-quality images and can be opened using standard image-viewing software. MATLAB .fig versions preserve the MATLAB figure object and allow users with MATLAB to inspect the plotted elements, axes, and underlying figure structure and to modify or regenerate figure presentation.
Where both formats are supplied, the PNG and FIG files represent the same figure or analysis product in accessible raster and editable MATLAB formats, respectively.
tropics_IR3D.png and tropics_IR3D.fig provide the corresponding tropical-region infrared three-dimensional analysis/visualization in static and MATLAB figure formats.
3. ERA5 vector-difference comparison figures
Files beginning with MAVD_ contain the mean absolute vector difference (MAVD) comparison between stereo and non-stereo HRW winds and ERA5.
Examples:
MAVD_IR_0-2_28-Mar-2025.figMAVD_IR_2-6_28-Mar-2025.figMAVD_IR_6-20_28-Mar-2025.figMAVD_VIS_0-2_28-Mar-2025.figMAVD_VIS_2-6_28-Mar-2025.figMAVD_VIS_6-20_28-Mar-2025.fig
Filename components:
MAVD— mean absolute vector difference relative to ERA5.IRorVIS— infrared or visible satellite channel grouping.0-2,2-6, or6-20— atmospheric height range in kilometers.28-Mar-2025— date analyzed.
For the ERA5 comparison, wind speed and direction are converted to horizontal u and v wind components. HRW observations are grouped on a 0.25° grid, ERA5 winds are interpolated to the observation heights, and vector differences relative to ERA5 are calculated. The MAVD figures summarize these differences for stereo and non-stereo HRW retrievals.
4. Stereo/non-stereo height-difference figures
Files beginning with DH_ contain height-difference comparisons between the non-stereo and stereo HRW products.
Examples:
DH_IR_0-2_28-Mar-2025.figDH_IR_2-6_28-Mar-2025.figDH_IR_6-20_28-Mar-2025.figDH_VIS_0-2_28-Mar-2025.figDH_VIS_2-6_28-Mar-2025.figDH_VIS_6-20_28-Mar-2025.fig
DH represents the mean height difference calculated as:
non-stereo HRW height − stereo HRW height
The files are separated by satellite channel (IR or VIS) and atmospheric height interval (0–2 km, 2–6 km, and 6–20 km). The figures also include the sample counts used in the comparison.
ERA5 comparison methodology represented by the files
For the full-day ERA5 comparison on 28 March 2025, the analysis:
- Uses ERA5
u,v, and geopotential-height fields. - Separates HRW observations by satellite channel:
- VIS = channel 2
- IR = channel 16
- Separates the stereo HRW observations into three height ranges:
- Low: 0–2 km
- Mid: 2–6 km
- High: 6–20 km
- Searches each UTC hour for an available stereo/non-stereo HRW file pair.
- Converts HRW wind speed and wind direction to
uandvcomponents. - Bins observations to a 0.25° geographic grid and retains grid cells containing both products.
- Interpolates ERA5 winds to the HRW observation heights.
- Computes absolute wind-vector differences relative to ERA5.
- Produces MAVD and HRW height-difference comparison figures for each channel and height interval.
File formats and reuse
PNG (.png)
PNG files are standard raster images intended to maximize accessibility and ease of reuse. They can be opened with common image viewers, web browsers, image-analysis packages, and scientific software without proprietary software.
MATLAB FIG (.fig)
MATLAB .fig files contain saved MATLAB figure objects corresponding to analysis and manuscript figures. They are provided in addition to accessible PNG versions where available so that users can inspect and manipulate the figure objects in MATLAB. MATLAB is required to open .fig files directly.
The .fig files are figure products rather than raw satellite or ERA5 source-data files.
Terminology
- AMV — Atmospheric Motion Vector: A wind estimate derived by tracking atmospheric features between sequential satellite images.
- HRW — High Resolution Winds: The NWC SAF satellite wind product used in this study.
- Stereo AMV: An atmospheric motion vector for which stereoscopic information is used to improve the determination of feature height.
- Non-stereo AMV: The corresponding HRW wind retrieval without the stereoscopic height determination used in the stereo product.
- ERA5: Global atmospheric reanalysis used here as an independent atmospheric reference for wind comparisons.
- MAVD — Mean Absolute Vector Difference: A measure of the magnitude of the vector wind difference between the satellite-derived HRW winds and ERA5.
- DH — Height Difference: Difference in assigned atmospheric height between the non-stereo and stereo HRW retrievals, calculated as non-stereo minus stereo.
- IR: Infrared satellite channel.
- VIS: Visible satellite channel.
- MTG-I: Meteosat Third Generation Imager.
Notes for users
The PNG products are the most broadly accessible versions of the visual data products and should be used when MATLAB is not available. MATLAB .fig files are supplied as supporting editable figure objects.
The dataset contains processed comparison and figure products rather than the complete raw source holdings from the satellite instruments or ERA5 archive. The filenames preserve key metadata—including date, UTC time, geographic coordinates, channel grouping, and/or atmospheric height range—to allow the products to be interpreted and navigated independently.
Sharing/Access information
Data products in this dataset were derived from NWC SAF GEO-I High Resolution Winds products and ERA5 reanalysis. The complete source satellite and ERA5 holdings are not reproduced in this Dryad dataset.
Associated analysis software is archived separately on Zenodo:
The software source is also available through the associated GitHub repository:
Code/Software
Software used to generate and analyze the NWC SAF GEO-I HRW figure products is archived separately from this dataset. Users seeking to reproduce or modify the analysis should consult the associated software archive and study documentation.
