Five-minute average horizontal wind velocity data combined from both sensors (which has been corrected for air-flow distortion) from the Antarctic Circumnavigation Expedition (ACE) 2016/2017 legs 0 to 4.

***** Dataset abstract *****

The horizontal wind velocity data from the Antarctic Circumnavigation Expedition (ACE) 2016/2017 legs 0 to 4 has been corrected for air-flow distortion.
The measurements from both the port and starboad side anemometer were averaged to five-minute resolution and have been combined via vector averaging of the data.
The ten meter neutral wind speed (U10N) has been estimated using ERA-5 surface heat fluxes, which were interpolated onto the ship's track, and the COARE 3.5 drag coefficient.
This data set provides a continous and high-resolution record of the wind speed and direction near to the ship's location.

***** Original data collection *****

The intermediate processing steps and files which take us from the input datasets (10.5281/zenodo.3801718, 10.5281/zenodo.3772895, 10.5281/zenodo.3832045, 10.5281/zenodo.3831980, 10.5281/zenodo.3836185, 10.5281/zenodo.3836200, 10.5281/zenodo.3836211) to this output dataset are summarised in the python scripts ace_flowdistortion.py and read_ace_data.py, which are published in Landwehr, 2020 (DOI: 10.5281/zenodo.3836050). The scripts use utilities from the pyantarctica package, which are published in Volpi and Landwehr, 2020 (DOI: 10.5281/zenodo.3827478).

***** Data processing *****

All steps to produce this quality-checked data file are contained in the python script ace_flowdistortion.py (Landwehr, 2020; DOI: 10.5281/zenodo.3836050) and are fully described in Landwehr et al. (2019; DOI: https://doi.org/10.5194/amt-2019-366). 
The main steps are as follows:
- Horizontal wind speed and direction observations relative to the platform are filtered for outliers and corrected for air-flow distortion using the flow distortion bias factors from the dataset Landwehr et al., 2020 (DOI: 10.5281/zenodo.3836200)
- The data are vector-averaged to one-minute resolution.
- The ship's cruise track, velocity, course over ground and heading at one-minute resolution are used to compute the true wind speed according to Smith, Shawn R., Mark A. Bourassa, and Ryan J. Sharp. ‘Establishing More Truth in True Winds’. Journal of Atmospheric and Oceanic Technology 16 (1999): 14.
- True and relative wind data are vector-averaged to five-minute resolution and averaged between both sensors.

The repository, which was used to produce this data, can be checked out at: https://renkulab.io/projects/ACE-ASAID/wind-speed-correction

Python3.7.3 was used to run the code. All packages and versions can be found in requirements.txt in Landwehr, 2020 (DOI: 10.5281/zenodo.3836050) and the repository used to create this dataset.

***** Quality checking *****

The data processing included basic methods of outlier detection based on the difference of the individual observations to the rolling median and the removal of implausible values (wind_speed_relative_to_platform > 100 m/s, wind_speed > 40 m/s). Outliers are removed prior to the time averaging.

***** Standards *****
 
This dataset does not conform to any data formats or standards. 

***** Further information for interpreting the data and using the dataset *****

Timescales: The original wind record was at 0.33 Hz resolution. The averaging to five-minute resolution removes the instantaneous platform motion signal, which has not been accounted for in the horizontal mean-motion correction.

Interpolation: The wind observations are suspect to turbulent motion at time scales between seconds to one hour. Therefore this data should not be interpolated.

Aggregation to lower temporal resolution: No problem but correct vector averaging should be done using (eastward_wind, northward_wind) and (bowward_relative_wind, portward_relative_wind).

***** Dataset contents *****

- wind-observations-port-stbd-corrected-combined-5min-legs0-4.csv, data file, comma-separated values
- data_file_header, metadata, text format
- README.txt, metadata, text format

***** Dataset contact *****

Sebastian Landwehr, Ecole Polytechnique Federale de Lausanne, Switzerland. ORCID: 0000-0003-2622-577X. Email: sebastian.landwehr@gmail.com

***** Dataset license *****

This five-minute averaged wind velocity dataset is made available under the Creative Commons Attribution 4.0 International License (CC BY 4.0) whose full text can be found at https://creativecommons.org/licenses/by/4.0/

***** Dataset citation *****

Please cite this dataset as:
Landwehr, S., J. Thomas and J. Schmale. (2020). Five-minute average horizontal wind velocity data combined from both sensors (which has been corrected for air-flow distortion) Antarctic Circumnavigation Expedition (ACE) 2016/2017 legs 0 to 4. (Version 1.0) [Data set]. Zenodo.