A turbulence data reduction scheme for autonomous and expendable profiling floats: datasets

Dataset associated with the following two papers:

• Hughes et al. – A turbulence data reduction scheme for autonomous and expendable profiling floats.  Ocean Sci. doi:10.5194/os-19-193-2023
• Moum et al. – Flippin’ χSOLO, an upper ocean autonomous turbulence profiling float.  
  J. Atmos. Oceanic Tech. doi:10.1175/JTECH-D-22-0067.1

The description below is also included in _fcs_data_readme.pdf

During field tests in 2019, we deployed two FCS floats (units 4002 and 4003). Between the two, approximately 650 profiles were taken. For each profile, we are archiving five configurations of the data. We are providing these in both Matlab .mat format and NetCDF .nc format. The content is the same, but the formatting differs.

For the Matlab files, each of the five configurations is contained within a struct within the same Matlab .mat file. For the NetCDF files, the variable names are prepended with the names of the structs. For example avg.S1 in the Matlab file corresponds to avg_S1 in the NetCDF file.

To facilitate downloading from Zenodo, the files are grouped into days (May 13, 14, 15, 16, and 17) and direction (UP and DN).

In the Matlab files, blk, phi, avg, and avgr contain readmes with full details about the individual quantities that were measured or derived. In the NetCDF files, these details are attributes associated with each variable.

The structs avg and avgr are similar, but avg was derived with a standard processing approach, whereas avgr was derived using the data reduction scheme described in the paper. The number of depths bins may differ between the two due to differences in the calculation of W_min and hence the identification of the end of the profile. Further, avgr (but not avg) includes bad values of ε and χ in the lowest tens of meters of upward profiles when the ballast pump is running. (The data reduction scheme does not include a method to identify when the ballast pump is on within individual profiles. Therefore, we have left the bad values in avgr. However, the bad values are easily identified in post-processing by comparing successive casts.)

Both blk and blkr include shear data after deskiping. The despiking routines, however, are not the same. See Appendix E of the associated paper for details.

Each Matlab file also includes the struct head, which is a header file that includes calibration coefficients.

Each filename is of the form UD_YYYYMMDDhhmmss_00X.ext where UD is UP or DN, YYYYMMDDhhmmss is the time stamp when the raw file was first created, 00X is 001 for unit 4002 and 002 for unit 4003, and .ext is either .mat or .nc.

Sea surface height spectra are contained within a substruct of avgr within the DN profiles in the Matlab files.

Some files, especially early ones, may not contain any values in avg or avg. These can be ignored.