• Home
  • NOAA ESRL Tropical Dynamics and Climate

TDC: Wind Profiler Database: Gridded File Format Description

Home Data Description Data Inventory Data Download Contacts

Data Interpolation

To produce the gridded wind profilerdata files, we begin with the data that has been through the Aeronomy Lab's standard "TOGA-COARE" wind profiler data processing, which produces half-hourly-averaged wind profiles.These profiler wind profilesare collected with a "native" vertical spacing that varies considerably: with the operating mode, with the site, and with time at most sites as operating parameters or equipment are changed. Placing these data on a uniform vertical grid makes some types of analysis easier, especially those involving long time series or multiple sites. However, there is always the danger of introducing spurious artifacts into a data set during this process, since interpolation uses samples of data to reconstruct a presumed function and then resamples that "function" at the new points.

We have recently gridded most of our archive of profiler winds using a fairly conservative set of procedures that we believe retains most of the information originally observed by the wind profilers without creating data. The resultant wind profiles are on a vertical grid with a 100m spacing that starts at 100m ASL. Key points are as follows: Only data deemed to be research quality have been used. Data are interpolated between 2 consecutive original levels, but are not extrapolated. With one exception, the scheme uses a cubic spline. We chose a cubic spline becase it is one of the smoothest interpolating functions and we believe it minimizes artifacts. The exception is that for winds collected in a mode with a pulse length >= 150m, if a solitary observation exists (i.e. there are missing observations above and below it in the original profile), its value is assigned to the nearest new gridpoint if that gridpoint is no more than 50m away - data collected simultaneously (by collocated profilers, or by interleaved operating modes of one profiler) but using different scanning volumes have not been merged - no attempt has been made to interpolate in time The results have been thoroughly tested at the Aeronomy Laboratory; a manuscript describing the details and the testing is in preparation. Users who wish to be less restrictive are always welcome to download the original data and apply their own gridding algorithms.

Gridded File Format

The gridded data is stored in ASCII text files. Each file consists of two sections: a single header section followed by a singledata section.

Header Section

View a sample header section. Each line of the header begins with a "#" character in column 1, followed by a description of the contents of theheader line, followed by a colon.Thecontents ofthe header information beginat column 21.The first header linegives the total number of header linesfor this file. Following this aredescriptive lines about thestation, the profiler, andthe processing.

At the very end of the header section there may be zero or more comment lines (beginning with "#Comment:"),which contain text strings of additional information.

The header lines that begin with "#Data column:" describe the data fields (columns) that appear in the data section. This header line gives four comma-separated pieces of information:

  1. the column number from left to right,
  2. a label containing the quantity being displayed and (in parentheses) the units of measurement,
  3. a FORTRAN-like format specifier, and
  4. the value that is displayed when there is no data for this quantity.
So, for example, the header line
#Data column:11, u (m/s), f7.2, 9999.00
means that column 11 will contain values for u, given in m/s; will occupy 7 columns, with 2 digits to the right of the decimal point; and will indicate missing data by the value 9999.00. Note that there will be an additional space character between each column of data.

Data Section

Here is more detail on the specific data columns in the gridded data file:

  1. 3-character station abbreviation
  2. 1-character code for operating mode ('a'=915-MHz high mode, 'b'=915-MHz low mode, 'e'=50-MHz)
  3. North latitude
  4. East longitude
  5. Year (dates/times, in UTC, arefor thebeginning of each (half-hour) averaging period)
  6. Day of year (1-366)
  7. Hour
  8. Minute
  9. Number of seconds since Jan. 1, 1970
  10. ht, the height, in meters above sea level, for quantities measured by obliquely pointing radar beams
  11. u, eastward component of horizontal wind
  12. v, northward component of horizontal wind
  13. wid1, average Dopplerspectral width, in m/s, of radar signalreceived by oblique beam #1
  14. wid2, average Dopplerspectral width, in m/s, of radar signal received by oblique beam #2
  15. snr1, average signal-to-noise ratio, in dB, of the radar signal received by oblique beam #1
  16. snr2, average signal-to-noise ratio, in dB, of the radar signal received by oblique beam #2
  17. n12,number of measurements making up the half-hour average ofoblique beam quantities
  18. sumwt12, an internal diagnostic statistic from the interpolation procedure
  19. wht, the height in meters above sea level for quantities measured by the vertically pointing radar beam
  20. w, vertical component of the wind
  21. wid3, average Doppler spectral width, in m/s, of radar signal received by vertical beam
  22. snr3, average signal-to-noise ratio, in dB, of the radar signal received by the vertical beam
  23. n3, number of measurements making up the half-hour average ofvertical beam quantities
  24. sumwt3, an internal diagnostic statistic from the interpolation procedure

The data section of the gridded data fileis organized as follows. There is one row of data for each height for each averaging time period. The first time period is given by the "#Start time:" header line. For each time period the heights will start at 100 m above sea leveland continue at 100 m intervalsas long as there was valid data. Following the last height for each time period is the 100 m height data for the next half-hour time period. Different time periods can have different numbers of heights with valid data. The header line "#Most heights:" gives the maximum number of heights in any one data period throughout the entire file. The header line "#Highest height:" gives the value in meters of that maximum height. The last data period in the file is given by the header line "#End time:".

File Naming Convention

The format forgridded file names is "sss_ix_yyyy.txt", where "sss " is replaced by the 3-character station abbreviation, "x" is replaced by the profiler/mode character ('a'=915-MHz high mode, 'b'=915-MHz low mode, 'e'=50-MHz), andyyyy is the year. In some cases, data may be disseminated in files containing less than one year of data. In those cases, file names will be of the form "sss_ix_yyyy_ddd.txt" or "sss_ix_yyyy_ddd_ddd.txt", whereddd represents a 3-digit day-of-year number. The first case is for a single day and the second is for a range of days.

"o" Files and "i" Files

As mentioned above, the gridded files are created from an interpolation of the original TOGA-COARE data onto a uniform vertical grid. At the same time that we created the gridded files, we put the original TOGA-COARE data into files using the same format as the gridded files. These files have names of the form "sss_ox_yyyy.txt", which are like the gridded (or interpolated or "i") file nameswith the "i" replaced by "o", for "original". The internal format of the "o" files is the same as the "i" files, but notethat the heights of these data are the original measurement heights, not the interpolated 100-m grid heights. Also, since the oblique beam and vertical beam measurement heights are not the same, the ht and wht values(columns 10 and 19) are not identical. This is indicated in the header by the line "#Vertical scales:", whose value is 2 in the "o" files and 1 in the gridded "i" files. Therefore, in the "o" files, the header lines "#Lowest height:" and "#Highest height:" will each contain 2 values, the first one for the oblique height scale, "ht", and the second one for the vertical height scale, "wht". An example of "o" file header lines is

#Vertical scales:   2 
#Lowest height:        292 m     312 m  
#Highest height:      2349 m    2516 m

The data in the "o" filesare identical to those in the original TOGA-COARE format files, except that we haveapplied the same quality control adjustments used on the input to the interpolation process.