Output of .was and .wdr Files

Hi there,
Can someone tell me how the amplitude and direction reported in the .was and .wdr are derived? Is this a simple PUV, or some averaged version of the full directional spectra as calulated by MLM-AST?

Cheers

Kurt

Hi there,

Torstein is back on Monday -- he will probably give the better answer  " />'>

- Atle Lohrmann

Hi Kurt,

I am guessing that you are using QuickWave to process you wave data.  There is a help button in the dialog box which provides you with the information that you requested.  I will recap here since it is useful for anyone using Quickwave.

Energy Density Spectra
The Energy Density Spectra is output in the *.WAS file.  The units are meter^2/Hz.  The ASCII file is intended to be easily read into MatLab or Excel, where the first row contains the frequencies of each estimate and each of the following rows is the energy spectra for each wave burst measurement.

The user can select which measurement the enegy density is based; The options for the AWAC are Acoustic Surface Tracking, Pressure, Velocity, or Optimized.  The optimized setting uses AST as the primary unless one of the quality control checks deems the AST too poor to use, in this case the energy estimate is falls back on either the pressure (depth < 20 m) or velocity measurements (depth > 20 m).

For PUV instruments (Aquadopp Current Meter, Aquadopp Profiler, Vector), the pressure is the only measurment used for energy estimates.

Direction Spectra
The Direction Spectra is output in the *.WDR file.  This file contains an estimate of the mean direction at each frequency.  The units for this is degrees, and reference to magnetic North.

The processing method for the direction is either the SUV or MLM-ST (for AWACs with AST), MLM (non-AST AWACs),  or PUV (Aquadopp Current Meter, Aquadopp Profiler, Vector).

kind regards,
Torstein

Hi Torstein,
Thanks for the reply. I did check the help file, and it does not outline the method used to get the directions in the .wdr file. It lists the output as being "directional spectra", but not full directional spectra (energy by degree and frequency bin). So you're saying the mean direction is also an output of the MLM method?

Thanks

Kurt

Hi Kurt,

I think there are two points of confusion here (1) directional spectra (*.WDR) may be confused with full directional spectra (*.WDS), and (2) it looks like the edits to the help file never "took" the last time I compiled QuickWave  

I have pasted in the section covering output files below.  The part on the WDS files discusses what the contents are and how to use it to get the energy density per degree-frequency bins.

on a side note, the directional spectra in the *.WDR file is the mean direction at each frequency; these values are based on the first pair of Fourier coefficients (A1 & B1).  The energy distribution found in the *.WDS file is based on the first two pairs of Fourier coefficients (A1, B1, A2, B2).  The directional processing method to arrive at these Fourier coefficients is selected by the user in the processing parameters option of QuickWave.  The processing settings used are also indicated in the Header file (*.WHR).

Sorry for the confusion with the help file.  Below is what you are looking for.

regards,
Torstein

-------------------- QuickWave File Output ----------------


There are several output files available from QuickWave.  The user has the choice of which files will be output.  There is one file which is always output, this is the header file FILENAME.WPR.  The header file contains information on how the instrument was configured, the parameters used for wave processing, and a description of the remaining output files.

Wave parameter file *.wap[/U]
This file presents the estimates of the standard wave parameters for each wave burst measurement a description can be found in the section on Definition of standard wave parameters.

Wave engery spectra file *.was
This file presents the energy spectra for each burst and begins with the frequency values used (Hz).  The units for the energy spectra are (meter2/Hz).  The spectra used for this is described in the section Definition of standard wave parameters.  Note that the frequency limit (Nyquist) for this spectra can be greater than the directional spectra if the measurements include the Acoustic Surface Tracking (AST); this is because the AST samples at twice the frequency as the standard measures of pressure and velocity.

Wave directional spectra file *.wdr
This file presents the direction at each frequency, or the “directional spectra” for each burst and begins with the frequency values used (Hz).  The units for the energy spectra are (degrees).

Wave full directional spectra file *.wds
The full directional spectra file is a presentation of energy distribution over both direction and frequency.  This is sometimes of interest when viewing the distribution as a contour plot.  This file tends to be quite large and therefore is often not output unless used.  The units are normalized and reported as (Normalized-Energy/degree), this means that if the directional spectrum for single given frequency was integrated from 0 to 360 degrees, then the result would be 1.  

In order to calculate the full energy-direction spectrum, one must multiply the results from the energy spectra with the normalized full directional spectra.  This would provide the units of meter2/Hz /degree.  The full directional spectra is reported as normalized spectra because the file size would be enormous if we tried to cover all sea states with the required significant figures (i.e. dynamic range demands here are considerable).

It is possible that the Energy spectra file has a frequency range that this greater than that found in the full directional spectra.  This has to do with the different Nyquist limits for the directional and non-directional spectra when using AST spectra.  Matching frequency lengths can be achieved by setting the upper frequency limit (Processing parameters) below the Nyquist limit for the directional spectra.

Wave Fourier coefficient file *.wcf
This file presents the first two Fourier coefficient pairs at each frequency.  This data is useful if one intends to perform further post processing of directional estimates.

Wave time series file *.wst
This file presents the AST time series after it has been cleaned up with the despiking routines.  This means the bad detects have been identified and linearly interpolated by the neighboring values that are considered acceptable.  This file tends to be large so it is not necessary to output unless intended to be used.

I did see that section in the help file, but it was not detailed enough (i.e., it doesn't explicitly state that the direction in the .wdr file is a mean direction based on the first two Fourier coefficients). Thank you - that was the information I was looking for. I just want to be able to provide the correct metadata when I write the processed data out to netCDF.

Cheers

Kurt " />'>

One other question - does the fourth beam measure along-beam velocity also? I've read papers that show how to get turbulence estimates from different beam configurations, and I was under the impression that Beam 4 on the AWAC also measured velocity (which I understand is a better measurement of vertical velocity?), but I don't see it in the output...

Kurt

Hi Kurt,

We have actually decided to not measure along beam velocity for the vertical beam.  The reason for this is that the three beams at an angle are sufficient for determining the 3D flow during the current profiling mode and more importantly we do not transmit on this transducer and thus save 30% power during current profiling.

As you already know, the reason we have the vertical pointing beam is for the Acoustic Surface Tracking which would not really work that well for the non-vertical beams.  In wave mode the transducer is dedicated for AST and not the velocity.

When comes to turbulence studies, we have found the velocimeter to be a better instrument because of its (a) higher sampling rate, (b) smaller measurement volume, © measurements are not spatially separated as with a current profiler.

hope this helps,
Torstein

Previously KurtRosenberger wrote:

Hi Kurt,

We have actually decided to not measure along beam velocity for the vertical beam.  The reason for this is that the three beams at an angle are sufficient for determining the 3D flow during the current profiling mode and more importantly we do not transmit on this transducer and thus save 30% power during current profiling.

As you already know, the reason we have the vertical pointing beam is for the Acoustic Surface Tracking which would not really work that well for the non-vertical beams.  In wave mode the transducer is dedicated for AST and not the velocity.

When comes to turbulence studies, we have found the velocimeter to be a better instrument because of its (a) higher sampling rate, (b) smaller measurement volume, © measurements are not spatially separated as with a current profiler.

hope this helps,
Torstein

Dear Torstein:

I have been following your discussion with Kurt regarding spectral estimates with QuickWave software. I have a few related queries.

1. What is the threshold value for AST quality (what value is "good quality data" and why)

2. I need to know more details on the methodology for spectra calculation. For example, If I extract the pressure time series and the velocity time series from the *.wad file, and perform the spectral estimates myself (Welch, hanning,128), I cannot get the high frequency peaks I observe if I use Quickwave, SUV, pressure, 128 to calculate the 1D spectra..Why?...what is the depth of the reported velocity time series of the *.wad file?...

Reading about how ADCPs calculate directional spectra, I learned that RDI's ADCPs measure the wave orbital velocities below the surface fairly accurately.  Time series of velocities are accumulated and from these time series, velocity power spectra are calculated. To get a surface height spectrum the velocity spectrum is translated to surface displacement using linear wave kinematics. The depth of each bin measured and the total water depth are used to calculate this translation. To calculate directional spectra phase information must be preserved. Each bin in each beam is considered to be an independent sensor in an array. The cross-spectrum is then calculated between each sensor and every other sensor in the array. The result is a cross-spectral matrix that contains phase information in the path between each sensor and every other sensor at each frequency band. The cross-spectrum at a particular frequency is linearly related to the directional spectrum at a particular frequency....is this the way AWAC calculates the spectra as well?

Is it possible to access the time series used by the Quick Wave to estimate the spectra (say if I want to apply other analysis such as bispectrum or wavelet) as the spectra of the time series extracted from the *wad file are so different to the spectra calculated with QuickWave.

 

Thanks and sorry for the lenght

Ismael

Dear Ismael,

 

Torstein is out of the office for a couple of weeks. He will be back on the 18th, and I am sure you will have a reply from him shortly after that.

 

Best regards,

Ketil

Hello Ismael,

Answers to your questions are as follows:

 

1.  There is no real threshold value for the AST.  The algorithm we use is to simply find the largest peak in the vertical profile.  Sometimes this can be the maximum of 256 and sometimes it can be as low as 50.   The AST quality is baiscally an indication of the peak strength and this can depend on several things: attenuation of the acoustic signal, amount of instrument tilt (AST vertically oriented), scattering from the surface, etc.  Generally speaking you can assume the AST is working well if the AST quality is above 100.

 

2.  It is hard for me to say why you are not getting a similar peak spectra structure with your pressure spectra to that from the Quickwave software.  I do know that the AST spectra and pressure spectra agree very well so the Quickwave methodology is correct.  Any differences you have with the velocity based spectra probably has to do with how you are using the velocity found in the *.WAD file.  Velocity data here is in beam coordinates.  The position of the measurement cells is found in the Wave Header file (*.WHD). 

The procedure you describe for estimating wave direction and estimating wave energy from the velocity spectra is commonly known as the Maximum Likelihood Method (MLM).  This is method is offered in Nortek's wave processing software as well as the SUV method.  You will note that both MLM and MLMST are available; the latter is simply the MLM with the AST measurment included in the directional processing.

When the MLM procedure is used we keep it in beam coordinates, and therefore this is why it is stored in beam coordinates.  The directional processing with the MLM is relatively complicated and we tend to find that very few perform their own directional processing with MLM.

 

kind regards,

Torstein