SOFIA - Poster - Salvaging Sontek ADVM Data Using a One Beam Solution for a Bad Transducer

Collecting velocity data with Sontek SL (side-looker) acoustic instrumentation requires two beams with comparable signal to noise (SNR) values to properly measure velocity data. On occasion SNR values for each beam may begin to diverge due to a bad transducer, interference from debris in the path of only one beam, or from severe bio-fouling over one beam face (Fig. 1). Data collected during these conditions can exhibit significant bias depending on the magnitude of difference in SNR between the two beams.

graph showing example of separation of signal to noise values for a Sontek side-looker at Taylor River at the Mouth

Figure 1. Example of separation of signal to noise values for a Sontek SL at Taylor River at the Mouth. Values are approximately 15 dB in difference between beam 1 and beam 2. [larger image]

Data collected during conditions such as these can be salvaged with some simple techniques and trigonometry. The standard method for beam transformation from any frequency Sontek SL is the xyz coordinate system. This system assumes transducer calibration and similar signal strength is put out by both beams. When one beam has been affected by the issues listed above then the calculations used for this coordinate system can bias velocity X data. This bias is evident when data is examined in beam coordinate transformation (Fig. 2). Data displayed in beam coordinates should be a mirror image of the other beam when transducers are functioning properly. When there is a biased beam the beams will begin to show differences when plotted in beam coordinates. The bias will make velocities appear lower then they may actually be (Fig. 3) resulting in lower computed discharge.

graph showing velocity values output in beam coordinates at Trout Creek

Figure 2. Velocity values output in beam coordinates at Trout Creek. Each beam should be the mirror opposite of the other when transducers are functioning properly. This plot shows beam 2 is not responding the same as beam 1. [larger image]

graph showing velocity X before and after transformation

Figure 3. Velocity X (3 months) from XYZ coordinates before using the beam coordinates translation equation (yellow line) plotted with velocity X after using the translation equation (black line). [larger image]

Despite a bad transducer, data can be salvaged by exporting the raw data from View Argonaut © in beam coordinates. The data from the good beam must then be run through the following formula so that you can derive velocity X values from the good beam.

It is important when examining the velocity data in beam coordinates that the sign of beam you will be transforming matches the original raw data file velocity direction. The bank the instrument is deployed on will determine which beam may need a sign change to translate into the proper flow direction. The raw data file while biased in xyz coordinates, will give the correct flow direction provided one of the two transducers is still working properly.

Using biased velocity data will also lead to biased discharge data when the velocity rating is applied (Fig. 4). The uncorrected velocity data, when used to compute discharge with the existing rating, had a percent error of approximately 30%. Once the data was transformed using the equation the percent error reduced to approximately 9%. Care must be taken to not assume that the beam with the lower SNR values is the bad one. Examination of beam velocities in the beam coordinates plot will be necessary to verify which transducer is actually experiencing problems. An example of a bad transducer with a higher SNR value then the good transducer is shown in figure 5.

If an instrument is identified as having problems with one of the transducers then it is critical that ADCP measurements are performed prior to instrument removal. Data collected from the ADCP measurement will be critical for determining if this method to correct your data will be valid when related to your index-velocity rating.

graph comparing biased data plotted on a rating before re-computation using beam coordinates

graph comparing biased data plotted on a rating after re-computation using beam coordinates

Figure 4. Comparison of biased data plotted on a rating before and after re-computation using beam coordinates. Plot A is the rating with the data in question reported in XYZ coordinates. Plot B. is the rating with data re-computed using only the good beam. The circled area shows the data affected by a bad transducer. [click on images above for larger version]

graph showing signal to noise separation where beam 2 was bad

Figure 5. Signal to noise separation where beam 2 was bad. Lower SNR for one particular beam is not enough evidence to determine if a transducer is bad. Only careful examination of velocity data in beam coordinates coupled with examination of SNR will help determine which transducer is bad. Collection of diagnostic parameters is imperative for any corrections that may be needed in post processing of data. [larger image]

Special thanks to Christian Lopez, Shane Ploos, Carrie Boudreau, and Stephen Huddleston for their long hours in the field collecting and analyzing data.

Poster displayed at 2008 National Data Conference, sponsored by CHIDER, Tunica, MS June 16-20, 2008