Tolerance and calibration adjustments

When should gains be adjusted based on calibration results?  Several users have chosen not to adjust gains when differences are small arguing that the errors inherent in the calibration itself is larger than the suggested change.  Foote (1983) and Simmonds and MacLennan (2005) suggest that calibrations should be able to give gain values for on-axis TS measurements of ± 0.2 dB (at 38 kHz).  If this holds for the higher frequency transducers, calibration results within 0.2 dB may be consider an indication that the unit is working properly and that no changes are required.

Calibrations of the Cornell University units have generally given larger differences than 0.2 dB and Rudstam generally adjusts gains according to the calibration done in conjunction to the surveys even if the differences are small.  Calibrations of Cornell University’s 70 kHz split beam Simrad transducer (EY-500) with 0.2 ms pulse duration has resulted in measured TS of the ball ranging 3.4 dB (± 1.7 dB, since 1996).  Twenty-seven of 29 calibrations considered acceptable were within ± 0.8 dB of the overall mean.  This is larger than the ± 0.2 dB calibration error and suggests that each survey should be related to its own calibration.  Calibrated gains also vary with pulse length; TS of the ball with default setting was 2.2 dB lower with 0.6 ms pulse length than with 0.2 ms pulse length.  For the Cornell split beam 120 kHz Biosonics Dt-X unit, calibrations in 2005 and 2006 ranged over ± 0.5 dB and also varied with pulse length, higher TS were measured at 0.2 ms than at 0.4 ms for the same ball (-39.01 dB at 0.2 ms and 40.22 dB at 0.4 ms for a -40.4 dB calibration sphere). However, this unit did give the same values for the ball for 0.4 to 0.8 ms pulse lengths.  Somewhat higher variation (but similar to the Cornell Simrad unit) has been observed for Biosonics equipment used by USGS Great Lakes Science Center and somewhat lower variation for Biosonics equipment used by Vermont Fish and Wildlife during 2007 (5 calibrations within ± 0.3 dB).  Since the calibration constants stored in the Biosonic transducer is for the default pulse duration of 0.4 ms, it may be important to use gain offsets when analyzing data collected with other pulse durations than 0.4 ms, especially when using shorter pulse lengths. Therefore, it is imperative to calibrate at all pulse length used in the surveys.

We recommend using the same methods and software for calibration as for data analysis as there are some differences between analysis programs.  This will result in a calibration that accounts for such differences.