FISHERY BULLETIN: VOL. 71, NO. 3 



DISCUSSION 



A major advantage of the digital processing 

 system is the ability to estimate fish densities 

 in a number of depth strata simultaneously. 

 This ability allows detailed examination of the 

 variation of density with depth. During the 1971 

 surveys in Port Susan the Pacific hake were 

 uniformly distributed over both depths and 

 transects, so that it was possible to consider 

 all observations as estimates of the mean fish 

 density. A slightly different distribution was 

 found during the 1970 acoustic surveys, when 

 two density strata were identified (Thorne et 

 al., 1971). During the acoustic surveys in Port 

 Susan in 1969 and 1970, integration was done 

 directly off the sounders and a single output 

 was obtained, including the entire depth range 

 of interest. Thus only one estimate of density 

 was obtained for each transect, and only 9 ob- 

 servations were made over a standard set of 

 nine transects; whereas 26 observations were 

 obtained from the 1971 survey processed with 

 the analog integrator and 125-145 observations 

 were taken with the DDAPS from 10-m depth 

 strata and three outputs per transect. The in- 

 creased number of outputs greatly decreased 

 the variance of the estimated mean density. For 

 example, Thorne et al. (1971) calculated that 

 36 transects (18 per stratum) were required 

 to reach a precision of ± 15% in 1970, neglecting 

 variance in the calibration relationship. This 

 level was reached in nine transects processed 

 in three depth intervals with the analog inte- 

 grator from magnetic tape in 1971, whereas 

 precisions of about ±8% were shown for nine 

 transects processed by the DDAPS, and a pre- 

 cision less than 5% for the combined four series 

 (549 observations). 



The estimates of total population from the 

 four series processed with the DDAPS varied 

 even less than the estimates of mean density 

 because of an apparent inverse relationship 

 between density and average bottom depth. It 

 is possible that shallower areas had slightly 

 higher fish densities. It is interesting to note 

 that the range in estimated fish density for the 

 four series was not much greater than the range 

 in estimated average depth of Port Susan (mean 

 98 m, range 95-105 m). 



A good linear relationship was obtained be- 

 tween catch and integrated voltage for both 

 the digital and analog systems in 1971. The 

 relative error of the beta coefficient for the re- 

 lationship between integration and catch ob- 

 tained in 1970 was quite large, 29% compared 

 to less than 10% for both systems in 1971. Part 

 of the variability in the 1970 results probably 

 was associated with instability in the analog 

 voltage-squared integrator. 



The highly linear relationship between catch 

 and integration from the 1971 analog voltage 

 integrator was somewhat surprising. Since 

 the integrated voltages were unsquared, the 

 relationship should have deviated from linearity 

 at higher densities (Thorne, 1971). However, 

 densities were relatively low compared to the 

 resolution of the system. The pulse resolution 

 volume at the mean haul depth (80 m) was 

 about 40 m^, if an 8° cone and 0.6 msec pulse 

 length are assumed. Thus a catch of about 525 

 lb (750 fish) would be equivalent to an average 

 density of one fish per pulse resolution volume. 

 Only four catches were greater than 525 lb. 

 There is some indication that the highest two 

 observations may be deviating from linearity. 

 The ratio of fish density to integrated voltage 

 associated with multiple targets would be 

 greater and would tend to increase the magni- 

 tude of the beta coefficient over that expected 

 from all single targets. This effect may account 

 for the greater population estimate derived 

 from the survey processed with the analog echo 

 integrator. 



Estimates of density of fish from an echo in- 

 tegrator are dependent upon determination of 

 the individual fish's mean target strength 

 (Moose, Ehrenberg, and Green, 1971). In acous- 

 tic studies of Pacific hake populations in Port 

 Susan since 1969, the mean target strength 

 has been determined indirectly from comparison 

 of integrated voltage with net catches, on the 

 assumption of 100% net efficiency. The proce- 

 dure provides a valuable relative index for an- 

 nual comparisons, but suffers from a large 

 variability associated with the net catch and 

 from possible bias in the assumptions concern- 

 ing net efficiency. A relative error of 9.4% was 

 associated with the calibration relationship 

 between the catches from the 21 net hauls and 



842 



