FISHERY BULLETIN: VOL. 72. NO. 2 



A note of caution should accompany the conclu- 

 sions, however. They are valid only for the data 

 collected under the conditions of the experiment, 

 and therefore extrapolation to other areas or to the 

 same area under different experimental condi- 

 tions might not be valid. 



The commercial fishing data demonstrated a 

 Condition 4, i.e., means and standard deviations 

 different with respect to water depth, salinity, 

 Forel-Ule color, and secchi disc transparency 

 (Table 2). Temperature and sea state were not 

 tested, and data were not available for chlorophyll 

 a and currents. The subset of fishing data included 

 measurements from 237 "fish sets" and the total 

 population of oceanographic conditions included 

 measurements from 29 June, 30 June, 6 July, 7 Au- 

 gust, 25 August, and 28 September 1972. For each 

 parameter, a negative correlation is indicated as 

 the mean parameter values for the fishing subsets 

 were significantly less than the mean values for 

 the total parameter populations. The lack of high 

 significance levels for mean salinity and 

 Forel-Ule color value differences was not particu- 

 larly surprising in that the subset approach tends 

 to preclude such significance. In any case, the rela- 

 tionships shown in Table 2 substantiate those 

 shown in Table 1. 



A second approach was used to substantiate still 

 further the correlations formed between fish dis- 

 tribution and salinity, Forel-Ule color, secchi disc 

 transparency, and water depth. Mean parameter 

 values for conditions where menhaden were 

 photographically detected (Table 1) were com- 

 pared with similar values from the fishing subset 

 (Table 2). None of these values were significantly 

 different at levels down to 80% (^-test). 



In summary, water depth, secchi disc visibility 

 depth, surface salinity, and Forel-Ule color were 

 found to correlate negatively with the distribution 

 of menhaden. Chlorophyll a correlated positively 



with fish distribution, although independent data 

 were not available with which to corroborate this 

 relationship as in the case of the other four 

 parameters. 



ERTS- Imagery and Fish 

 Distribution Relationships 



Analysis 



The only complete docket of quality ERTS-1 

 MSS imagery coincidental with main day acquisi- 

 tion events was from 7 August 1972. Band 5 imag- 

 ery from 25 August 1972 was of poor quality and 

 no imagery was available for 28 September 1972. 



The four MSS bands from 7 August 1972 were 

 examined to determine if their density levels re- 

 lated to fish distribution. Bands 6 and 7 did not 

 contain any readily apparent useful density de- 

 tail. Band 4, for reasons which are still unclear, 

 seemed to contain too much density detail. Den- 

 sity levels in Band 5, however, appeared to relate 

 to menhaden distribution. 



Results 



Figure 3a shows a portion of the ERTS- 1 Band 5 

 imagery covering the western portion of the Mis- 

 sissippi Sound and adjacent offshore waters as 

 displayed on a I^S DIGICOL video screen. 

 Superimposed on the image are locations of 23 

 photographically detected menhaden schools. 

 Water imagery densities were divided into two 

 density ranges and color-enhanced (Figure 3b). 

 All menhaden schools were found to lie in the less 

 dense range, enhanced as orange. This density 

 range was further reduced by slicing it to the nar- 

 rowest range possible with the instrument. All of 

 the fish schools can be found to either lie in or 

 immediately adjacent to this range, enhanced as 



Table 2. — Comparison of total parameter populations (with and without fish) and fish 

 parameter population subsets (with fish). 



Total 

 population 



Fishing subset 

 population 



Level of signi- 

 ficant difference 

 (%)' 



'f-tests for differences between means for populations witti unequal variances and F-tests for 

 differences between standard deviations (Ostle, 1963) 



382 



