KEMMERER ET AL : ERTS-I MENHADEN EXPERIMENT 



orange (Figure 3c). The 10 tightly grouped school 

 location indicators in the middle-left portion of 

 the image overlie a small orange enhanced area 

 making the latter difficult to see. 



Unfortunately, the lack of additional data to 

 test the persistence of the relationship between 

 menhaden distribution and MSS Band 5 imagery 

 density levels precludes any but the most tenta- 

 tive of conclusions. However, the data are suffi- 

 cient to warrent an observation that the imagery 

 does appear to contain information relating to the 

 distribution of menhaden schools. 



ERTS-1 Imagery and Oceanographic 

 Parameter Relationships 



Analysis 



An analysis was performed on the MSS Band 5 

 imagery for 7 August 1972 to determine if image 

 densities could be explained based on oceano- 

 graphic parameter measurements. An isodensity 

 tracing was made of that portion of the imagery 

 covering the study area to provide quantitative 

 relative density data. The tracing was not particu- 

 larly satisfactory because of instrument limita- 

 tions which caused more than one density range 

 to be represented by the same color trace, but 

 accurate enough to demonstrate relationships. 



Results 



Water depth, secchi depth visibility, and the 

 interaction between the two parameters (formed 

 by their product, Mott, 1967) were regressed 

 against relative image densities. Simple correla- 

 tions (r) between these parameters and image 

 density were 0.56, 0.73, and 0.69, respectively, 

 significant at the 99% confidence level. A slight 

 improvement in precision (r = 0.77) was realized 

 when the parameters were combined through 

 multiple regression (Table 3) into the following 

 equation: 



Image Density = 0.5776 + 0.0222B + 0.0762T 



-0.00515T (4) 



where: 

 B = water depth in meters, 

 T= secchi disc transparency in meters, 



BT = interaction formed as the product of B and T 



Of the parameters, secchi disc transparency was 



Table 3. — Analysis of variance for the relationship between 

 ERTS-1 image density and two oceanographic parameters. 



the most important one in the equation as indi- 

 cated by the relative magnitude of the coefficients 

 and the simple correlation coefficients. The most 

 meaningful facet of this analysis is that the two 

 parameters correlating significantly with image 

 density levels also correlated significantly with 

 menhaden distribution (Tables 2 and 3). Thus, it 

 appears that the apparent correlation between 

 menhaden distribution and Band 5 density levels 

 (Figure 3) is more than a chance occurrence and 

 can be explained based upon secchi disc transpar- 

 ency and water depth measurements. 



PREDICTION MODELS FOR 



RESOURCE MANAGEMENT 



AND UTILIZATION 



A potential management and utilization benefit 

 from this experiment is identification of an ap- 

 proach through which remotely sensed environ- 

 mental data could be used to provide distribution 

 information about menhaden stocks in the study 

 area. This information could be used to reduce 

 search time for commercial concentrations of 

 menhaden by fishermen and as a means to develop 

 efficient survey designs by resource managers. 

 Ideally, distribution information should be valid 

 for the entire Gulf Coast menhaden fishery; how- 

 ever, this ideal case cannot be supported with re- 

 sults from this experiment but can be realized only 

 through future experiments specifically designed 

 to test demonstrated relationships in other areas. 



Model Development 



Demonstrated menhaden distribution- 

 oceanographic parameter relationships (Table 1) 

 were placed into a context potentially useful to 

 commercial fishermen and resource managers. 

 Multiple regression analysis was used to develop 

 eight empirical models to predict menhaden 

 distribution (D) in the study area based on 



385 



