Applying Geostatistics into an acoustic survey

                                      Satoshi Honda
 
                               Hokkaido Natl.Fish.Res.Inst.
                      116,Katsurakoi,Kushiro,Hokkaido,085 JAPAN
                               E-mail:ponch@hnf.affrc.go.jp

     When  you want  to apply  orthodox statistics  into fish  stock  estimation, you 
should  pay  attention   to  an  important  assumption  that  each   sample  data  is 
"independent" mutually.
     However, is it really suitable  to apply this condition in our common survey? In 
many cases,  our study field, ocean,  has two or three-dimensional  expanse. So it is 
natural to  imagine that  the data  taken from  closer points  may more  similar, and 
samples taken from remote points may less similar than that of closer one.
     A new  statistical technique called  "Geostatistics" is gradually being  introduced 
into fisheries stock assessment. One  remarkable feature of this method is, taking the 
idea of autocorrelation among sample points in proportion to  their distance, instead of 
"independence".
     Under a certain condition of both distribution pattern of  resources and particular 
survey design,  this new  method may  give us  higher confidence  of the  estimated 
average density or total quantity in comparison with the orthodox statistics.
     I will  briefly give  presentation about  a case  applying "Geostatistics"  into an 
acoustic survey  for young  Walleye pollock  stock distributing  offshore southeastern 
Hokkaido, Japan.