重回帰式による赤潮発生予測〔広島湾奥部の赤潮発生予察に関する研究-2-〕

概要

論文の詳細を見る
This study was conducted to ascertain the possibility of predicting red tide by the multiple linear regression model based on the surface water quality data of the northern part of Hiroshima Bay. The monthly regression equations from May to September were calculated using the data from 1976 through 1980. The results obtained are as follows. 1) Partial regression coefficients found to be significant at the 1% level were salinity, TDP and DIN in June, July and August, and those at the 5% level were WT in June and DON in Septmber. 2) In the monthly regression equations by multiple regression analysis (MRA), the coefficients of determination were found to exceed the 50% level in July and August and their equations are PON=262+7.79WT-12.9Sal.+5.72TDP-0.33DIN+0.05DON in July and PON=681+7.88WT-27.42Sal.+4.22TDP-0.62DIN-0.05DON in August. 3) In the best regression equation by the stepwise method of MRA, multiple correlation coefficients ranged from 0.31 to 0.81 and the coefficient of determination exceeded 50% only in August. The regression equation in August is PON=631+8.44WT-26.66Sal.+4.26TDP-0.60IN. 4) In all cases, salinity was highly correlated with PON. Decrease of salinity in Hiroshima Bay was caused by the discharge of ricer water, factory wastes and urban sewage which contains iron, manganese, organic matters and vitamins. These substances are considered to play an important role in the occurrence of red tides.
日本水産學會の論文