中文摘要：

以海州湾2004-2006年赤潮实测资料和同期的水文气象资料为研究对象，通过构建多元logistic回归模型，从各种环境要素中提取出对赤潮暴发有显著影响的无机氮（DIN）、pH值、盐度、水温、硅酸盐和风速等环境要素。根据样本分布状况确定每个环境要素阈值的遍历范围和遍历步长，采用含哑变量的logistic回归模型对有显著影响的环境要素值进行遍历，根据回归的结果建立风险度和相对风险度等指标，并依据这些指标筛查出与赤潮暴发显著相关的各环境要素的阈值分别为：DIN：0．09mg／L；pH值：8．09；盐度：28．4；水温：25．6℃；硅酸盐：0．24mg／Lg风速：3．2m／s，为在该海域控制赤潮的危害提供一种简便有效的手段。

英文摘要：

The present study analyzes the data of red tides, hydrology and meteorology collected during May to October of each year from 2004 to 2006 by means of logistic regression, and distills 6 environmental factors including DIN, pH, salinity, temperature, SiO3^2- and wind speed. The ergodic range and step length are confirmed with the distribution of the samples. The ergodic processing adopts the model of logistic regression with dummy variables in ergodic range after establishing the guide line of dangerous degree and relative dangerous degree. The thresholds of environmental factors which affect the course of red tides remarkably are confirmed by the variety of dangerous degree and relative dangerous degree. The thresholds of environmental factors are： DIN： 0.09mg/L; pH： 8.09; salinity： 28.4‰; temperature： 25.6℃; SiO3^2-： 0.24 mg/L; wind speed： 3.2m/s. The coastal productions can adopt preventive steps when the values of environmental factors exceed the threshold to control the damage of red tides easy and efficiently.