金枪鱼围网沉降性能影响因子的多元回归分析

摘要

We evaluated the relationship between sinking depth and a number of factors using multiple regression to determine the external patterns of sinking a purse seine. We collected data on sinking depth, gear operation, and shooting duration(T), shooting velocity(S), current speed(Cs), current direction(Cd), and purse line(L) length from tuna purse seiners owned by Shanghai Fisheries General Corp. between September and December in 2011. Current direction and shooting velocity had no effect on sinking depth (P>0.05, df=53). Conversely, the shooting duration, current speed, and length of the purse line had a significant effect on sinking depth (P<0.05, df=53). The best fit multiple regression model was: D=0.069T−144.5Cs2+0.022L+158. Principal component analysis suggested that the majority of factors associated with gear operation have little effect on the sinking depth. Under normal oper-ating conditions (shooting duration:550 s ,purse line length:2 000 m), the model predicts that the average sinking depth is 236.78 m with a95%confidence interval ranging from 211.51 to 262.04 m in the absence of wind and cur-rents. Similarly, the average sinking depth will be 92.27 m (5%confidence interval:60.56 to 123.97 m) in 1 kn current.%　　根据2011年9—12月金枪鱼围网渔船调查期间测定的围网沉降深度数据和收集的放网时间、放网速度和括纲松放长度等渔具操作数据,以及流速、流向等渔场环境数据,通过多元回归统计定量分析围网沉降深度与有关影响因子之间的关系,试图寻找影响网具沉降的有效外部因子。结果表明,流向和放网速度对网具沉降深度的影响不明显(P>0.05, df=53),而放网时间、流速及括纲长度对网具沉降深度的影响均显著(P<0.05, df=53),最终得到沉降深度(D)和放网时间(T)、流速(Cs)及扩纲长度(L)的多元回归模型为D=0.069T−144.5Cs2+0.022L+158。主成分分析表明,中国金枪鱼围网作业在一定程度上没有利用有关渔场海况信息对捕捞操作进行有益的改变。在一般操作条件(放网时间为550 s,括纲长度为2000 m)时,模型预测表明：在无风无流的情况下,网具沉降深度平均值为236.78 m,95%置信区间为[211.51,262.04],而遇海流急促的情况(流速为1 kn)时,网具沉降深度平均值为92.27 m,95%置信区间为[60.56,123.97]。