摘要
根据丰满库区及下游河段1985-2010年的水质监测资料,选取其中监测数据全面、完整的3个水质因子(NH3-N、DO、CODMn),分别对其总体统计特征进行分析,得出这3种因子的沿程变化规律,然后采用均生函数拓展进行时间序列分析,分别作线性回归和多项式回归,比较选择较优方法进行水质因子预测。分析结果表明:水质因子CODMn、NH3-N在3个水期(丰水期、平水期、封冻期)的沿程变化规律基本一致,即先降低后增大,而后又降低,而DO在3个水期的沿程变化不大;另一方面,对水质资料进行均生函数的时序分析是合理的,采用多项式回归对水质因子进行预测是可行的,且比多元线性回归精度要高。
Based on the water quality monitoring data of the Fengman Reservoir and downstream from 1985 to 2010,three comprehensive water quality parameters（NH3-N,DO,CODMn） are selected.Then their overall statistical features are analyzed and their variation law is obtained.In this paper the mean generation function is used to finish the time series analysis with linear regression and polynomial regression.Then the optimum method is chosen to carry out the prediction of water quality factors.The results show that the variation of water quality factor（CODMn,NH3-N）is basically consistent during the three water periods（high-flow period,normal water period,freezing period）.That decreases firstly and then increases,then decreases again,while the DO changes little along the way.In addition,it is reasonable to use the timing mean function to analyze water quality data;it is feasible to use polynomial regression to predict water quality factors,and its accuracy is higher than the multiple linear regression.
出处
《中国农村水利水电》
北大核心
2011年第3期58-61,65,共5页
China Rural Water and Hydropower
关键词
统计特征
时间序列分析
回归
水质
丰满大坝
statistical features
time series analysis
regression
water quality
Fengman Dam
