研究论文
高学民,陈静生,王立新.Logistic Regression在我国河流水系氮污染研究中的应用[J].环境科学学报,2000,20(6):676-681
Logistic Regression在我国河流水系氮污染研究中的应用
- Application of logistic regression to the study of nitrogen in surface water
- 基金项目：国家自然科学基金项目(编号:49671017)
- 作者
- 单位
- 高学民
- 北京大学城市与环境学系, 北京 100871
- 陈静生
- 北京大学城市与环境学系, 北京 100871
- 王立新
- 北京大学城市与环境学系, 北京 100871
- 摘要：对四川省岷江、沱江及嘉陵江流域和江西省的赣江流域及鄱阳湖地区共1.70多个水文站的数据进行了相关分析和多元回归分析.结果表明, 河流水中硝态氮浓度与年降雨量、人口密度、氮肥施用量、牲畜饲养量、农作物及粮食作物种植面积等因素有较好的相关性.以以上数据资料为基础, 将河流水NO3--N的浓度划分为背景浓度(<0.7mg/L)、受人类活动的显著影响的NO3--N浓度(>3.0mg/L)以及中间类(0.7—3.0mg/L)进行LogisticRegression分析, 两个Logistic模型的准确度分别达82.46%和89.19%.运用Logistic模型对整个长江流域河流水中NO3-N浓度进行估计, 结果与实测值基本相符合.
- Abstract：The correlation analysis and multi-regression analysis were applied to data of more than 170 gauging stations in R.Min, R.Tuo, R.Jialing, R.Gan watersheds and Lake Poyang (located in Sichuan and Jiangxi province).The results indicate that there exist a close relationship between the nitrate nitrogen concentration in surface water and social-economic development indexes and climatic conditions such as precipitation, population density, application of nitrogen-fertilizer, number of live-stock, area of farm crop etc..Based on the data of Sichuan and Jiangxi province, the nitrate nitrogen concentration in surface water were divided into three categories:backgroud concentration (<0.7mg/L), NO-3--N concentration that affected by human activity significantly (>3.0mg/L), middle category (0.7—3.0mg/L), then analyzed with Logistic Regression.The result indicates that the accuracy of two Logistic models are up to 82.46% and 89.19% respectively.The predicted results of the nitrate nitrogen concentration in surface water of the Yangtze River wastershed with the models agree with measured values.In addition, the Logistic models can be used to predict that in what situations NO-3--N concentration in surface water polluted heavily.
摘要点击次数： 1735 全文下载次数： 2997
