基于北京市城市河道水体发臭大量感官数据及水质监测数据,采用多元线性回归及有序Logistic回归建立北京市水体发臭回归模型.研究发现多元线性回归法与有序Logistic回归法整体准确率相似(78%-80%),但发臭评价准确率明显高于后者,更适合作为水体发臭回归方法进行评价.将该方法应用于北京市地表水发臭分级评价中,筛选确定了适合做水体发臭回归自变量的监测指标,发现溶解氧及氨氮起主要作用,对水体发臭贡献率约95%;有机类指标对解释水体发臭效果显著程度从大到小依次为:高锰酸盐指数>五日生化需氧量>化学需氧量,由于五日生化需氧量分析时间长、高锰酸盐指数多用于Ⅰ—Ⅲ类水体监测、化学需氧量是水环境污染物总量控制的关键指标,综合考虑建议采用化学需氧量为回归自变量;总磷不适合作为北京市城市河道水体发臭回归自变量.对2012年北京市部分城市河道水体发臭情况进行分析,结果发现本标准用于评价发臭程度与国家颁布的黑臭标准的评价结果吻合程度为95.6%,通过选取典型发臭水体进行分级评价,再次验证了该评价方法可行.
Based on two regression methods (multiple linear regression,ordinal logistic regression),the correlation was determined between Human's sensory data and the water quality monitoring data to grade and assess urban creek's malodor in Beijing.After comparing and analyzing,it was found that the two methods shared similar total accuracy rate of prediction (78%-80%),while the multiple linear regression's accurate rate of malodor predication was significantly higher.Therefore the multiple linear regression method was selected.Through screening,some monitoring indicators were considered as appropriate independent variables to explain the urban creek's malodor in Beijing.Among them,dissolved oxygen (DO) and ammonia nitrogen (NH3-N) were determined as the main indicators causing malodor,sharing the contribution rate of about 95%.As for organic indicators,Permanganate index (CODMn) was the optimized indicator following by biochemical oxygen demand (BODs) and chemical oxygen demand (CODcr).However,five days were needed for monitoring BOD5,and CODMn was usually used to monitor water bodies of Grade Ⅰ —Ⅲ.Thus,CODcr was selected as the appropriate organic indicator which was also the key indicator of overall control of water pollutants' quantity in China.It was also found that total phosphorus (TP) might not be an appropriate independent variable.During application,it was shown that the evaluation results of Beijing's urban creek malodor in 2012 with the method proposed by this research had the concordance rate of 95.6% with the results in the national standard.Eventually,the method proposed by this research was also verified in the case study.
参考文献
- 下载量()
- 访问量()
- 您的评分:
-
10%
-
20%
-
30%
-
40%
-
50%