基于随机森林模型的干旱绿洲区张掖盆地地下水水质评价

doi:10.7522/j.issn.1000-694X.2017.00026

摘要/Abstract

摘要： 为合理准确评价地下水水质，建立了基于随机森林（RF）模型的地下水水质评价模型，并根据张掖盆地81个地下水采样点的pH值、Cl^-、SO₄^2-、NO₃^-、Na⁺、NH₄⁺含量及总硬度的监测数据，对研究区的地下水水质进行了综合评价。结果表明：盆地地下水水质主要为Ⅱ、Ⅲ、Ⅳ类水，其中甘州区地下水埋藏较深，水体不容易受到来自地面的污染，水质较好，大多数地方为Ⅱ类水；临泽县和高台县地下水埋藏较浅，水质较差，大多数地方为Ⅲ类水，尤其高台县的水位最浅，再加上地处河段下游，污染更为严重，部分地区达到Ⅳ类。根据指标的重要性度量发现影响研究区域地下水水质的主要因子是NO₃^-含量；其次是NH₄⁺、SO₄^2-、Na⁺、Cl^-含量及总硬度、pH值。为验证模型的有效性，将地下水水质评价结果与基于支持向量机（SVM）和人工神经网络（ANN）的地下水水质综合评价模型模拟结果进行对比，3个模型均能很好地评价研究区地下水水质，但RF模型的评价结果更为准确。

关键词: 水质评价, 地下水, 随机森林, 张掖盆地

Abstract: In this study, a groundwater quality evaluation model was established to assess the groundwater quality reasonably and accurately in the Zhangye Basin by using random forest model (RF). Based on the pH, Cl^-, SO₄^2-, NO₃^-, Na⁺, NH₄⁺, and total hardness observation values of 81 groundwater sampling points in the basin, a comprehensive evaluation of groundwater quality for the whole study area was made. Results indicated that the water quality in the study area can be mainly classified into class Ⅱ, Ⅲ, and Ⅳ. Specifically, water quality in most of the local Ganzhou District was class Ⅱ because groundwater was difficult to be contaminated by surface for the deep water table. However, with shallow water level and poor water quality, most areas in Linze county and Gaotai county had class Ⅲ water quality, especially some areas in Gaotai county even reached class Ⅳ for the highest water level and being located in downstream of the river. Moreover, according to the index of importance, the main factor affecting the groundwater quality in the study area was found to be NO₃^- and the order of the other ions was NH₄⁺, SO₄^2-, Na⁺, Cl^-, total hardness and pH successively. In order to test the validity of the developed model, comparisons were made to the support vector machine (SVM) model and the artificial neural network (ANN) model. Results showed that performances obtained by the three aforementioned models were satisfactory and RF model performed much better than the SVM and ANN models.

Key words: water quality evaluation, groundwater, random forest, Zhangye Basin

中图分类号:

P342

吴敏, 温小虎, 冯起, 尹振良, 杨林山. 基于随机森林模型的干旱绿洲区张掖盆地地下水水质评价[J]. 中国沙漠, 2018, 38(3): 657-663.

Wu Min, Wen Xiaohu, Feng Qi, Yin Zhengliang, Yang Linshan. Assessment of Groundwater Quality Based on Random Forest Model in Arid Oasis Area[J]. Journal of Desert Research, 2018, 38(3): 657-663.

参考文献

[1] 倪深海,白玉慧.BP神经网络模型在地下水水质评价中的应用[J].系统工程理论与实践,2000,20(8):124-127.
[2] Cui Y l,Shao J L.The role of groundwater in arid/semiarid ecosystems,Northwest China[J].Ground Water,2005,42(4):471-477.
[3] 于海姣,温小虎,冯起,等.运用小波变换与支持向量机耦合模型(WA-SVM)预测干旱区地下水埋深[J].中国沙漠,2016,36(5):1435-1442.
[4] 王根绪,杨玲媛,陈玲,等.黑河流域土地利用变化对地下水资源的影响[J].地理学报,2005,60(3):456-466.
[5] 席海洋,冯起,司建华,等.黑河下游绿洲NDVI对地下水位变化的响应研究[J].中国沙漠,2013,33(2):574-582.
[6] 胡晓利,卢玲.黑河中游张掖绿洲地下水时空变异性分析[J].中国沙漠,2009,29(4):777-784.
[7] 甄亿位,郝敏,陆宝宏,等.基于随机森林的中长期降水量预测模型研究[J].水电能源科学,2015,33(6):6-10.
[8] 徐红敏.基于支持向量机理论的水环境质量预测与评价方法研究[D].长春:吉林大学,2007.
[9] Han H G,Chen Q L,Qiao J F.An efficient self-organizing RBF neural network for water quality prediction[J].Neural Networks,2011,24(7):717-725.
[10] Senkal O,Yıldız B Y,Sahin M.Precipitable water modelling using artificial neural network in Cukurova region[J].Environment Monitoring and Assessment,2012,184:141-147.
[11] 岳丹丹,梁海涛,王天慧,等.BP人工神经网络模型在西鞍山铁矿地下水水质评价中的应用[J].环境监测管理与技术,2016,28(4):23-26.
[12] 黄鹤,卢文喜,伊燕平,等.基于支持向量机模型的水质评价研究[J].节水灌溉,2012,2:57-59.
[13] He Z B,Wen X H,Liu H,et al.A comparative study of artificial neural network,adaptive neuro fuzzy inference system and support vector machine for forecasting river flow in the semiarid mountain region[J].Journal of Hydrology,2014,509:379-386.
[14] 康有,陈元芳,顾圣华,等.基于随机森林的区域水资源可持续利用评价[J].水电能源科学,2014,32(3):34-38.
[15] Ding Y R,Cai Y J,Sun P D,et.al.The Use of Combined Neural Networks and Genetic Algorithms for Prediction of River Water Quality[J].Journal of Applied Research and Technology,2014,12:493-499.
[16] 丁世飞,齐丙娟,谭红艳.支持向量机理论与算法研究综述[J].电子科技大学学报,2011,40(1):2-10.
[17] Iverson L R,Prasad A M,Matthews S N,et al.Estimating potential habitat for 134 eastern US tree species under six climate scenarios[J].Forest Ecology and Management,2008,254:390-406.
[18] Wang Z L,Lai C G,Chen X H,et al.Flood hazard risk assessment model based on random forest[J].Journal of Hydrology,2015,527:1130-1141.
[19] 张颖,高倩倩.基于随机森林分类算法的巢湖水质评价[J].环境工程学报,2016,10(2):992-998.
[20] 崔东文,金波.基于随机森林回归算法的水生态文明综合评价[J].水利水电科技进展,2014,34(5):56-60.
[21] 温小虎,仵彦卿,苏建平,等.额济纳盆地地下水盐化特征及机理分析[J].中国沙漠,2006,26(5):836-841.
[22] 肖生春,肖洪浪,宋耀选,等.2000年来黑河中下游水土资源利用与下游环境演变[J].中国沙漠,2004,24(4):405-408.
[23] 温小虎,仵彦卿,常娟,等.黑河流域水化学空间分异特征分析[J].干旱区研究,2004,21(1):1-6.
[24] 简富缋,宋晓谕,虞文宝,等.水资产负债表编制中水资源资产核算账户的建立与分析——以黑河中游张掖市为例[J].中国沙漠,2016,36(3):851-856.
[25] Breiman L.Random forests[J].Machine Learning,2001,45(1):5-32.
[26] 崔东文.随机森林回归模型及其在污水排放量预测中的应用[J].供水技术,2014,8(1):31-36.
[27] 余胜男,陈元芳,顾圣华,等.随机森林在降水量长期预报中的应用[J].南水北调与水利科技,2016,14(1):78-83.
[28] 刘蔚,王涛,高晓清,等.黑河流域水体化学特征及其演变规律[J].中国沙漠,2004,24(6):755-762.