Evaluation of surface water quality in Heilongjiang Province, China: Based on different quantities of water quality indicators

作者：Qi Wei, Qi Wei, Qi Wei, Wei Qi, Siying Li, Junzeng Xu, Zihan Yang, Zhenyang Liu, Peng Chen, Yuzhou Liu, Yimin Ding, Junyi Tan, Jiuying Li · 发表于：Ecological Indicators · 年份：2023 · DOI：10.1016/j.ecolind.2023.110472 · 被引用次数：21 · 研究领域：Water Quality and Pollution Assessment、Water Quality Monitoring and Analysis、Hydrological Forecasting Using AI

Accurately evaluating surface water quality grade (WQG) is of great significance in improving the efficiency of regional water quality management. Based on seven main surface water quality indicators (including dissolved oxygen (DO), chemical oxygen demand (COD), permanganate index (CODMn), ammonia nitrogen (NH4+-N), fluoride (FL), total phosphorus (TP) and total nitrogen (TN)) data from 910 measuring points in Heilongjiang Province from May to August 2017, an improved Nemerow index method (IN method) based on combined weighting was constructed, and the WQG was evaluated. On this basis, the multiple linear or non-linear regression models (MLR or MNR) under different quantities of water quality indicators were established, and the prediction results were evaluated. Results show that: 1) WQG in Heilongjiang Province above grade III accounts for more than 92.6%, evaluated by the IN method; 2) The accuracy of WQG evaluated by MLR based on different quantities of water quality indicators are maintained at 77%-90% and decreases with the increase of the number of water quality indicators (N), which may be related to the selected combination of water quality indicators and key water pollutants. The accuracy is improved by 3.9%-17.4% for MNR as compared to MLR, and especially the accuracy can reach as high as 95.6% when N is 5; 3) Compare to the MLR, the MNR is less affected by the data fluctuation characteristics, and performs best when N is 5 under both the great and small data fluc...