| In the process of studying soil heavy metal pollution, data is the foundation. High quality data could accurately characterize the pollution status of soil heavy metals in the region. However, data which is mixed with outliers or false data may lead to an inaccurate evaluation results and even cause erroneous judgment and it would bring much inconvenience to the follow-up work such as the soil remediation and management. So it is of great importance to control the data quality and disposal of outliers of soil heavy metal data.Mathematical statistics method, background value method and local spatial autocorrelation method were used to detect the outliers of As, Cd and Pb of Beijing in this paper. Results showed that there were 8, 34, and 38 outliers for the As, Cd, and Pb concentrations in the Beijing soil,respectively. The detected outliers in the Beijing soil were re-analyzed.The analysis revealed that 75.0%, 76.5 %, and 92.1% of the As, Cd, and Pb outliers, respectively, were caused by systematic and artificial errors.After the correction, the interpolation accuracy improved significantly.The mean relative error (MRE) of As, Cd, and Pb outliers decreased by 48.0%, 44.6%, and 54.7%, the mean absolute error (MAE) decreased by 51.8%, 46.7%, and 54.4%, while the root mean square error (RMSE) of these outliers decreased by 34.2%, 33.3%, and 46.4%, respectively. The MRE values of the nearest neighboring points, which were influenced by the outliers decreased by 5.2%, 20.6%, and 27.6%, MAE values decreased by 5.4%, 25.5%, and 29.8%, while the RMSE values of these points decreased by 5.3%, 17.3%, and 33.2%, respectively. Results indicate that the proposed combined method can effectively detect the outliers of soil heavy metal. The quality of the survey data improved under the premise of adding a finite sample size and analysis time. In addition, an effective tool was provided to investigate on regional soil and guarantee high data quality. |
PDF Full Text Request