Predicting heavy metal recontamination in farmland

Taoyuan City ranks second in the nation in terms of land pollution area. Over the years, while the government has actively remediated a significant portion of contaminated land, the evolving nature of land characteristics continues to pose a threat of recontamination. Therefore, this study primarily focuses on cadmium (Cd), copper (Cu), and zinc (Zn) heavy metals and utilizes Python to construct a Random Forest model based on soil monitoring concentration changes and incremental heavy metal concentrations. Through a literature review, the relevant parameters of the model were determined. A dataset of 1,555 records from Taoyuan City's managed agricultural land over the past 18 years (2004-2021) was collected. Four commonly used algorithms were applied for prediction, with the Random Forest demonstrating the best performance, achieving an accuracy of 75.76% and an error rate of 0.05%. This indicates a certain level of reliability in the model. Furthermore, potential applications of the model are discussed, with the hope that relevant authorities can use this model to predict potential recontamination sites in the future. Environmental agencies can then strategically deploy monitoring and control measures, thereby avoiding unnecessary widespread testing and saving resources such as time, manpower, and finances.