桃園市列管農地重金屬再汙染預測模型之研究;Study on Prediction Model for Heavy Metal Recontamination in Regulated Farmland in Taoyuan City

NCUIR > Engineering College > Graduate Institute of construction engineering > Electronic Thesis & Dissertation > Item 987654321/93722

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/93722

Title:	桃園市列管農地重金屬再汙染預測模型之研究;Study on Prediction Model for Heavy Metal Recontamination in Regulated Farmland in Taoyuan City
Authors:	葉明忠;Yeh, Ming-Chung
Contributors:	土木系營建管理碩士班
Keywords:	預測;重金屬;再汙染;農地;汙染;Prediction;Heavy Metal;Recontamination;Farmland;Pollution
Date:	2023-07-12
Issue Date:	2024-09-19 17:32:36 (UTC+8)
Publisher:	國立中央大學
Abstract:	桃園市的土地污染面積在全國排名第二。多年來，雖然政府積極地整治大部分受污染的土地，但土地的特性在不斷變化，仍然容易再次受到污染的威脅。然而，目前尚無專門針對桃園市列管農地重金屬再污染的預測方法。因此，本研究以鎘、銅和鋅這三種重金屬為主，利用Python建立了一個基於土壤監測濃度變化和土壤重金屬濃度增量的隨機森林模型。通過文獻回顧確定模型的相關參數，並收集近18年（2004-2021年）的1555組桃園市列管農地數據，應用了四種常見的演算法進行預測，其中隨機森林表現最佳，準確率為75.76%，誤差率為0.05%，顯示出模型具有一定的可靠性。此外，提出了模型的潛在應用價值，期望未來相關單位能夠利用此模型預測潛在的再污染地點，並由環保單位針對性地部署監測和控制措施，從而避免不必要的廣泛測試，以節省時間、人力和財力等資源。;Taoyuan City has the second largest area of land pollution in the country. Despite the government′s active efforts over the years to remediate most of the contaminated land, the characteristics of the land continue to change, and farmland remains susceptible to potential recontamination. Currently, there is no dedicated method for predicting heavy metal recontamination specifically for Taoyuan City′s regulated farmland. Therefore, this study focuses on cadmium, copper, and zinc, three heavy metals, to establish a random forest model using Python, based on changes in soil monitoring concentrations and soil heavy metal concentration increments. A comprehensive literature review indicates the factors, followed by data collection which involves 1,555 datasets in recent 18 years (2004-2021). There are four common prediction methods applied and random forest performed the best at an accuracy of 75.76% with 0.05% of error, demonstrating a certain level of reliability. Furthermore, potential applications of the model are proposed, with the hope that relevant agencies in the future can use the constructed model to predict potential recontamination sites and enable targeted deployment of monitoring and control measures by environmental protection units, thereby avoiding unnecessary extensive testing and conserving resources such as time, manpower, and finances.
Appears in Collections:	[Graduate Institute of construction engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	24	View/Open

社群 sharing

Loading...