| Summary: | 碩士 === 國立臺灣大學 === 農業化學研究所 === 90 === In a contaminated site, the spatial distribution of pollutant is essential for delineating hazardous areas and risk assessment. The spatial distribution of pollutant in contaminated soils is non-homogeneous. The data of pollutant concentrations are usually found to exhibit great variation. Assessing the spatial variability of soil pollutant is thus important for a cleanup regime. In practice, for a cleanup action, it is necessary to determine a unit volume suitable for cleanup. However, an observed value of pollutant concentration is measured as the average over a certain volume, called support, which is usually smaller than the unit for cleanup. If there are not enough observations in each unit, it is difficult to determine whether cleanup is needed or not in each unit. Moreover, changing support will lead to different results when assessing the spatial variability. For instance, the shape of the variogram model will change, too. Thus, to determine the unit size for cleanup should be dependent upon the support size and the number of observed values. In this study, we used the additivity property, called Krige’s relationship, D 2(v|G)= D 2(v|V)+ D2(V|G) (v V G ), to measure the dispersion variance, which can be used to describe the support effects on determining the unit size for cleanup. D2(V|G) is the variance of pollutant concentrations between cleanup units. D 2(v|V) is the variance of pollutant concentrations within each cleanup unit. Based on the Krige’s relation using the variogram models corresponding to various support sizes, D2(V|G) and D 2(v|V) can be calculated. Taking both D 2(v|V) and D2(V|G) into account, a risk function: is proposed to calculate the risk of using the finite observed values to represent the values of pollutant concentrations in each cleanup unit. In the risk function, N is the number of units, n is the numbers of units that has observed values in it, and f(N) is a function of D 2(v|V) vs. N. An optimized unit size with the minimum risk then can be determined by the risk function. The optimized unit size can be used for setting as the cleanup units. Two real data sets of heavy metal concentrations in two contaminated sites located in Taoyuan County, Taiwan were used for illustration.
|
|---|
