以初次採樣密度與管制值為基礎的調整式叢集採樣改善克金利推估應用於污染場址的界定

Using adaptive cluster sampling based on first sampling density and regulation thresholds to improve kriging estimation on the delineation of contaminated sites

廖婉君(Wan-Jiun Liao)

指導教授 ： 李達源

國立臺灣大學/生物資源暨農學院/農業化學研究所/碩士(2005年)

https://doi.org/10.6342/NTU.2005.01663

摘要

污染物在土壤中的空間分佈是污染區風險評估及整治復育的重要依據。近來，地理統計的克利金法常被應用於推估土壤污染物含量的空間分佈，可有助於界定土壤污染範圍並做為危害風險評估與整治復育的參考。然而當污染物濃度接近危害管制值，且其所在位置周圍附近採樣密度又偏低的情況下，可能會使克利金法推估的空間分佈圖應用在界定污染危害範圍時，產生較大的誤判率，而導致整治資源的浪費或造成環境的潛在威脅。因此本研究同時考慮初次採樣密度與污染濃度管制值，利用調整式叢集採樣方法，以期能降低克利金推估應用在界定危害範圍時的錯判率。調整式叢集採樣方法是以初次採樣的樣本資料為基礎，依據調查目的設定再次採樣的需求指標(requirement index)，符合需求指標的位置會優先選取進行再次採樣。本研究利用交叉驗證法，計算初次採樣位置的克利金推估值和標準差，以推估值和管制值的差(deviation)與克利金標準差(kriging standard deviation)的比值做為再次採樣的需求指標，該指標的絕對值愈小即表示該採樣點附近的濃度愈接近污染管制值且初次採樣密度較低，應優先選出並在其附近的適當位置進行再次採樣。為確認本方法的適用性，故在本研究中模擬調整式叢集採樣和簡單隨機採樣取得土壤樣本，以比較不同
採樣方法對應用克利金推估界定危害範圍的影響。本研究以臺灣彰化縣境內的重金屬污染農地為例，選用土壤中鎳濃度資料進行本研究的測試模擬與比較說明。結果顯示，考慮初次採樣密度與管制值的調整式叢集採樣相較於簡單隨機採樣，能將再次採樣的樣本位置優先配置於鎳濃度接近管制值的區域和初次採樣密度較低的位置，因此有助於提升克利金推估的準確度。而且由模擬結果發現，相較於簡單隨機採樣，使用調整式叢集採樣能減少以克利金推估值界定污染危害範圍的錯判率。如此可更精準的將克利金推估之污染分佈圖，應用於危害範圍的界定，以提供土壤整治復育或進一步細密調查之用。

關鍵字

採樣設計 ； 調整式叢集採樣 ； 簡單隨機採樣 ； 克利金 ； 錯判率 ； 風險評估 ； 土壤復育

並列摘要

Spatial distributions of a pollutant in contaminated soil are essential for risk assessment and soil
remediation. Recently, the kriging technique is frequently used in spatial interpolation of pollutant concentrations to delineate contamination for hazardous risk assessment and soil remediation. However, high misclassification rates of hazardous area delineation with kriging estimated values of pollutant concentration will happen at the locations, where pollutant concentrations are close to the regulation thresholds for declaring hazardous areas and first sampling density is low. This situation consequently results in high cost of remediation and disadvantage impacts on environment health. Thus, in order to reduce the misclassification rate of hazardous area delineations with kriging pollutant concentrations, an adaptive sampling approach, called adaptive cluster sampling, based on the first sampling density and regulation threshold of pollutant concentrations was proposed to improve kriging estimation in this study. The adaptive sampling given a selected requirement index, which determining the priority of candidate locations for further sampling, was used to draw the additional samples at the locations where pollutant concentrations are close to the regulation thresholds and first sampling density is low. For assessing the feasibility of the adaptive sampling approach, a comparison of adaptive cluster sampling (ACS) and simple random sampling (SRS) was carried out in simulation. A data set of soil Ni concentrations in a heavy-metal contaminated site in Changhua County, Taiwan, was used for illustration. The results showed that the additional samples with Ni concentrations close to the regulation threshold (Ni = 200 mg kg-1) were drawn in ACS but not in SRS. Simultaneously, the sampling configuration of additional samples in ACS was on the locations where first sampling density is low but was that in SRS randomly spreading on the site. The precision of kriging estimation of Ni concentrations based on the observations sampled in ACS was thus relatively higher than that based on the observations sampled in SRS. Compared with using SRS, using ACS could reduce the misclassification rate of hazardous area delineations with kriging Ni concentrations in the results of simulation.

並列關鍵字

sampling design ； adaptive cluster sampling ； simple random sampling ； kriging ； misclassification rate ； risk assessment ； soil remediation

參考文獻

Alloway, B. J. 1995. Heavy Metals in Soils. London Glasgow, Blackie Academic and Professional.

Armstrong, M., and R. Jabin. 1981. Variogram models must be positive-define. Math. Geol. 13 : 455-459.

Burgess, T., R. Webster, and A. MacBratney. 1981. Optimal interpolation and isarithmic mapping of soil properties. IV. Sampling strategy. J. Soil Sci. 32:643-659.

Candela, L., R. A. Olea, and E. Custodio. 1988. Lognormal kriging for the assessment of reliability in groundwater quality control observation networks. J. Hydrol. 103:67-84.

Christakos, G. and R. A. Olea. 1992. Sampling design for spatially distributed hydrogeologic and environmental processes. Adv. Water Resour. 15:219-237.