Abstract
Five years have passed since the first set of environmental samples was taken in 2011 to represent various ecosystems which would help future generations lead back to the past environment. Those samples have been preserved cryogenically in the National Environmental Specimen Bank(NESB) at the National Institute of Environmental Research. Even though there is a strict regulation (SOP, standard operating procedure) that rules over the whole sampling procedure to ensure each sample to represent the sampling area, it has not been put to the test for the validation. The question needs to be answered to clear any doubts on the representativeness and the quality of the samples. In order to address the question and ensure the sampling practice set in the SOP, many steps to the measurement of the sample, that is, from sampling in the field and the chemical analysis in the lab are broken down to evaluate the uncertainty at each level. Of the 8 species currently taken for the cryogenic preservation in the NESB, pine tree samples from two different sites were selected for this study. Duplicate samples were taken from each site according to the sampling protocol followed by the duplicate analyses which were carried out for each discrete sample. The uncertainties were evaluated by Robust ANOVA; two levels of uncertainty, one is the uncertainty from the sampling practice, and the other from the analytical process, were then compiled to give the measurement uncertainty on a measured concentration of the measurand. As a result, it was confirmed that it is the sampling practice not the analytical process that accounts for the most of the measurement uncertainty. Based on the top-down approach for the measurement uncertainty, the efficient way to ensure the representativeness of the sample was to increase the quantity of each discrete sample for the making of a composite sample, than to increase the number of the discrete samples across the site. Furthermore, the cost-effective approach to enhance the confidence level on the measurement can be expected from the efforts to lower the sampling uncertainty, not the analytical uncertainty. To test the representativeness of a composite sample of a sampling area, the variance within the site should be less than the difference from duplicate sampling. For that, a criterion, ${i.e.s^2}_{geochem}$(across the site variance) <${s^2}_{samp}$(variance at the sampling location) was proposed. In light of the criterion, the two representative samples for the two study areas passed the requirement. In contrast, whenever the variance of among the sampling locations (i.e. across the site) is larger than the sampling variance, more sampling increments need to be added within the sampling area until the requirement for the representativeness is achieved.
국가환경시료은행에서는 과거 환경 재현을 목적으로 다양한 생태계를 대표하는 시료를 채취 저장하고 있다. 지난 5년간 8종의 생태계 시료종이 엄격한 표준운영절차(SOP)에 따라 채취되어 왔으나 수행절차에 대한 비용효율성이나 시료의 대표성에 대한 논리적 통계적 검증은 이루어 진 바 없었다. 따라서 본 연구에서는 시료채취 및 분석과정으로 구성된 표준운영절차의 각 단계에서 비롯되는 불확도(uncertainty) 수준에 대한 평가를 실시하였다. 이를 위해 표준운영절차에서 규정된 채취방법에 의해 채취된 두 지역의 침엽수 시료를 대상으로 중복시료(duplicate sample)를 채취하였고, 이에 대한 중복분석결과를 대칭설계(balanced design)하여 분산분석을 실시하였다. 시료채취 및 분석의 각 단계에서 산출된 불확도 수준은 각 해당지역 대표시료에 대한 측정불확도로 통합되었다. 그 결과 시료채취단계와 분석단계 중 측정불확도의 대부분은 시료채취단계에서 비롯되고 있음이 확인되었다. 또한 측정불확도 수준을 저감하기 위해서는 표준운영절차에서 규정하고 있는 시료채취방법이 개선되어야 하는데, 본 연구에서 확인된 채취지역의 상대적으로 큰 국지적 이질성(small-scale heterogeneity)으로 말미암아 지역내에서의 채취대상 개체수를 확대하는 것보다 각 개체에서 채취되는 시료량을 늘리는 것이 비용효율적인 개선에 대한 기준이 되었다. 또한 채취방법이 채취지역에서 분포하는 개체들의 이질성을 충분히 극복하며 대표성을 확보할 수 있는가에 대한 검증으로서 분산분석을 적용한 결과, 지역전체의 변화량보다 국지적 변화량이 더 커야 하는 조건을 제시할 수 있었다.
