Anderson Donovan, Okuda Kei, Hess Ann, Nanba Kenji, Johnson Thomas, Takase Tsugiko, Hinton Thomas
Institute of Environmental Radioactivity, Fukushima University, 960-1248, Fukushima Prefecture, Fukushima, Kanayagawa, Japan.
Faculty of Human Environmental Studies, Hiroshima Shudo University, 731-3195, Hiroshima Prefecture, Hiroshima, Japan.
J Environ Radioact. 2019 Feb;197:101-108. doi: 10.1016/j.jenvrad.2018.12.009. Epub 2018 Dec 10.
Aggregated transfer factors (T; m kg) are often used to predict radionuclide activity concentrations in biota (Bq kg) from soil contamination levels (Bq m). Inherently large uncertainties in T values severely limit their predictive power. Many published T values have been derived from radionuclide deposition onto soil following weapons fallout, or the accidents at Chernobyl and Fukushima. In many cases the soil data used to derive a T value were collected for other purposes, and the spatial resolution of the soil data is much less than that of the biota data to which it is paired. We hypothesized that this disassociation and imprecision in paring deposition density and biota data may contribute to the large variations observed in T values. We tested the hypothesis by deriving T values for Japanese wild boar in two ways. One method used paired deposition density-biota contamination levels, with the soil data collected from each boar trap site. The second method used a soil radioactivity density map, of relatively low spatial resolution, generated by the Japanese government agency MEXT for fallout from the Fukushima accident. We hypothesized that T values derived from the method using paired deposition density-wild boar data would have less variation. Initial statistical test suggested significant differences in the predictive power of the two methods. However, removal of suspected outliers in the MEXT data set decreased the statistical differences and indicated that collecting Cs soil deposition density measurements in the field did not reduce the large variation in our T values. More importantly, both methods revealed that soil contamination levels are a poor predictor of radiocesium concentrations in boar (r < 0.23). The inadequacies of T to predict wild boar Cs concentrations is an ominous indication of the lack of applicability of the T model as a rigorous research parameter. T values are best suited for their original intended purpose: upper tier, screening level computations. Further studies on how to reduce uncertainty when predicting Cs concentrations in biota are needed to thoroughly understand the transfer of radiocesium within the environment.
聚集转移因子（T；m/kg）通常用于根据土壤污染水平（Bq/m）预测生物群中的放射性核素活度浓度（Bq/kg）。T值本身存在的巨大不确定性严重限制了它们的预测能力。许多已发表的T值来自武器 fallout 后放射性核素在土壤上的沉积，或切尔诺贝利和福岛事故。在许多情况下，用于推导T值的土壤数据是为其他目的收集的，并且土壤数据的空间分辨率远低于与其配对的生物群数据。我们假设在配对沉积密度和生物群数据时的这种脱节和不精确可能导致观察到的T值存在很大差异。我们通过两种方式推导日本野猪的T值来检验这一假设。一种方法使用配对的沉积密度-生物群污染水平，土壤数据从每个野猪陷阱地点收集。第二种方法使用日本政府机构文部科学省为福岛事故的fallout生成的空间分辨率相对较低的土壤放射性密度图。我们假设使用配对沉积密度-野猪数据的方法推导的T值变化较小。初步统计测试表明两种方法的预测能力存在显著差异。然而，去除文部科学省数据集中的可疑异常值降低了统计差异，并表明在现场收集铯土壤沉积密度测量值并没有减少我们T值的巨大变化。更重要的是，两种方法都表明土壤污染水平对野猪体内放射性铯浓度的预测能力很差（r < 0.23）。T值不足以预测野猪铯浓度，这是T模型缺乏作为严格研究参数适用性的不祥迹象。T值最适合其最初的预期用途：高层筛选水平计算。需要进一步研究如何在预测生物群中铯浓度时降低不确定性，以彻底了解放射性铯在环境中的转移。
