Various ¹³⁷Cs conversion models have been used to predict soil erosion. A knowledge of the initial distribution of ¹³⁷Cs is vital to improve soil loss prediction. The objectives were to (1) characterize the dynamic distributions of ¹³³Cs near the soil surfaces in multiple rains spiked with ¹³³Cs using a rainfall simulator, and (2) further estimate the relaxation mass depth (H) for use in an improved ¹³⁷Cs mass balance model. Three silt loam soils were packed to soil boxes (1 × 0.5 × 0.1 m) and rained on at 30 mm h⁻¹ and 10% slope in five consecutive rains. Soil cores (62.5 mm i.d.) were taken following each rain and were sectioned in the 1–10-mm intervals. Event runoff and sediment were collected. Cs⁺ was extracted with an NH₄OAc method using a centrifuge procedure and analyzed with Inductively Coupled Plasma-Mass Spectrometer. The results showed initial Cs distributions near the soil surfaces were approximately exponential, and the mean H value could adequately characterize the initial Cs distributions for each soil for any sequence of rains. The averaged H values were 0.338, 0.304, and 0.354 g cm⁻² for the three soils, equivalent to the soil depths of 2.76, 2.20, and 2.58 mm, indicating that majority of Cs was distributed within the top 3-mm depth. Cs⁺ was strongly adsorbed by the soils as indicated by the large K_d values of >32.6 L kg⁻¹, showing the limited mobility of the freshly deposited Cs⁺ in the soils. Better H estimation should lead to more accurate soil loss prediction using the improved Cs mass balance model.

各种¹³⁷ Cs转换模型已用于预测土壤侵蚀。了解¹³⁷ Cs的初始分布对于改善土壤流失预测至关重要。目的是（1）使用降雨模拟器描述在多次降雨中以¹³³ Cs加标的土壤表面中¹³³ Cs的动态分布，以及（2）进一步估算弛豫质量深度（H），以用于改进的¹³⁷ Cs质量平衡模型。将三种淤泥质壤土包装到土箱（1×0.5×0.1 m）中，并在30 mm h ^-1上降雨连续五次降雨的坡度为10％。每次下雨后取土芯（内径62.5毫米），并以1-10毫米的间隔切开。收集事件径流和沉积物。使用NH ₄ OAc方法通过离心程序提取Cs ⁺，并使用电感耦合等离子体质谱仪进行分析。结果表明，土壤表面附近的初始Cs分布近似呈指数分布，并且平均H值可以充分表征任何降雨序列下每种土壤的初始Cs分布。平均H值为0.338、0.304和0.354 g cm ^-2对于这三种土壤，其土壤深度分别为2.76、2.20和2.58 mm，这表明大多数Cs分布在顶部3 mm深度内。如> 32.6 L kg ^-1的大K _d值所示，Cs ⁺被土壤强烈吸附，表明新鲜沉积的Cs ⁺在土壤中的活动性有限。使用改良的Cs质量平衡模型，更好的H估计将导致更准确的土壤流失预测。