The hydrologic soil group (HSG) is required for the determination of curve numbers (CN), which are used in a large number of hydrological models to calculate surface runoff during rain events of ungauged catchments. Despite the paramount importance of HSG, it is still based on an assumption made in 1955. We assessed HSG using 249 rainfall simulation experiments carried out in two sets of experiments examining seedbed conditions. One set analyzed the variation within a small (60 ha) catchment while the other examined 32 benchmark soils selected to cover the full range of arable soils on sloping land in temperate humid areas. The assumption made in 1955 that sandy and stony soils have a low HSG (A), while clayey soils of low permeability have a high HSG (D), cannot be supported. Instead, HSG appeared to be governed by surface sealing because CN increased with increasing silt content in the bulk soil and decreasing soil organic matter content in the fine earth fraction. An equation for calculating CN and, in turn, a HSG from these two parameters is provided. This equation also avoids classification-induced errors. The relation holds true for a wide range of rain properties (e.g., rain amount between 0.2 mm and 103 mm; event intensities between 2 mm h⁻¹ and 80 mm h⁻¹), slope gradients (2 %–24 %) and various soil properties (e. g., stone contents up to 40 %, clay contents up to 61 %, sand contents up to 87 %). The CN and, in turn, HSG remained constant throughout the events from the start of runoff until the end of the event, which was up to 49 h later in 32 cases. Autocorrelation within the catchment was strong and the full variation was found within 600 m. Even small catchments are, hence, heterogeneous and their runoff cannot be assigned to a single HSG or a single CN. Rejecting the assumption developed in 1955, which had not been verified since then, has important implications for hydrological modelling but also for the development, implementation and assessment of water conserving land use.

确定曲线数（CN）需要使用水土土壤组（HSG），曲线数（CN）用于大量水文模型中，以计算无污染集水区降雨时的地表径流。尽管HSG极为重要，但它仍基于1955年做出的假设。我们使用249个降雨模拟实验对HSG进行了评估，该实验是在两组检查苗床状况的实验中进行的。一组分析了小面积流域（60公顷）内的变化，而另一组分析了32种基准土壤，这些土壤被选为覆盖温带潮湿地区坡地上的所有耕地。1955年的假设不能支持砂质和石质土壤的HSG（A）低，而低渗透性粘土质的土壤HSG（D）高。反而，HSG似乎受表面密封的控制，因为CN随散土中淤泥含量的增加而增加，而细土部分中土壤有机质的含量则降低。提供了用于根据这两个参数计算CN并进而计算HSG的方程式。该公式还避免了分类引起的错误。该关系适用于广泛的降雨特性（例如，降雨量在0.2 mm和103 mm之间；事件强度在2 mm h之间^-1和80 mm h ^-1），坡度梯度（2％–24％）和各种土壤特性（例如，石头含量最高40％，粘土含量最高61％，沙子含量最高87％）。从径流开始到事件结束，CN和HSG在整个事件中保持恒定，在32例中，直到49小时之后。流域内的自相关性很强，并且在600 m内发现了全部变化。因此，即使是小流域，也是不同的，其径流不能分配给单个HSG或单个CN。拒绝1955年提出的假设，此假设此后一直未得到验证，这对水文建模具有重要意义，对节水土地利用的开发，实施和评估也具有重要意义。