Abstract Legacy soil data currently represents an irreplaceable source of data for many national and international applications. Their full use at the international level largely depends on adequate harmonization. In this study, we adopted different methods of taxonomic distance and reclassification of soil properties to convert a large-scale database originating from an extensive national mapping campaign of agricultural soils in Czechoslovakia (1961–1971), according to the criteria of the World Reference Base for Soil Resources (WRB), version 2014. The harmonization was effectuated at two levels of detail represented by two different types of soil mapping units [soil type and general soil representative (GSR)] and resulted in the conversion of original soil units into their counterparts in the WRB (Reference Soil Groups). The results showed a good potential of the adopted methods for soil data harmonization. The conceptual differences of the two harmonised classifications decreased the applicability of the taxonomic distance in the case of broadly defined and less developed soil classes lacking specific diagnostic features. The replacement of a simple conversion by soil association was a relevant approach, especially at the soil type level, where the simple one-to-one conversion showed only poor accuracy (39.5%). The accuracy at the GSR level was satisfactory (reaching ca 60%) and increased with soil association (72%). The conversion accuracy of different soil classes significantly differed from very high (e.g. 92% in Luvisols) to completely mismatched (Technosols and Histosols). The latter can be attributed to a questionable determination of soil classes in legacy maps. The study showed the potential of taxonomic distance methods in soil data harmonization and the limits of historical databases resulting from the way data were collected and recorded.

中文翻译：

---

大型国家土壤数据库与 2014 年世界土壤资源参考基地的协调

摘要 传统土壤数据目前代表了许多国家和国际应用不可替代的数据来源。它们在国际层面的充分利用在很大程度上取决于充分协调。在这项研究中，我们采用了不同的分类距离和土壤特性重新分类方法，根据世界参考基地的标准，转换源自捷克斯洛伐克（1961-1971 年）广泛的国家农业土壤测绘活动的大型数据库土壤资源 (WRB)，2014 年版。协调是在由两种不同类型的土壤制图单位 [土壤类型和一般土壤代表 (GSR)] 代表的两个详细级别上实现的，并导致将原始土壤单位转换为它们的WRB（参考土壤组）中的对应物。结果表明所采用的土壤数据协调方法具有良好的潜力。两种统一分类的概念差异降低了分类距离在定义广泛和缺乏特定诊断特征的欠发达土壤类别的情况下的适用性。用土壤关联代替简单转换是一种相关的方法，特别是在土壤类型级别，简单的一对一转换的准确度很低（39.5%）。GSR 水平的准确度是令人满意的（达到约 60%）并且随着土壤关联性增加（72%）。不同土壤类别的转换精度从非常高（例如 Luvisols 为 92%）到完全不匹配（Technosols 和 Histosols）有显着差异。后者可归因于遗留地图中对土壤类别的可疑确定。该研究显示了分类距离方法在土壤数据协调方面的潜力以及由于收集和记录数据的方式而导致的历史数据库的局限性。