基于土壤属性相似性聚类与成土环境推测的土壤空间溯源比较研究
Comparative Study on Soil Spatial Provenance Based on Soil Property Similarity Clustering and Pedogenic Environment Inference
摘要
土壤空间溯源在法医土壤学和司法鉴定领域具有重要应用价值.本研究基于安徽省265 个表层土壤样本,比较两种土壤溯源策略:一是基于土壤属性相似性匹配与空间聚类分析,利用土壤光谱与理化属性相似度筛选相似样本,结合DBSCAN 算法确定未知样品可能的来源区域;二是基于成土环境要素逆向推断,采用随机森林模型预测母质、土地利用、地形、气候和植被等环境要素,结合环境要素空间分布图进行来源地推断.通过模拟溯源分析,评估不同溯源策略的精度与适用性.结果表明,相似性匹配策略在空间邻近性强、数据库完善的条件下具有较高的定位精度,而成土环境推测策略在数据库有限或空间异质性强的区域溯源中表现出更优的空间限定能力.两种策略各具优势,融合应用有望进一步提升土壤空间溯源的准确性与空间解析分辨率.
[Objective]Soil spatial traceability has significant application value in forensic soil science and judicial identification.[Method]This study,based on 265 surface soil samples from Anhui Province,compares two soil provenance strategies:(1)a similarity-matching and spatial clustering approach,which filters similar samples using spectral and physicochemical property similarities and applies the DBSCAN algorithm to determine the potential source area of unknown samples;and(2)an inverse inference approach based on pedogenic environmental factors,which employs a random forest model to predict environmental variables such as soil parent material,land use,topography,climate,and vegetation,and infers provenance by integrating spatial distribution maps.By simulating provenance analysis,the accuracy and applicability of the two strategies were evaluated.[Result]The results indicate that the similarity-matching strategy achieves higher localization accuracy under conditions of strong spatial proximity and well-established databases,while the inverse pedogenic environment inference strategy demonstrates superior spatial constraint capabilities in regions with limited databases or strong spatial heterogeneity.[Conclusion]Each strategy has its own advantages,and their integrated application holds promise for further improving the accuracy and resolution of soil spatial provenance analysis.
年,卷(期):
2026, 63(4)
分类号:
P934
机标分类号:
TP391.41
P24
F252
在线出版日期:
2026年7月29日