基于分形理论和地质统计学的表层土壤颗粒大小分布变化特征

Characterizing Variation of Topsoil Particle Size Distribution Based on Fractal Theory and Geostatistics

【目的】探究多角度、多尺度全面揭示土壤颗粒大小分布特征的方法体系,探索更加简单而又可综合定量分析评价土壤质量及其演变过程的手段。【方法】以计算的土壤颗粒体积分形维数(Dv)作为分析对象,建立集分形理论、传统统计学、分离土壤颗粒Dv的双对数图和地质统计学等理论和方法为一体的表层土壤颗粒大小分布变化特征分析方法体系,并基于此从点和区域2个尺度系统全面地研究表层土壤颗粒大小分布变化特征。【结果】粒径<10μm的土壤颗粒累积体积百分含量与Dv呈显著正相关,相关系数为0.46,而>50μm以内的土壤颗粒累积体积百分含量与Dv呈显著负相关,相关系数为0.63,土壤颗粒Dv值越小,土壤颗粒越粗;最大和最小Dv的双对数拟合图主要变化靠近拟合的直线,R2值均在0.9以上,拟合结果理想,分离的Dv能够包含整个土壤颗粒大小分布变化的程度。不同土壤有机质含量组之间土壤颗粒Dv存在一定差异性,土壤颗粒Dv能够客观地表征农田土壤质量的变化;随着高程增加,Dv表现相对比较复杂;褐土土壤颗粒Dv均值最大,潮土均值最小;粮田、园地和草地之间的土壤颗粒Dv差异性不明显。基于土壤颗粒Dv与环境因素关系分析的回归克里格法预测结果较为准确,区域和样点土壤颗粒Dv空间分布格局一致,较客观地反映了区域表层土壤颗粒大小分布变化特征。【结论】建立的方法体系能够从多角度、多尺度全面地反映土壤颗粒大小分布特征,分析结果符合实际,土壤颗粒Dv可以作为定量化分析评价和表征土壤质量及其演变过程的手段。

[Objective] This study aimed to explore the methodology which can help fully reveal the characteristics of soil particle size distribution（PSD） on multi-angle and multi-scale, and to find more simple and comprehensive means of quantitative analysis and evaluation of soil quality and its evolution. [Method] The study calculated the volume fractal dimension（Dv） and analyzed the variation characteristics of soil particle size distribution based on fractal theory, traditional statistics, separated soil particles Dv log-log graph and geostatistics from point and regional scale. [Result] Cumulative volume percentage of soil particles with diameter〈10 μm showed a significant positive correlation with Dv while that of soil particles with diameter〉50 μm was negatively related to Dv. The smaller soil particles Dv was, the coarser soil texture was. The end portions of log-log scatter plot for the maximum and minimum Dv bent downward with the fitting straight lines close to major changes, the value of R2 are above 0.9, and the fitting results are ideal. The separation of Dv contained the entire soil particle size distribution（PSD） of the degree of change. There were some differences of soil particles Dv between different groups of soil organic matter content, and soil particles Dv was able to objectively characterize the changes of soil quality of a farmland. With elevation increasing, the performance of soil particles Dv was relatively complex. The mean value of soil particles Dv for cinnamon soil was maximum, mean was minimum for fluvo-aquic soil. The results of variance analysis showed that the difference in Dv values between grain field, garden land and vegetable land was not obvious. Prediction results with regression kriging were more accurate based on the analysis results about the relationship between soil particle Dv and environmental variables. Spatial distribution pattern of the regional and sampling Dv was consistent, and objectively reflected the variation characteristics of PSD. [Conclusion] Methodology established in the study can reflect the characteristics of PSD from multi-angle and multi-scale. The study results matched the actuality, and soil particles Dv can be used as a means to characterize and evaluate the soil quality and its changing process.