Weipeng Wang, Jianli Liu, Bingzi Zhao, Jiabao Zhang, Xiaopeng Li, Yifan Yan
State Key Laboratory of Soil and Sustainable Agriculture, Institute of Soil Science, Chinese Academy of Sciences, Nanjing, China; Graduate University of Chinese Academy of Sciences, Beijing, China.
State Key Laboratory of Soil and Sustainable Agriculture, Institute of Soil Science, Chinese Academy of Sciences, Nanjing, China.
PLoS One. 2015 Apr 30;10(4):e0125048. doi: 10.1371/journal.pone.0125048. eCollection 2015.
Mathematical descriptions of classical particle size distribution (PSD) data are often used to estimate soil hydraulic properties. Laser diffraction methods (LDM) now provide more detailed PSD measurements, but deriving a function to characterize the entire range of particle sizes is a major challenge. The aim of this study was to compare the performance of eighteen PSD functions for fitting LDM data sets from a wide range of soil textures. These models include five lognormal models, five logistic models, four van Genuchten models, two Fredlund models, a logarithmic model, and an Andersson model. The fits were evaluated using Akaike's information criterion (AIC), adjusted R2, and root-mean-square error (RMSE). The results indicated that the Fredlund models (FRED3 and FRED4) had the best performance for most of the soils studied, followed by one logistic growth function extension model (MLOG3) and three lognormal models (ONLG3, ORLG3, and SHCA3). The performance of most PSD models was better for soils with higher silt content and poorer for soils with higher clay and sand content. The FRED4 model best described the PSD of clay, silty clay, clay loam, silty clay loam, silty loam, loam, and sandy loam, whereas FRED3, MLOG3, ONLG3, ORLG3, and SHCA3 showed better performance for most soils studied.
经典粒度分布（PSD）数据的数学描述常用于估算土壤水力特性。激光衍射法（LDM）如今能提供更详细的PSD测量结果，但推导一个能表征整个粒径范围的函数是一项重大挑战。本研究的目的是比较18种PSD函数对来自广泛土壤质地的LDM数据集的拟合性能。这些模型包括5种对数正态模型、5种逻辑模型、4种范格内uchten模型、2种弗雷德隆德模型、1种对数模型和1种安德森模型。使用赤池信息准则（AIC）、调整后的R2和均方根误差（RMSE）对拟合进行评估。结果表明，弗雷德隆德模型（FRED3和FRED4）对大多数研究土壤的性能最佳，其次是一种逻辑生长函数扩展模型（MLOG3）和三种对数正态模型（ONLG3、ORLG3和SHCA3）。大多数PSD模型对粉粒含量较高的土壤性能较好，对黏粒和砂粒含量较高的土壤性能较差。FRED4模型最能描述黏土、粉质黏土、黏壤土、粉质黏壤土、粉质壤土、壤土和砂壤土的PSD，而FRED3、MLOG3、ONLG3、ORLG3和SHCA3对大多数研究土壤表现出更好的性能。
