根据颗粒大小分布估计土壤水分特征曲线:分形模型的应用

根据UNSODA数据库和文献中的 1 1种质地共 5 5 4个样品的颗粒大小分析和水分特征曲线资料 ,对Tyler Wheatcraft、Brooks Corey和Rieu Sposito三种分形模型在预测土壤水分特征曲线中的适用性进行了研究。结果表明 ,Brooks Corey形式的分形模型预测精度高于其它两种模型。同时本文还指出了这三种模型适用的土壤质地范围 ,即Brooks Corey模型对于中、粗质地的土壤预测效果好于另外两种模型 ,Rieu Sposito分形模型则适用于细质地土壤 ,Tyler Wheatcraft模型的预测误差界于二者之间 ,也适用于中、粗质地的土壤。     

Model Development to Predict CEC Using the Intelligence Data Mining Approaches

ABSTRACT

Using easily measurable soil properties and pedotransfer functions (PTFs) is a time-saving, non-destructive and cost-saving way in the prediction of the cation exchange capacity (CEC). The purpose of this study was to compare and evaluate the regression tree (RT), multiple linear regression (MLR) and Mamdani fuzzy inference system (MFIS) in estimating CEC. For this work, 100 soil samples from unsaturated soil hydraulic database (UNSODA) data-set were used. %Organic matter (OM), bulk density (BD), the geometric mean particle diameter (dg) and fractal dimension of particle size (D) were applied as the input predictive variables. First, the type of relationship between easily measurable soil properties and CEC was investigated and, then used for the development of PTFs and fuzzy membership functions. The results showed that MLR method was developed only based on %OM (r = 0.68, p < .01) and D (r = 0.68, p < .01). While in the RT method, all of the predictive variables were appeared in the tree-like based on their correlation coefficient with CEC. The D and %OM also were considered as input variables in developing fuzzy membership functions. Results also revealed that RT method had a higher performance than MLR and MFIS in the estimation of CEC with the highest coefficient of determination (R² = 0.77), smallest root-mean-square error (RMSE = 5.14 meq/100_gsoil), normalized root-mean-square error (NRMSE = 0.25 meq/100_gsoil) and mean error (ME = ?1.80 meq/100_gsoil). In addition, the MFIS had a higher efficiency than the MLR in the CEC estimation.     

Developing Pedotransfer Functions for Predicting FC and PWP

Using pedotransfer functions (PTF) is a useful way for field capacity (FC) and permanent wilting point (PWP) prediction. The aim of this study was to model PTF to estimate FC and PWP using regression tree (RT) and stepwise multiple linear regressions (SMLR). For this purpose, 165 and 45 soil samples from UNSODA and HYPRES datasets were used for development and validation of new PTFs, respectively. %Clay, geometric mean diameter (dg), and bulk density (BD) were selected as predictor variables due to the highest correlation and lowest multicollinearity. The results showed that clay percentage with W* = 0.89 and dg with W* = ?0.57 were the most effective variables to predict PWP and FC, respectively. The RT method had a better performance (R² = 0.80, ME = ?0.002 cm³cm^?3, RMSE = 0.05 cm³cm^?3 for FC and R² = 0.85, ME = 0.003 cm³cm^?3, RMSE = 0.03 cm³ cm^?3 for PWP) than SMLR in estimation of FC and PWP.     

Simple methods for estimating field capacity using Mamdani inference system and regression tree

Using easily measurable soil properties could save time and cost for field capacity (FC) prediction. The objective of this study was to compare Mamdani fuzzy inference system (MFIS) and regression tree (RT) for FC predicting using such properties. One hundred and sixty-five soil samples from Unsaturated Soil hydraulic database data-set and 45 from Hydraulic Properties of European Soils data-set were used for the development and validation of MFIS and RT, respectively. Fuzzy rules and tree diagram based on the relationships between these predictor variables and the response variable FC were defined and 48 rules were written. Results showed a positive linear relevancy in terms of standardized independent variable weight, W*, between clay content and FC and negative linear relevancy between geometric mean particular size diameter (d_g) and FC. Among predictor variables, d_g (W* = 0.81) and bulk density (BD) (W* = 0.49) had the highest and lowest influence on FC, respectively. A tree diagram is presented for the prediction of FC using clay content, dg, and BD. Overall, based on statistical parameters, RT method (R² = 0.78, geometric mean error (GME) = 1.02, mean error (ME) = 0.01 cm³ cm^?3, and root mean square error (RMSE) = 0.04 cm³ cm^?3) showed a higher performance than MFIS method (R² = 0.72, GME = 1.16, ME = 0.08 cm³ cm^?3, and RMSE = 0.06 cm³ cm^?3) to predict FC.