Estimating the soil water retention curve: Comparison of multiple nonlinear regression approach and random forest data mining technique

• Published in: Computers and electronics in agriculture 2020-07, Vol.174, p.105502, Article 105502
• Main Authors: Rastgou, M., Bayat, H., Mansoorizadeh, M., Gregory, Andrew S.
• Format: Article
• Language: English
• Subjects: Bulk density; Data mining; Moisture content; Pedotransfer functions; Performance evaluation; Reliability; Soil properties; Soil structure; Soil texture; Soil water; Soil water retention curve; Soils; Training; van Genuchten
• ISSN: 0168-1699; 1872-7107
• DOI: 10.1016/j.compag.2020.105502

•The RF was compared to NLR method and Rosetta-based PTFs to predict the SWRC.•The NLR method had better performance due to higher reliability in the testing step.•The RF and NLR-based PTFs performed better than the Rosetta-based PTFs.•In the absence of moisture points, OM and Ks can be suitable predictors for SWRC.•dg and δg can be suitable alternatives for textural fractions in predicting SWRC.•Total porosity and bulk density have the same effect in predicting the SWRC. This study evaluates the performance of the random forest (RF) method on the prediction of the soil water retention curve (SWRC) and compares its performance with those of nonlinear regression (NLR) and Rosetta-based pedotransfer functions (PTFs), which has not been reported so far. Fifteen RF and NLR-based PTFs were constructed using readily-available soil properties for 223 soil samples from Iran. The general performance of RF and NLR-based PTFs was quantified by the integral root mean square error (IRMSE), Akaike’s information criterion (AIC) and coefficient of determination (R2). The results showed that the accuracy of the RF-based PTFs was significantly (P