Prediction of the soil water retention curve for structured soil from saturation to oven‐dryness. (20th December 2016)

• Record Type: Journal Article
• Authors: Karup, D.; Moldrup, P.; Tuller, M.; Arthur, E.; de Jonge, L.W.
• Is Part Of: European journal of soil science. Volume 68:Number 1(2017:Feb.)
• Page Start: 57
• Page End: 65
• Publication Date: 2016-12-20
• Subjects: Soil science -- Periodicals; 631.4
• DOI: 10.1111/ejss.12401
• Languages: English
• ISSNs: 1351-0754

Abstract

Abstract : The soil water retention curve (SWRC) is the most fundamental soil hydraulic function required for modelling soil–plant–atmospheric water flow and transport processes. The SWRC is intimately linked to the distribution of the size of pores, the composition of the solid phase and the soil specific surface area. Detailed measurement of the SWRC is impractical in many cases because of the excessively long equilibration times inherent to most standard methods, especially for fine textured soil. Consequently, it is more efficient to predict the SWRC based on easy‐to‐measure basic soil properties. In this research we evaluated a new two‐stage approach developed recently to predict the SWRC based on measurements for disturbed repacked soil samples. Our study involved undisturbed structured soil and took into account the effects of bulk density, organic matter content and particle‐size distribution. Independently measured SWRCs for 171 undisturbed soil samples with organic matter contents that ranged from 3 to 14% were used for model validation. The results indicate that consideration of the silt and organic matter fractions, in addition to the clay fraction, improved predictions for the dry‐end SWRC. The dry‐end results revealed that the smallest matric potential at hypothetical ' zero‐water‐content ' varies between −45 000 and −125 000 m (pF 6.65–7.1) even for soils with similar clay mineralogy. The sensitivity analysis of the two‐stage approach indicated that predicted SWRC results are more sensitive to bulk density than to organic matter content or soil texture. For the soils studied, the two‐stage approach showed reasonable agreement with measured data, with a root mean square error of 0.04 cm 3 cm −3 for the matric potential range from pF 1 to pF 6.6. Highlights: Is a new approach to modelling the soil water retention (SWR) curve applicable to structured soil? The model accurately predicts the full SWR curve with an RMSE of 0.04 cm 3 cm −3 . Prediction accuracy of the wet part of the SWR curve was sensitive to variation in bulk density. The pF at zero water content, which affects the prediction of dry SWR, ranged from 6.65 to 7.10.