Hesham M. Ibrahim1,2*and Ali M. Al-Turki1
1Department of Soil Science, King Saud University, P.O. Box 2460, Riyadh, 11451, Saudi Arabia.
2Department of Soils and Water, Faculty of Agriculture, Suez Canal University, Ismailia, 41522, Egypt.
Accurate measurement or estimation of the hydrological soil parameters are needed to determine the availability of soil water to plants, and to model water flow and solute transport in the vadose zone. In dry regions, effective irrigation management is crucial to maintain crop production and sustain limited water resources. Effective irrigation requires good knowledge of soil water content (SWC) in the root zone. However, measurement of SWC in the root zone at the watershed scale is expensive and time consuming. Alternatively, pedotransfer functions (PTFs) are used to predict the soil water retention characteristics (SWRC) and SWC from easily measured soil properties. In this study, we used multiple linear regression (MLR) to develop two site-specific PTFs, a point (MLRP) and a parametric (MLRF), using soil properties of 219 soil samples collected from Jazan watershed, southwest of Saudi Arabia. The accuracy of the developed PTFs and four existing PTFs to determine the SWRC, predict soil water content (SWC) at -10, -33, and -1500 kPa, and to estimate available water content (AWC) was assessed. The predicted SWC was compared with measured SWC that was continuously monitored at 18 locations in the Jazan watershed. SWC was measured at four depths of 25, 50, 75, and 100 cm using 5TE Decagon sensors that were connected to an EM50G micrologger. The predicted AWC values were georeferenced to a base map of the Jazan watershed and the geostatistical analyst function within ArcGIS 10.3 was used to interpolate the AWC values, and to produce a general map showing the AWC distribution in the Jazan watershed. Results showed that the MLRP and the Schaap et al. (2001) PTFs produced the best estimate of SWC, with smaller RMSE (0.023-0.053 cm3 cm-3), and larger D-index (0.8-0.9) and AIC (-303.7 to -240.7), respectively. The other tested PTFs: Rawls et al. (1982); Gupta and Larson (1979); and Vereecken et al. (1989) produced much larger RMSE and showed an overestimation in the predictions of SWC at all matric potentials. The largest prediction errors in the estimation of SWC were observed at matric potential close to FC (-33 kPa). For the AWC, the Schaap PTF provided the best prediction (RMSE = 0.014 cm3 cm-3, D-index = 0.93, AIC = -359.9), followed by the MLRP PTF (RMSE = 0.027 cm3 cm-3, D-index = 0.83, AIC = -302.1), whereas the Vereecken, Gupta and Larson, and the MLRF PTFs produced less accurate predictions of the AWC. The MLRP PTF proved to be more accurate compared to other tested PTFs in the prediction of SWC at both FC and PWP. In contrast, the MLRF PTF produced relatively large error in the estimation of SWC at FC. Nevertheless, the predicted values are usually accurate enough for the simulation processes, especially if direct measurement of the SWRC is not available.
*Corresponding author. Tel.: +966 50 172 8656; fax: +966 1 467 8440. E-mail address: firstname.lastname@example.org
This research was funded by the National Plan for Science, Technology and Innovation (MAARIFAH), King Abdulaziz City for Science and Technology, Kingdom of Saudi Arabia, Award Number 12-AGR2575-02.