Using an inverse modelling approach to evaluate the water retention in a simple water harvesting technique

• Published in: Hydrology and earth system sciences Vol. 13; no. 10; pp. 1979 - 1992
• Main Authors: Verbist, K, Cornelis, W.M, Gabriels, D, Alaerts, K, Soto, G
• Format: Journal Article
• Language: English
• Published: Katlenburg-Lindau Copernicus GmbH 01.01.2009; Copernicus Publications
• Subjects: arid zones; infiltration (hydrology); mathematical models; rainfall simulation; semiarid zones; simulation models; soil water content; soil water retention; water harvesting; Chile
• ISSN: 1027-5606; 1607-7938; 1607-7938
• DOI: 10.5194/hess-13-1979-2009

In arid and semi-arid zones, runoff harvesting techniques are often applied to increase the water retention and infiltration on steep slopes. Additionally, they act as an erosion control measure to reduce land degradation hazards. Nevertheless, few efforts were observed to quantify the water harvesting processes of these techniques and to evaluate their efficiency. In this study, a combination of detailed field measurements and modelling with the HYDRUS-2D software package was used to visualize the effect of an infiltration trench on the soil water content of a bare slope in northern Chile. Rainfall simulations were combined with high spatial and temporal resolution water content monitoring in order to construct a useful dataset for inverse modelling purposes. Initial estimates of model parameters were provided by detailed infiltration and soil water retention measurements. Four different measurement techniques were used to determine the saturated hydraulic conductivity (Ksat) independently. The tension infiltrometer measurements proved a good estimator of the Ksat value and a proxy for those measured under simulated rainfall, whereas the pressure and constant head well infiltrometer measurements showed larger variability. Six different parameter optimization functions were tested as a combination of soil-water content, water retention and cumulative infiltration data. Infiltration data alone proved insufficient to obtain high model accuracy, due to large scatter on the data set, and water content data were needed to obtain optimized effective parameter sets with small confidence intervals. Correlation between the observed soil water content and the simulated values was as high as R2=0.93 for ten selected observation points used in the model calibration phase, with overall correlation for the 22 observation points equal to 0.85. The model results indicate that the infiltration trench has a significant effect on soil-water storage, especially at the base of the trench.