Comparison of Numerical and Steady-state Models to Predict Vertical Water Flux With Application to Water Delivery Canals

In the U.S. alone, as much as 50 percent of irrigation water carried within unlined water delivery canals may be lost during transport, resulting in wasted resources and degradation of cropland due to salinization and low water tables in some locations. To minimize this loss, it is important to quantify where and when the seepage occurs. The use of heat flux as a tracer has become an accepted technique for detecting interactions between surface and groundwater sources. Most commonly, flow is assumed to be only in the vertical direction, through an isotropic, homogeneous porous medium, allowing the use of the one- dimensional advection-conduction equation for heat transport. Both analytical and numerical approaches to the transient and steady-state forms of the equation have been employed; however, few studies have considered the variability between these approaches. This study uses synthetic data to explore systematic differences between numerical and analytical transient models and a steady-state model. The models were then applied to one year of temperature data collected between July 2006 and 2007 at several large water delivery canals in eastern Colorado. Preliminary results show that the numerical transient model best fits the observed data, and variability in error during the modeled period suggest that seepage rates vary throughout the year.