Uncertainty of canal seepage losses estimated using flowing water balance with acoustic Doppler devices

- Magnitude and uncertainty of canal seepage estimated for 60 water balance tests.
- Steady and unsteady flow, lagged and simultaneous discharge measurements considered.
- Estimated seepage uncertainty substantial, primarily influenced by canal flow error.
- Changing stage and storage have sizable impact on seepage magnitude and uncertainty.
- Recommendations for recognizing and reducing water-balance seepage test uncertainty.

SummarySeepage losses from unlined irrigation canals amount to a large fraction of the total volume of water diverted for agricultural use, posing problems to both water conservation and water quality. Quantifying these losses and identifying areas where they are most prominent are crucial for determining the severity of seepage-related complications and for assessing the potential benefits of seepage reduction technologies and materials. A relatively easy and inexpensive way to estimate losses over an extensive segment of a canal is the flowing water balance, or inflow-outflow, method. Such estimates, however, have long been considered fraught with ambiguity due both to measurement error and to spatial and temporal variability. This paper presents a water balance analysis that evaluates uncertainty in 60 tests on two typical earthen irrigation canals. Monte Carlo simulation is used to account for a number of different sources of uncertainty. Issues of errors in acoustic Doppler flow measurement, in water level readings, and in evaporation estimates are considered. Storage change and canal wetted perimeter area, affected by variability in the canal prism, as well as lagged vs. simultaneous measurements of discharge at the inflow and outflow ends also are addressed. Mean estimated seepage loss rates for the tested canal reaches ranged from about −0.005 (gain) to 0.110 m3 s−1 per hectare of canal wetted perimeter (or −0.043 to 0.95 m d−1) with estimated probability distributions revealing substantial uncertainty. Across the tests, the average coefficient of variation was about 240% and the average 90th inter-percentile range was 0.143 m3 s−1 per hectare (1.24 m d−1). Sensitivity analysis indicates that while the predominant influence on seepage uncertainty is error in measured discharge at the upstream and downstream ends of the canal test reach, the magnitude and uncertainty of storage change due to unsteady flow also is a significant influence. Recommendations are presented for conducting field water balance tests to recognize and reduce uncertainty in canal seepage estimates.