Constraints in calculations of evaporative losses in arid climates using the stable isotope composition of water

Thursday, 18 December 2014
Grzegorz Skrzypek1, Adam Mydlowski2, Shawan Dogramaci3, Paul Hedley3, John J Gibson4 and Pauline F Grierson1, (1)University of Western Australia, Crawley, WA, Australia, (2)Polish Geological Institute – National Research Institute, Wroclaw, Poland, (3)Rio Tinto Iron Ore, Perth, WA, Australia, (4)University of Victoria, Victoria, BC, Canada
Accurate quantification of evaporative losses to the atmosphere from surface water bodies is essential for calibration and validation of hydrological models, particularly in remote arid and semi-arid regions, where rivers and lakes are generally minimally gauged. In this study, we reviewed and combined the most recent equations for estimation of evaporative losses based on the revised Craig-Gordon model. We designed new software, called Hydrocalculator, which allows quick and robust estimation of evaporative losses based on the isotopic composition of water. We validated Hydrocalculator by testing the range of uncertainty in the estimation of evaporative losses in arid climates by cross-validating a simplified stable isotope model with field pan evaporation experiments. The use of standardized pans (1.2 m diameter, volume 300 dm3) in hot and dry climates (temperature 29°C and relative humidity between 19 and 26%) allowed simulation of fast evaporation from shallow water bodies. Several factors may contribute to the uncertainty in the evaporative loss calculations. The analytical uncertainty in the determination of the stable isotope composition of water may contribute to ~0.6% for δ18O and ~1.4% for δ2H. The model is less sensitive to uncertainty in climatic variables and an uncertainty of 1°C in air temperature will result only in the ~0.1% uncertainty in δ18O and δ2H. However, uncertainty in relative humidity of 10% will result in an uncertainty in the final outcome of 0.4% (δ18O) and 1.0% (δ2H). Significantly higher uncertainty in evaporative loss estimation is thus associated with uncertainty in ambient air moisture estimation or analysis. An error of 20‰ in δ2H and 5.0‰ in δ18O will result in a maximum difference of 2.4% (δ2H) and 1.7% (δ18O) in the final calculations. Hydrocalculator can thus provide accurate, rapid and cost-effective insight into the water balance of surface water pools. We used the new software to determine the origin of surface water and the progress of evaporation in pools along several dryland rivers in northern Western Australia.