Evaporation from water supply reservoirs: An assessment of uncertainty

Summary

Many water resource investigations require an estimate of evaporation from water supply reservoirs. Despite its recognised limitations, the pan coefficient method for estimating reservoir evaporation is widely used in industry due to its moderate data requirements. This paper presents a framework to quantify the uncertainties associated with estimates of reservoir evaporation generated using the pan coefficient method. The uncertainties associated with each of the inputs are quantified using a combination of statistical techniques, including Bayesian statistics and subjective judgement; before being combined using Monte Carlo simulation. The framework is applied to three reservoirs in the Werribee River catchment, Australia. The 95% probability intervals surrounding the estimates of reservoir evaporation are as large as ±40% of the best estimate. The largest contributing factor to uncertainty is shown to be the estimation of pan evaporation at reservoirs without a climate station, followed by the uncertainty surrounding the annual pan coefficient itself. The assessment shows that the largest reductions in uncertainty can be achieved by installing an evaporation pan at the reservoir.

Introduction

Understanding the magnitude of evaporation from water supply reservoirs is an important component of water resource management. The effect of evaporation is considered during the design of water supply reservoirs and subsequent reservoir yield investigations. During the operation of a water supply system, the losses due to evaporation are taken into account before water is allocated to consumptive users. Investments in water saving measures are also becoming common practice in water resources management. Managers are looking to reduce inefficiencies in water supply systems, including the evaporative loss of water from reservoirs. In response, methods to reduce evaporation from storages have been developed (e.g. Alvarez et al., 2006). In some regions, climate change is expected to lead to an increase in evaporation, and potentially a reduction in the water yield (e.g. Adeloye et al., 1999). Therefore, an estimate of reservoir evaporation is an important precursor to the design and ongoing operation of a water supply reservoir.

There are many methods available that estimate evaporation from an open water body, also known as lake evaporation and referred to as reservoir evaporation throughout this paper. Methods include the water budget method, energy budget method, eddy-correlation method, mass-transfer approach, the Penman method, combination equation and the pan coefficient method (Dingman, 1994). A significant limitation of most of these methods is that they require several meteorological variables, such as wind velocity and humidity, to be measured or estimated at the reservoir. While these measurements have been taken for specific studies, they are not commonly available at water supply reservoirs. The use of remote sensing information to estimate reservoir evaporation is promising and has been trialled in some studies (e.g. Guerschman et al., 2009), however, the application of these techniques is not widespread. Therefore, although the pan coefficient method is generally considered to be the least accurate method (Grayson et al., 1996, Winter, 1981), it is the only existing method that does not require site specific measurements and is, therefore, commonly used in water resources investigations.

Despite the limitations of the pan coefficient method being recognised (e.g. Tanny et al., 2008), there has been no attempt to quantify the uncertainty of reservoir evaporation calculated with this method. However, before making decisions such as an investment in the reduction of evaporation from a reservoir or the allocation of water to consumptive users, managers need to be aware of the magnitude of uncertainties associated with estimates of reservoir evaporation. For example, the economic viability of a water savings project will depend on the volume of water that can be recovered. In this instance the uncertainty surrounding the volume of evaporation contributes to the overall investment risk and this information should be disclosed to those making the investment decision. In some instances the magnitude of uncertainty may be unacceptably high and managers may consider initial investments that may lead to reductions in this uncertainty. A comprehensive uncertainty analysis identifies the main contributing factors to the overall uncertainty, and enables managers to more efficiently allocate resources to improve estimates of reservoir evaporation. The assessment also shows the expected benefit of reducing these uncertainties. This paper presents a framework to quantify the greatest sources of uncertainty in estimates of reservoir evaporation. First, an overview of the pan coefficient method and approach used to quantify and combine uncertainties is presented in the Section “Overview of approach”. Second, in the Section “The sources of uncertainty”, each step of the pan coefficient method is described and the uncertainties associated with each step identified and quantified. Third, the framework is applied to estimate the range of possible evaporation volumes from three water supply reservoirs in the Werribee River catchment, Australia (Section “The Werribee River catchment case study”). A final synthesis is provided in the Section “Discussion”.