Theory and Practice of Evaporation Measurement, with Special Focus on SLS as an Operational Tool for the Estimation of Spatially-Averaged Evaporation
Report no 1335/1/04

Nov 2004

Executive summary

1 Motivation

The need for increased food and timber production has led to increases in irrigated and forestry lands in South Africa. Agriculture and forestry face increased competition for water by industries, municipalities and other groups. This ever-growing demand for water makes it imperative that water resource management procedures and policies are wisely implemented and improved. The accurate assessment of evaporation is essential if this is to be done. Furthermore, the real and/or potential impact of global warming and climate change on all forms of water resources increases the imperative for a knowledge base on water use (and productivity) of commercially-managed lands and natural vegetation. In particular, what is the potential impact of these large-scale changes on evaporation? Implied in this question is the definition of a standard or a benchmark for evaporation measurement or estimation.

The 1998 Republic of South Africa National Water Act refers to the possible prescription, by the Minister, of methods for making a volumetric determination of water for purposes of water allocation. and charges in the case of activities resulting in stream flow reduction. Given this scenario and the above-mentioned demand on water resources it is important to consider how evaporation, one of the main components of the water balance, is to be measured or estimated routinely with reliable accuracy and precision.

So, what is the standard or benchmark for evaporation measurement or its estimation? This important and fundamental question needs to be addressed to achieve the ideal of accurate and reliable evaporation data. This investigation focuses on one method, using a surface layer scintillometer (SLS) and other measurements, to estimate evaporation. The SLS method involves the more fundamental aspects of micrometeorological research. Advances in this field will have direct benefits to water resources assessments and management processes. Examples of areas not easily addressed using the more traditional evaporation measurement and modelling methods include amongst others: measurement of evaporation in riparian zones; determination of open water evaporation from rivers and dams (both large and small). effluent water management from saline dams in the mining environment; measurement of evaporation from slimes tailings, water use of trees and crops in agro-forestry trials; remote sensing applications; determination of baseline evaporation from natural communities (e.g., indigenous forests); irrigation research.

Since John Dalton first introduced the mass transport equation for evaporation estimation, numerous methods for estimating or measuring evaporation have been developed. In certain instances, these methods are accurate and reliable; in others, they are unsuitable or unreliable for certain surfaces or certain times or provide only rough approximations. Currently, there is no single accepted method capable of providing both good spatial and temporal measurements or estimates of evaporation that is reliable and has adequate resolution. This has been confirmed by Drexler et al. (2004) who concluded in their review that there is no single approach that is best for estimating wetland total evaporation. Similarly, Hanson et al. (2004) concluded that no single tested model (using hourly, daily or annual time steps), for capturing intra- and inter-annual forest water and carbon cycle, consistently performed best at all time steps or for all variables considered. However, they did find that inter-model comparisons showed good agreement of water cycle fluxes.

South African researchers have had good success with the heat pulse velocity (HPV) (Dye et al. 1992, Dye and Olbricht 1993) and energy balance methods (Savage et al. 1997). The energy balance is defined by the balance of energy between the net irradiance at a surface and the energy terms used for evaporating water, heating the atmosphere, heating the soil and for photosynthesis. [Net irradiance refers to the flux density of net radiant energy comprised of the incoming solar energy minus the reflected solar plus the incoming long wave energy minus the outgoing long wave. Flux density is the change in an entity, for example net radiant energy per unit time per unit area. Hence net irradiance is the net radiant energy per unit time per unit area with unit W m-2].

The HPV method for estimating sap flow is based on using heat as a tracer for sap flow in plant stems. Temperature needles and a heater probe are inserted in the plant stem and a pulse of heat is applied at regular intervals of time. The rate of change in temperature as measured by the needles a known distance for the heater is related to the sap flow. However, the HPV method is suited only to mono-specific stands of trees, is problematic when scaling from single trees to whole stands and is a destructive method since the tree must be felled to calculate sapwood areas and wound size. The lack of wound size information at the time of measurement makes calculations of transpiration only possible at the end of an experiment. Another method of determining sap flow, the stem steady state heat energy balance technique, has also received attention in South Africa (Savage et al. 1993, 2000).

Another traditional method for estimating evaporation. the Bowen ratio (BR) method, uses measurements of profile differences in air temperature and water vapour pressure above the canopy, surface net irradiance and soil heat flux density. The method assumes that the coefficients of exchange between the surface and the atmosphere for sensible heat [heat transferred from the atmosphere to a surface, or nice versa, that results in a temperature change] and latent energy are identical. Latent energy refers to the energy removed from a surface when water from the surface evaporates. The evaporation causes a cooling effect due to the loss of energy. The term latent is used since largely the process is invisible and results in no temperature change. The BR method, compared to the HPV method, is well suited to mixed species communities such as grassland, but is limited by constraints imposed by both fetch distances and measurement height and is essentially a point measurement. Fetch refers to the upwind distance from the instrument location across a uniformly rough surface such as a forest or a grassland until there is a discontinuous change in surface roughness. Thus, the BR method is not suited to narrow strips (e.g., riparian zones, wetlands, etc.), where fetch distances are short, or above tall vegetation where there are small gradients of air temperature and water vapour pressure (caused by the increased wind turbulence above tall canopies) or above almost any surface in the dry season typical of even the humid areas of South Africa. These gradients may be outside the measurement resolution of the sensors particularly when placed above forest canopies.

The EC method is also a traditional method for direct and point measurement of one or more of sensible heat and latent energy flux density. The method uses high frequency (typically 10 Hz) measurements of air temperature and vertical wind speed (for sensible heat) and air temperature and water vapour pressure (for latent energy flux density). In recent years, direct measurements of turbulent fluxes have been achieved using the eddy covariance (EC) method in KwaZulu-Natal (Monnik et al. 1995, Savage et al. 1995a, b). If only sensible heat flux density is measured using the EC method; then latent energy flux density may be estimated by also measuring net irradiance and soil heat flux density. A drawback of EC measurements is that they, like BR measurements. are point measurements although both sets of measurements are affected by fluxes from upwind source areas. However. the application of the EC method is sometimes problematic. The necessary sensors for vertical wind speed, air temperature and water vapour pressure (or equivalent) must respond very quickly (frequency of 5 Hz or greater) and at the same time must not show noticeable drift. This makes them delicate, expensive and in some cases difficult to calibrate. However. more serious is the airflow distortions by the sensor, mast, etc., as well as the horizontal misalignment measurement errors. In addition, there are technical problems because temporal co-spectra, measured at a fixed local sensor, extend to very low frequencies. To achieve acceptable significance often demands averaging periods of several tens of minutes. Such long averaging periods reduce the temporal resolution and conflict with the requirement of atmospheric stationarity within averaging periods.

Because of the above-mentioned theoretical and practical difficulties associated with the various measurement methods. alternative or complementary flux measurement methods for estimating evaporation have been sought. Recent investigations have demonstrated the potential of using a scintillometer to measure areally-averaged sensible heat flux density. A scintillometer measures the intensity fluctuations of visible or infrared radiation after propagation above the plant canopy of interest. A scintillometer is an instrument that optically measures a parameter associated with refractive index fluctuations, over atmospheric path lengths from 50 m up to at least 3 km and beyond in some cases, from which sensible heat flux density may be calculated. The refractive index fluctuations are caused by interference after the radiation has been scattered by inhornogeneities of the refractive index of air, the latter caused by turbulent fluctuations of air temperature and atmospheric humidity. The parameter is referred to as the structure parameter for refractive index fluctuations Refractive Index Symbol. Once Refractive Index Symbol has been determined, the sensible heat flux density may be estimated from air temperature, atmospheric pressure, effective beam height and beam path length. In contrast to BR and EC measurements, scintillometers provide path-averaged results. Surface layer scintillometer (SLS) systems have beam paths between about 50 and 250 m and large apertures scintillometer (LAS) systems usually operate over distances of between 500 m and 3 km or even up to 10 km for boundary-layer scintillometer (BLS) systems.

The main objective of this project is to estimate a really averaged sensible heat flux density using the SLS method, previously untested in South Africa, and to compare these estimates with those obtained using the more traditional BR and EC methods. A preliminary investigation on the use of the so-called surface renewal (SR) method, also previously untested in South Africa, is also conducted. The SR method is a high frequency, typically 8 Hz; single-temperature measurement technique that also allows for the estimation of sensible heat flux density.

For most of the measurement methods used in this study, the use of the energy balance, requiring measurement of the net irradiance and soil heat flux density, is crucial. The measurements are required together with sensible heat flux density to estimate evaporation. Net irradiance and soil heat flux density measurements are made above and within the surface respectively and need to reflect the spatial variability of the surface and therefore the spatial variation of evaporation. Attention to these measurements for heterogeneous terrain is all the more crucial.

This project falls within the hydroclimatological research programme of the WRC. In this programme the project will contribute to research support for water resources assessment, management and sustainable utilisation in South Africa, by improving the methodologies for monitoring evaporation from surfaces with variable vegetation cover and water surfaces, possibly resulting in a standard for evaporation estimation by which other methods will be judged.