Agricultural and Forest Meteorology 106 (2001) 153–168A comparison of methods for determining forestevapotranspiration and its components: sap-flow, soil water budget,eddy covariance and catchment water balanceKell B. Wilsona,∗, Paul J. Hansonb, Patrick J. Mulhollandb,Dennis D. Baldocchic, Stan D. WullschlegerbaAtmospheric Turbulence and Diffusion Division, NOAA, P.O. Box 2456, Oak Ridge, TN 37831, USAbEnvironmental Sciences Division, Oak Ridge National Laboratory, P.O. Box 2008, Oak Ridge, TN 37831, USAcDepartment of Environmental Science, Policy and Management, University of California at Berkeley,151 Hilgard Hall, Berkeley, CA 94720, USAReceived 17 February 2000; received in revised form 13 July 2000; accepted 13 July 2000AbstractAmulti-year,multi-techniquestudywasconductedtomeasureevapotranspirationanditscomponentswithinanuneven-agedmixed deciduous forest in the Southeastern United States. Four different measurement techniques were used, including soilwater budget (1 year), sap flow (2 years), eddy covariance (5 years), and catchment water budget (31 years). Annual estimatesof evapotranspiration were similar for the eddy covariance and catchment water balance techniques, averaging 571 ± 16mm(eddycovariance)and582±28 mm(catchmentwater balance) peryear overa 5-yearperiod.There werequalitative similaritiesbetween sap flow and eddy covariance estimates on a daily basis, and sap flow estimates of transpiration were about 50%of annual evapotranspiration estimated from eddy covariance and catchment studies. Soil evaporation was estimated usinga second eddy covariance system below the canopy, and these measurements suggest that soil evaporation explains only asmall portion of the difference between sap flow estimates of transpiration and eddy covariance and catchment water budgetestimates of evapotranspiration. Convergence of the catchment water balance and eddy covariance methods and moderatelygood energy balance closure suggests that the sap flow estimates could be low, unless evaporation of canopy-intercepted waterwas especially large. The large species diversity and presence of ring-porous trees at our site may explain the difficulty inextrapolating sap flow measurements to the spatial scales representative of the eddy covariance and catchment water balancemethods. Soil water budget estimates were positively correlated with eddy covariance and sap flow measurements, but thedata were highly variable and in error under conditions of severe surface dryness and after rainfall events. © 2001 ElsevierScience B.V. All rights reserved.Keywords: Evapotranspiration; Eddy covariance; Catchment water balance∗Corresponding author. Tel.: +1-865-576-2317;fax: +1-865-576-1237.E-mail address: [email protected] (K.B. Wilson).1. IntroductionEvapotranspiration is an important process acrossa wide range of disciplines, including ecology, hy-drology and meteorology. Because of this multidisci-plinary focus, a number of methodologies have been0168-1923/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved.PII: S0168-1923(00)00199-4154 K.B. Wilson et al./Agricultural and Forest Meteorology 106 (2001) 153–168developed to measure evapotranspiration, or compo-nents of evapotranspiration (transpiration, soil evapo-ration and interception), across a spectrum of spatialscales ranging from individual plants, soil samples andsoil profiles, the atmospheric surface layer, and entirewatersheds. Examples of measurement techniquesinclude soil (Daamen et al., 1993) and plant weigh-ing lysimeters (Edwards, 1986), soil water budgets(Eastham et al., 1988; Jaeger and Kessler, 1997;Cuenca et al., 1997), sap flow (Smith and Allen,1996), plant chambers (Cienciala and Lindroth, 1995),chemical tracing (Calder et al., 1986; Kalma et al.,1998), Bowen ratio (Denmead et al., 1993), eddy co-variance (Baldocchi et al., 1988) and catchment waterbalance (Bosch and Hewlett, 1982; Swift et al., 1988).All of these methods have been used to estimatewater vapor exchange rates between the surface andatmosphere, but the techniques often vary consider-ably in at least three aspects. First, each technique isonly representative within a particular spatial and tem-poral scale, and either interpolation or extrapolationis necessary to infer evaporation rates outside thesescales. The techniques also differ in whether theymeasure evapotranspiration or just one or several of itscomponents. Thirdly, each of the techniques necessar-ily introduces a unique set of particular assumptions,technical difficulties, measurement errors and biases.As a result, specific inherent advantages and limi-tations are introduced for each of the measurementtechniques. Four of the less intrusive techniques thatare addressed in this study, soil water budget, sap flow,eddy covariance (both above and below canopy) andcatchment water balance, illustrate how these advan-tages and limitations differ with technique. A summaryof the approximate measurement scale, the compo-nent of evapotranspiration measured, and some of theTable 1Summary of the methods used to estimate evapotranspiration and its components in this studyaMethod Component Spatial scale (m2) Time scaleSoil water budget Et+ Es100DailySap flow Et102Half-hourEddy covariance (below canopy) Es102Half-hourEddy covariance (above canopy) Et+ Es+ Ei104Half-hourCatchment water budget Et+ Es+ Ei106AnnualaShown are the methods, the component of evapotranspiration measured (Et= transpiration; Es= soil evaporation; Ei= interception),the approximate representative spatial scale of the measurement and the highest meaningful resolution time scale used to estimate evaporation.major advantages and limitations of each of the fourmethods are shown in Table 1and discussed below.A soil water budget is a relatively simple methodfor estimating total water loss from the soil (transpira-tion and soil evaporation). Another advantage of thistechnique is that it can provide insight on the rela-tive contribution of various rooting depths to the totaltranspiration source (Eastham et al., 1988; Teskey andSheriff, 1996). However, closing the soil water balancerequires some estimate of drainage rates (Rutter, 1968;Cuenca et al., 1997), and evapotranspiration estimatesfrom this method do not account for canopy intercep-tion. The measurements are typically representative ofonly a small area, and high spatial variability of soilwater content results in sampling difficulties and prob-lematic extrapolations to larger