Estimation of Evapotranspiration andPhotosynthesis by Assimilation of RemoteSensing Data into SVAT ModelsAlbert Olioso,* Habiba Chauki,* Dominique Courault,*and Jean-Pierre Wigneron*Estimation of evapotranspiration and photosynthesis transfers, and particularly with evapotranspiration andCO2assimilation by vegetation.from remote sensing data frequently use soil–vegetation–atmosphere transfer models (SVAT models). These mod- Remote sensing data may be directly introduced insemiempirical models to compute energy and mass fluxesels compute energy and mass transfers using descriptionsof turbulent, radiative, and water exchanges, as well as at the time of measurements or at the daily scale usingan extrapolation procedure. Such approaches have beena description of stomatal control in relation with watervapor transfers and photosynthesis. Remote sensing data used to estimate sensible and latent heat fluxes from sur-face temperature (e.g., Heilman and Kanemasu, 1976;may provide information that is useful for driving SVATmodels (e.g., surface temperature, surface soil moisture, Jackson et al., 1977; Seguin and Itier, 1983; Kustas, 1990;Inoue et al., 1990, Lagouarde, 1991, Lhomme et al.,canopy structure, solar radiation absorption, or albedo).Forcing or recalibration methods may be employed to 1992; Moran et al., 1994; Norman et al., 1995; Zhan etal., 1996; Chehbouni et al., 1997; Troufleau et al., 1997).combine remote sensing data and SVAT models. In thisarticle a review of SVAT models and remote sensing esti- Some of these approaches are listed in Table 1, and thereader may find a detailed comparison of four recentlymation of energy and mass fluxes is presented. Examplesare given based on our work on two different SVAT mod- published models in Zhan et al. (1996). The simplifiedels. Eventually, some of the difficulties in the combinedrelationship, first derived at field scale by Jackson et al.use of multispectral remote sensing data and SVAT mod-(1977) and later analyzed by Seguin and Itier (1983), hasels are discussed. Elsevier Science Inc., 1999been used for mapping daily evapotranspiration overlarge areas (Lagouarde, 1991; Courault et al., 1994). Inanother wavelength domain, Chanzy et al. (1995) pro-REMOTE SENSING OFposed a semiempirical model, which may be driven byEVAPOTRANSPIRATION ANDsurface soil moisture measurements in the microwavePHOTOSYNTHESISdomain, for estimating bare soil evaporation. These au-thors proposed further to combine this model with theMonitoring energy and mass transfers of soils and vegetalsimplified relationship and thermal infrared data. Thus,canopies is a critical step for water and vegetal resourcesthey were able to specify the parameters of the soil evap-management. It is also useful for a better understandingoration model.of climate and for predicting its evolution. Remote sens-Semiempirical approaches have also been applied foring is an attractive tool to achieve these goals since itestimating net primary production (which represents anprovides information which is related to energy and massintegrator of vegetation photosynthesis). They mostlyrely, in a first step, on the estimation of photosyntheti-* INRA Bioclimatologie, Domaine Saint-Paul, Avignon, Francecally active radiation (PAR) absorption by vegetation can-Address correspondence to Albert Olioso, INRA Bioclimatologie,opies from spectral reflectance measurements. A secondDomaine Saint-Paul, F-84914 Avignon Cedex 9, France. E-mail: oliosostep is to convert this absorbed radiation into [email protected] 13 August 1997; revised 24 November 1998.through production models derived from Monteith (1972)REMOTE SENS. ENVIRON. 68:341–356 (1999)Elsevier Science Inc., 1999 0034-4257/99/$–see front matter655 Avenue of the Americas, New York, NY 10010 PII S0034-4257(98)00121-7342Olioso et al.Table 1. Some Semiempirical Models for Latent Heat Flux, Sensible Heat Flux, and Photosynthesis EstimationaSimplified relationshipSeguin and Itier (1983) LEd5Rnd2A12B1(Ts,14h2Tax)Methods based on an excess resistanceKustas (1990) Hi5qcpTs2Tara1rexLhomme et al. (1992)Moran et al. (1994) LEi5(12A2)Rni2HiApproaches based on a relationshipbetweenradiometric and a so-called “aerody-namic temperature”Troufleau et al. (1997) Hi5qcpTaer2TaraChehbouni et al. (1997) Taer2Ta5(12A3)(Ts2Ta)Two source approachNorman et al. (1995) Hi5qcp1Tv2Tara1Tg2Tara1rc2Tgand Tvmay be estimated from multiangular measurements ofTsor from one single measurement in combination with a Priestley-Taylor based energy balance calculationSoil evaporationChanzy et al. (1995)EdEpd5exp (A4h[025]1B4)C4[11exp (A4h[0–5])]1(12C4)Biomass productionKumar and Monteith (1981) DMS5eb1A51B5qnirqr2RPARaSymbols: A1, A2, A3, A4, A5, B1, B4, B5, and C4: empirical coefficients; cp: specific heat of air; Ed: daily soil evaporation; Epd: daily soil potential evaporation;Hi: instantaneous sensible heat flux; LEi: instantaneous latent heat flux, LEd: daily latent heat flux; RPAR: cumulated incident photosynthetically activeradiation; ra: aerodynamic resistance (above the canopy); rc: aerodynamic resistance at the soil surface; rex: excess resistance; Rni: instantaneous netradiation; Rnd: daily net radiation; Ta: air temperature at some height above the canopy; Taer: “aerodynamic temperature” (here, it corresponds to themean air temperature at some height in the canopy); Tax: daily maximum air temperature; Tv: vegetation surface temperature; Tg: soil surface temperature;Ts: radiometric surface temperature; Ts,14h: radiometric surface temperature at 14 h local solar time; DMS: biomass increase; eb: conversion efficiencyof absorbed radiation into biomass; q: air density; qrand qnir: red and near-infrared relfectances; h[0–5]: 0–5 cm top soil moisture.(cf. Table 1). This method was first proposed by Kumar which are acquired instantaneously. Conversely to sem-iempirical approaches, SVAT models give access to a de-and Monteith (1981), and used at field scale for exampleby Steven et al. (1983), Asrar et al. (1985), Steinmetz et tailed description of soil and vegetation canopy pro-cesses, and not only to a limited number of finalal. (1989). It has also been applied at regional and globalscales from satellite data (e.g., Prince, 1991; Gue´rif et al., variables such as evapotranspiration or net primary pro-duction. They simulate intermediary variables linked to1993; Ruimy et al., 1994; Hanan et al., 1995). In somestudies, the effect of water stress