Agricultural and Forest Meteorology 106 2001 215–231
Modification of DAISY SVAT model for potential use of remotely sensed data
Peter van der Keur
a,∗
, Søren Hansen
a
, Kirsten Schelde
b
, Anton Thomsen
b
a
Department of Agricultural Sciences, The Royal Veterinary and Agricultural University, Laboratory for Agrohydrology and Bioclimatology, Agrovej 10, DK-2630 Taastrup, Denmark
b
Department of Crop Physiology and Soil Sciences, Danish Institute of Agricultural Sciences, Research Centre Foulum, P.O. Box 50, DK-8830 Tjele, Denmark
Received 30 April 1999; received in revised form 7 August 2000; accepted 19 August 2000
Abstract
The SVAT model DAISY is modified to be able to utilize remote sensing RS data in order to improve prediction of evapotranspiration and photosynthesis at plot scale. The link between RS data and the DAISY model is the development
of the minimum, unstressed, canopy resistance r
min c
during the growing season. Energy balance processes are simulated by applying resistance networks and a two-source model. Modeled data is validated against measurements performed for a winter
wheat plot. Soil water content is measured by time domain reflectometry. Crop dry matter content and leaf area index are modeled adequately. Modeled soil water content, based on a Brooks and Corey [Brooks, R.H., Corey, A.T., 1964. Hydraulic
properties of porous media. Hydrology Paper no. 3, Colorado University, Fort Collins, CO, 27 pp.] parameterization, from 0 to 20, 0 to 50 and 0 to 100 cm is calibrated satisfactorily against measured TDR values. Simulated and observed energy fluxes are
generally in good agreement when water supply in the root zone is not limiting. With decreasing soil moisture content during a longer drought period, modeled latent heat flux is lower than observed, which calls for both improved parameterizations for
environmental controls and for a improved estimation of the r
min c
parameter. © 2001 Elsevier Science B.V. All rights reserved.
Keywords: Crop energy balance; Remote sensing; Minimum canopy resistance; DAISY model
1. Introduction
Spatially distributed information on land surface characteristics can be retrieved by means of remote
sensing from satellite or other platforms and has been used extensively in land use mapping, e.g. crop man-
agement, and subsequently stored in Geographical Information Systems. Another potentially powerful
application of remote sensing data RS data is pro-
∗
Corresponding author. Tel.: +45-3528-3560ext. 3544; fax: +45-3528-3384.
E-mail addresses: pvdkruc.dk P. van der Keur, shakvl.dk S. Hansen.
viding the link between measuring spatially varying biophysical properties and hydrological modeling
Tenhunen et al., 1999; Waring and Running, 1999. Soil moisture and vegetation development, usually
highly variable in both time and space and very difficult to quantify at larger scales, exert strong con-
trol on the surface energy balance and hydrological processes. These facts make the use of RS data in
modeling such processes very attractive. Models at- tempting to address landscape level processes need
to be deliberately designed to use remotely sensed variables Wessman et al., 1999.
Historically, basically three approaches have been adopted for coupling evapotranspiration to remote
0168-192301 – see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 8 - 1 9 2 3 0 0 0 0 2 1 2 - 4
216 P. van der Keur et al. Agricultural and Forest Meteorology 106 2001 215–231
sensing data see, e.g. Ottlé et al. 1996 for a review. The first method is based on conversion of observed
thermal radiance into surface temperature, which is then used to calculate sensible heat flux. The latent
heat flux is then calculated from the surface energy balance as a residual of the net radiation, estimated
ground heat flux and the RS estimated sensible heat flux Hatfield, 1983; Moran et al., 1989; Kustas, 1990.
The second method relies on estimation of surface energy fluxes from either remotely sensed vegetation
index data and surface radiant temperature Tucker et al., 1981; Gillies et al., 1994 or from surface bright-
ness temperatures as measured with a microwave ra- diometer Njoku and Patel, 1986. Surface brightness
temperature can be used to infer the soil moisture content of the upper few centimeters of the soil profile
e.g. Wang et al., 1989 and used as a boundary condi- tion for the calculation of the surface evaporation rate
where soil evaporation is the dominant component of the latent heat flux Sellers, 1991. Alternatively, ac-
tive microwave RS radar can be used to infer top soil moisture content or in combination with passive mi-
crowave RS Chauhan, 1997. The third method, and the one pursued in this study, relies on the ability to
infer information on the photosynthetic capacity and the minimum canopy resistance r
min c
from spectral vegetation indices e.g. Asrar et al., 1984; Monteith,
1977; Sellers, 1985, 1987; Sellers et al., 1992a,b. Specifying the correct change in minimum canopy re-
sistance with time is crucial and incorporates changes in both leaf area index and stomatal resistances Dol-
man, 1993. This link is here taken as the point of departure for the use of remotely sensed data in mod-
eling evapotranspiration processes in soil–vegetation– atmosphere–transport schemes SVATS models, at
various spatial scales. No direct means is yet available to monitor minimum stomatal resistance from space,
but subtle shifts in the reflectance spectrum in visible wavelengths that relate to diurnal changes in photo-
synthetic efficiency also mirror changes in stomatal resistance Gamon et al., 1992. In this study, however,
focus is on the unstressed stomatal resistance r
min c
, i.e. minimum canopy resistance, that is upscaled through
LAI. It can be inferred by RS data, and therefore in- herently contains information on plant physiological
status through r
min c
and LAI. Sellers 1991 summa- rizes the limitations of all three approaches to convert
satellite sensed data to the desirable surface param- eters including problems with sensor calibration,
atmosphericgeometric correction, conversion of radi- ance to surface parameters and finally conversion of
surface parameters to biophysical quantities.
The soil–plant–atmosphere system model DAISY Hansen et al., 1991 was prepared to accommodate
use of remotely sensed, initially ground based, data for simulation of evapotranspiration. In the present
approach, simulated actual evapotranspiration was ei- ther at potential rate and estimated empirically from
standard meteorological data e.g. Makkink, 1957 or less than potential rate being controlled by the extrac-
tion of soil water by plant roots Hansen et al., 1991. This method precluded the incorporation of remotely
sensed data in the model in the sense proposed in this paper. Instead, an energy balance approach based
on a two-source resistance network, allowing sparse canopy cover, is added to the model. Stomatal resis-
tance is part of this resistance network and regulates the amount of water available through stomata path-
ways for plant transpiration and intake of carbon diox- ide for photosynthesis. Thus, in summary, unstressed
stomata resistance scaled to the canopy level by LAI can be related to both RS data, i.e. spectral vegetation
indices, and actual canopy resistance and has the po- tential to provide a link between SVAT modeling and
RS data.
The purpose of this paper is to describe the method followed to prepare the DAISY model for RS data in-
put as envisaged within the framework of the Danish funded RS-MODELearth observation project. The
two-source model, allowing for sparse canopy cover Shuttleworth and Wallace, 1985; Shuttleworth and
Gurney, 1990, is added to the DAISY model struc- ture and modeled surface energy fluxes, soil moisture
content and crop development are validated against experimental data from a winter wheat plot under
Danish conditions.
2. Model concepts