493 
Figure 8. Ratio of cumulative actual to potential 24 
hour évapotranspiration over the entire growing season 
of 1982 (ELE 24 /ELE 24 ) for grass and maize on Typic 
Haplaquod soil with drainage class V depending on 
groundwater table drawdown as calculated with the 
SWATRE-model (after Thunnissen 1984b) 
Water-tabel drawdown (cm) 
Figure 9. Relative 24 hour évapotranspiration rate 
(LE^^/LEp 4 ) on 30 July 1982 as calculated with the 
SWATRE-model. A. grass on Typic Haplaquod soil with 
drainage class V; B. for maize on the same soil with 
drainage class VI depending on groundwater table 
drawdown (after Thunnissen 1984b) 
water table is present under natural conditions (Fig 
ure 9A). For a relatively deep groundwater table crop 
évapotranspiration of maize is, even under natural 
conditions, low on 30 July 1982 and the effect of a 
lowering of the groundwater table is then limited 
(Figure 9B). 
The results obtained with the remote sensing ap 
proach are confirmed by the SWATRE-model calculations. 
According to the remote sensing approach crop évapo 
transpiration strongly decreases on 30 July 1982 in 
the direction of the centre of the extraction in case 
of shallow groundwater tables under natural conditions. 
This is shown in Figure 7A. Figure 9A shows that also 
according to SWATRE-model calculations for the con 
cerning conditions crop évapotranspiration strongly 
decreases with lowering of the groundwater table. 
Also for relatively deep groundwater tables under 
natural conditions both results show good agreement 
(Figures 7B and 9B). If no suppletion from groundwater 
occurs crop évapotranspiration is very low on 30 July 
1982 and a lowering of the groundwater table has no 
perceptible influence on crop évapotranspiration. 
It can be concluded that with remote sensing de 
tailed information about the regional distribution 
of évapotranspiration on flight days is obtained. 
With SWATRE-model calculations the occurrence of 
drought damage can be explained. Thunnissen (1984b) 
concluded that an important improvement of the hy 
drological description of an area can be achieved by 
integrating SWATRE-model calculations with remote 
sensing. 
Information about the regional distribution of 
évapotranspiration can also be obtained with regional 
agrohydro logical simulation models like GELGAM (De 
Laat and Awater 1978). The applicability of such a 
model depends on the schematization. The size of the 
grid cell applied in the GELGAM-model has been in 
dicated in Figure 6. 
It can be seen that within grid cells large varia 
tions in évapotranspiration can occur. With model 
calculation for each grid cell separately mean val 
ues are obtained while with remote sensing informa 
tion is obtained about the deviations which may oc 
cur. In this way insight is obtained in the meaning 
of results obtained with a regional hydrological 
simulation model. 
As in the Netherlands we have to deal with inten 
sively used agricultural land an integrative ap 
proach based on the application of remote sensing 
and SWATRE-model calculations is preferred in water 
management above the application of a regional agro- 
hydrological simulation model as GELGAM. 
4 CONCLUDING REMARKS 
With multispectral scanning remote sensing detailed 
information can be obtained about land use and crop 
water supply. A method has been developed for the 
mapping of évapotranspiration from digitally taken 
reflection and thermal images. The accuracy of an 
évapotranspiration map obtained with remote sensing 
is strongly dependent on the availability of an ac 
curate crop map. The possibilities to distinguish 
different crops is dependent on the growth stage. 
This being the case a multi-temporal approach has to 
be applied. 
With some examples the applicability of remote 
sensing techniques to determine effects of certain 
water management measures has been demonstrated. For 
a unique interpretation of remote sensing images 
additional information is, however, indispensable. 
The main advantage of remote sensing is, that the 
actual crop water supply is registrated for large 
areas. With models crop water supply can be simulat 
ed for certain locations during the entire growing 
season, while verification of the obtained results 
on a regional scale is nearly impossible with con 
ventional methods. Therefore an integrative approach 
based on the application of remote sensing tech 
niques and agrohydrological model simulations is 
propagated. 
The possibilities of remote sensing techniques to 
detect drought damage has been extensively investi 
gated. In the near future damage caused by excess of 
water has to be emphasized as in humid areas from 
economical point of view this is an important prob 
lem. 
The application of MSS-techniques in practice will 
mainly depend on the availability of useful and pay 
able images. The satellites SPOT and LANDSAT 5 pro 
vide reflection images with a spatial resolution 
useful for different small plots (10-30 m). The 
spatial resolution of thermal images obtained with 
the TM-scanner on board LANDSAT 5, however, is 120 
m. Moreover, especially for humid areas the temporal 
resolution is questionable. The frequency is 16 days 
and the time of acquisition (about 9.30 AM local 
time) is far from optimal to detect drought damage. 
The increase in crop temperature because of a reduc 
tion in évapotranspiration is maximal in the early 
arteiuuon. Before 11.Ou AM une temperature dirter- 
ences caused by a lack of water in the root zone will 
be small. For operational applications it is neces 
sary that digital multi-spectral and thermal images 
are recorded with a frequency of at least once in 
every 5 days around midday with a spatial resolution 
of about 15-20 m. A space-borne system with these