mote
gated
ts, in
cally,
rious
every
mm,
unit,
now
large
op in
lation
es do
ow a
is as
n for
/. The
it. the
mated
rence
Its for
ation.
> 25th
e 11th
spatial
ed by
pleted
This graph clearly shows that CGMS results deviate considerably
from the remote sensing measured values. The simulation results
show much more extreme values. The CGMS simulated crop
transpiration for irrigated wheat (Figure 5B) is also relatively
high. So under irrigated conditions, the potential transpiration rate
as simulated by CGMS do not agree very well with the remote
sensing results. Also note that the difference between irrigated
and rainfed wheat is, according to the remote sensing results, in
fact very small, only 0.5 mm.
At the 11th of May the situation has changed dramatically,
according to CGMS. The crops are now experiencing serious
reductions in transpiration due to water restrictions and for every
simulation unit a reduced transpiration is calculated of 1 to 2 mm,
depending on the soil physical properties of the simulation unit,
the transpiration values for the several simulation units are now
distributed between 0 and 3 mm (Figure 5C), indicating a large
spatial variability.
The remote sensing information does indeed show a drop in
evapotranspiration of 0.5 mm, or | mm when the overestimation
is taken into account. Although the evapotranspiration values do
match better, the remote sensing information does not show a
larger spatial variability. It is thus unlikely that the situation is as
worse as CGMS simulates.
Figure SD shows the distribution of the evapotranspiration for
wheat growing under irrigated conditions at the 11th of May. The
arrow represents the potential transpiration for wheat at the
concerning day. Again CGMS has strongly overestimated
transpiration of irrigated wheat and typically the difference
between irrigated and rainfed wheat is only 0.5 mm.
In general, it can be concluded that CGMS simulation results for
wheat deviate considerably from the remote sensing information.
Crop transpiration has been over- and underestimated at the 25th
of April and, while the absolute values do match better at the 11th
of May, the distributions are not in agreement. The spatial
distribution in transpiration of wheat has been overestimated by
CGMS. This implies that the water balance of CGMS is depleted
to early.
6. CONCLUDING REMARKS
Multi-temporal NOAA/AVHRR satellite data has been applied to
map land cover to be able to perform a crop specific analysis for
agricultural regions in Europe. We developed a strategy to
establish a 1-km land cover data base of Europe in the near future.
Two methods were applied to integrate crop growth modelling
and remote sensing. VI time series were derived from multi-
temporal high and low resolution satellite data. Moreover,
Landsat-TM images were applied to map crop transpiration.
Van Dijk et al. (1991) found that NDVI time series are suitable as
multi-annual monitoring and alarm indicators. An overview of the
integrative use of crop growth modelling and remote sensing is
presented by Genovese (1994). Bouman et al. (1996) showed the
applicability of remote sensing derived parameters to determine
initial crop growth parameters as applied in WOFOST, the crop
growth model used within CGMS.
We found that it is possible to fit simulated LAI profiles with
actual VI profiles derived from HR satellite data trough
recalibration of crop parameters like temperature sum and sowing
date. The study showed that for regions like Andalusia crop
growth monitoring systems should account for differences in
sowing date for irrigated and non-irrigated land. Calibration
separately for irrigated and non-irrigated fields may yield also
differences in other crop parameters, such as cultivar (Van der
Wal and Miicher, 1996).
The agreement between simulations and low resolution satellite
data products are less pronounced. The applicability of NDVI
seems to be questionable, while WDVI shows an agreement with
crop development.
Crop transpiration maps as derived from HR satellite data supply
. detailed spatial information on crop water status. We found that
this information is very helpful to evaluate the water balance
component as applied in CGMS. Further research should focus on
which parameters of CGMS are suitable for calibration and, once
these parameters are selected, how this could be done with remote
sensing derived evapotranspiration maps. Limitations arise from
the fact that the processing of the images is time consuming and
that some field data are needed to get more reliable
evapotranspiration maps. In this respect, field measurements of
atmospheric transmittance or surface albedo, NDVI, roughness
length and net daily outgoing long wave radiation would greatly
reduce uncertainties in the evapotranspiration maps.
Finally it can be concluded that with CGMS quantitative yield
forecasts are only reliable when they are applied to large regions
or countries. An improvement of the forecasting system could be
realised by integrating detailed and synoptic information as
obtained with remote sensing and quantitative information about
seasonal effects on yield as obtained with crop models. Future
satellite systems might offer new opportunities. Especially the
SPOT-4 system could be relevant, as it will deliver high quality
low resolution reflection images on a regular base both in the
visible, near infrared and middle infrared part of the spectrum.
For specific dates in addition HR images are acquired
simultaneously.
ACKNOWLEDGEMENT
This paper results from a study financed by the Space Application
Institute (SAI/JRC, Ispra), where the integration of the satellite
data and agro-ecological crop modelling is investigated to
improve agricultural monitoring and yield forecasting in the
European Union. In this study results are used on remote sensing
applications in agriculture which were financed by the Dutch
National Remote Sensing Programme and the Directorate of
Scientific Knowledge transfer of the Dutch Ministry of
Agriculture, Nature Conservation and Fisheries.
REFERENCES
Bastiaanssen, W.G.M. 1995. Regionalization of surface flux
densities and moisture indicators in composite terrain; A remote
sensing approach under clear skies in Mediterranean climates.
PhD thesis, Wageningen Agricultural University, Wageningen,
the Netherlands, 273 pp.
Bastiaanssen, W.G.M., H. Pelgrum, T. van der Wal and R.A.
Roebeling, 1996. Aggregation of evaporative fraction by remote
sensing from micro to macro scale. Proc. Remote Sensing for
land degradation and desertification monitroing in the
Mediterranean Basin 13-15 June, Valencia. 20 pp. Commision of
European Communities, Luxembourg. In press.
Bouman, B.A.M. 1995. Crop modelling and remote sensing for
yield prediction. Neth. Journal of Agriculture Science, 43; 143-
161.
Bouman, B.A.M., HW.J. van Kasteren and D. Uenk. 1992.
Standard relations to estimate ground cover and LAI of
Intemational Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 149
