JAPRS & SIS, Vol.34, Part 7, “Resource and Environmental Monitoring”, Hyderabad, India, 2002
etford
g filter.
. for its
aximum
vere the
y one of
ı for the
‘ocessed
ollected
1e Earth
jues on
he table
la) plus
1a7x7
la7x7
lb).
otatoes
roducer
ccuracy
%
53.8
79.8
79.6
79.7
ng three
Kappa
)0).
cent for
iccuracy
still be
r,....the
producer
ble level
il.
The conclusion from this research is that the most suitable
filters are:
1. Gamma MAP where computing time availability is
short
2. Simulated annealing if computing time is readily
available
2.4 Processing chain
The overall purpose of this paper is to examine the use of radar
data for fraud detection. From the results of the filter evaluation
we can conclude that SAR data can be used as a suitable tool
for operational agricultural field monitoring and so we can
develop a processing chain to provide a field monitoring
system. Not all the elements of this processing chain are
described in this paper, but figure 2 presents the components of
the chain using the SAR data discussed here.
SPOT data
——ÀN.
Maps and
digital data
SAR data Principal ; Maximum Tag EE —
Simulated —> | Components Masking ^ H—*| Likelihood NT E
annealing Analysis Classification Statistics and |:
CHEN ' #
area data
Figure 2. The preferred processing chain for potato field
identification and monitoring.
In sustainable agriculture it is important to have a reliable and
regular data supply, and figure 2 shows how radar data can
used as a key input to the production of maps, digital data and
area statistics for agricultural areas.
3. IRRIGATION MONITORING
3.1 Context
Satellite radar data respond to two environmental components
in a radar scene, roughness and moisture content. The data on
moisture can potentially be used to detect changes in crop and
soil moisture. If such changes are the result of illegal irrigation
then radar data can be used to detect such fraud.
3.2 CHIPS model
The key environmental variable that we require to measure in
connection with irrigation is moisture, that is the moisture of
the soil or the crop (or a combination of the two) in a radar
scene. A new model has been developed to exploit radar
backscatter data for this purpose. The model is termed
Controlled Hydrology and Irrigation Practice with SAR
(CHIPS). CHIPS is a coupled water use and radar backscatter
model that links radar return to irrigation applications through a
series of physically based equations (Graham, 2001). The full
details of CHIPS are given in Graham (2001): this paper
presents some of the key features and table 2 summarises the
main outputs of the CHIPS model.
The CHIPS model was used to simulate soil moisture and crop
canopy moisture for a set of test fields growing potatoes in the
Cambridgeshire Fens area, UK. The period of simulation
covered 183 days from 1 April 1999 to 30 September 1999.
Modelling scale Single field
Number of soil layers Two
Dimensionality 1 dimension
Soil type User defined
Evapotranspiration equation Penman-Monteith
Planting date User defined, Variable
Haulm destruction date User defined, Variable
Emergence date Variable
Maximum canopy height User defined, Variable
Surface resistance calculation MORECS method
Outputs Cumulative PE
Cumulative AE
AE - 2 layers + crop
SMD - 2 layers
Moisture - 2 layers
Moisture - crop
Canopy cover
Canopy height
Radar backscatter
Table 2. Characteristics of the CHIPS model. PE = potential
evapotranspiration, AE = actual evapotranspiration, MORECS
= The Meteorological Office Rainfall and Evaporation
Calculation System. Source: Graham, 2001.
The CHIPS model uses the water cloud model (Attema and
Ulaby 1978) to simulate SAR backscatter for a given field.
Figure 3 plots 1999 ERS-2 SAR backscatter against backscatter
simulated in the CHIPS model for 15 data points from eight
potato fields, mainly for dates where a crop canopy is present.
The coefficient of determination (r) value of 0.81 shows the
good fit of the simulated with the actual backscatter. The
regression line is almost coincident with the 1:1 line suggesting
that the model, and its inherent assumptions, is sufficient to
improve the simulation of radar backscatter values when
compared to traditional methods using in situ data (Graham,
2001).
Actual backscatter (dB)
(gp) 1e3e9syeq pere[nuis SdiHO
Figure 3: Plot of actual versus simulated backscatter (dB)
bounded by the 95% confidence intervals. The 1:1 line is also
shown.
The CHIPS model can be used to estimate soil moisture and
vegetation canopy moisture. The soil moisture estimates are
calculated for two layers in the soil: soil layer 1 corresponds to
0-10 cm depth and soil layer 2 corresponds to 11-70 cm depth
which is the rooting zone for potatoes. Figures 4 and 5 show
