| 2004
——— —
band 3
tations
le or à
of the
vigour
plants
mation
ground
values
of the
ponse.
etation
1 good
| those
5-40%
In the
portion
er (20-
pectral
niform
bright.
udy of
re the
ugh its
re, the
nation,
aching
r water
values
and the
out for
‘all the
aluated
4. This
set and
s.
> linear
of the
er. The
ne that
| COV.
nature
ith that
> of the
coffee
effects
ur and,
of that
ie most
' coffee
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
fields in remote sensing studies. In this way, the greater the
coverage of the terrain by coffee plants, the greater the spectral
response in band 4. Due to these results, analyses of this
parameter values, placing the data in two different groups for a
better analysis, COV under 50% and COV over 50%, were
carried out. These analyses can be used in the identification and
survey of areas occupied by coffee by TM/Landsat images.
Although the slope gradient of the terrain does not present
significant correlation with the REFB4, it is nonetheless a
variable that interferes in the spectral response by orbital
imaging, as other works in remote sensing show (Strahler ef a/.,
1978; Justice et al, 1981; Stohr and West, 1985). The
topographic effect over reflectance is defined as the variation in
the spectral response of an inclined surface, compared to the
spectral response of a horizontal surface and it is a function of
the orientation of the surface in relation to the source of light
and position of the sensor (Holbern and Justice, 1981).
Therefore, the data was separated into two different groups for
statistical analysis: the first composed by coffee fields with
slope gradient over 15% and the second, composed by coffee
fields with slope gradient under or equal to 15%. Thus the
coffee fields with flat to gently undulating slopes were
separated from those with undulating to steep slopes. The data
presented in Table 2 show that, in the smoother landscapes
(slope gradient < 15%), the spectral response in band 4
(REFB4) was more significant (lower levels of significance)
than that observed in the other types of landscapes (slope
gradient > 15%).
In the declivity < 15% group, other variables besides COV, such
as size, diameter and average production, also presented lower
levels of significance.
4 CONCLUSIONS
The results show that the coffee crop presents a complex
spectral response that is influenced by the many variables that
influence its characterization.
Statistical analysis showed that, among the variables studied,
COV, which corresponds to the percentage of the terrain
occupied by coffee plants, presented the best reflectance results
in band 4. COV reflects the effects of other coffee agricultural
variables, such as size, diameter, plant density, vegetative
vigour and average production. Therefore, the use of
TM/Landat satellite images in coffee fields, in which the ground
Zz
covered by the plants is over 50%, is recommended.
Relief also has great influence in the coffee’s spectral response.
Therefore, it is recommended that remote sensing be used,
preferably coffee is cropped over flat to gently undulating
landscapes.
In formed coffee fields in good productive conditions, the
spectral response allows their survey and monitoring by
TM/Landsat images, especially in areas favorable to orbital
imaging, that is: regions with smoother slopes, ideal
atmospheric conditions and large, contiguous areas occupied by
coffee fields. These conditions occur in Patrocínio. In the other
regions, it is advisable to associate more precision remote
sensing products of satellites with better spatial resolution.
S. REFERENCES
Covre, M., 1989. /Influéncia de parámetros culturais de citrus
sobre os dados TM/Landsat. INPE, Sao José dos Campos, 1989.
241p. (INPE -4856- TDL/367).
Epiphanio, J. C. N.; Leonardi, L.; Formaggio, A. R., 1994.
Relacgóes entre parámetros culturais e resposta espectral de
cafezais. Pesq. Agropec.Bbras., 29 (3), pp. 439-447.
Leonardi, L., 1990. /nfluéncia de parámetros culturais de
cafezais sobre os dados TM/Landsat-5. INPE, Säo José dos
Campos, 141p. (Dissertaçäo de Mestrado).
Ranson, K. J.; Biehl, L. L.; Bauer, M. E., 1984. Variation in
spectral response of soybeans with respect to illumination,
view, and canopy geometry. West Lafayette, IN. LARS, 27p.
(LARS TR-073184).
SAS INSTITUTE, 1999. SAS/STAT®: user's guide: North
Carolina.
Valerio Filho, M.; Pinto, S.A.F., 1996. Imagens orbitais
aplicadas ao levantamento de dados do meio fisico:
Contribuigdo ao planejamento de microbacias hidrograficas. In:
Congresso Brasileiro e Encontro Nacional de Pesquisa sobre
Conservacdo do Solo, 8, Londrina, Paraná. Anais... pp. 77-94.
m Statistical Crop variables
| Parameter REFB4 SIZE DENS VIG DIAM PROD COV SLO
Average value 30.05 2.44 4305 8.10 1.99 3.64 59.87 10.40
Standard deviation 4.68 0.99 3140 1.18 0.99 3.25 28.34 11.10
Minimum value 16.86 0.80 1000 5.00 0.28 0.00 10.00 1.00
Maximum value 43.14 5.00 13333 10.0 5.00 12.0 100.0 70.00
Table 1 - Average, strandard deviation, minimum and maximum values of the data set collected in the 75 coffee fields surveyed.
Where:
REFB4: average reflectance values in band 4.
SIZE: average height of coffee plants in a field in meters;
DENS: Plant Density - number of plants per hectare;
VIG: Vegetative vigour - indices from 1 to 10, according to field survey;
DIAM: Diameter - average diameter of plants in a coffee field in meters;
PROD: Production - average plant coffee production in liters of coffee berries per plant;
COV: Ground cover - percentage of the ground covered by plants canopies in a coffee field;
SLO: Slope gradient - percentage of slope gradient of a coffee field.
