CROP PARAMETERS FOR THE EVALUATION OF THE SPECTRAL RESPONSE OF 
COFFEE (Coffea Arabica) AREAS WITHIN THE STATE OF MINAS GERAIS, BRAZIL 
T. G. C. Vieira **, H. M. R. Alves ^, M. P. C. Lacerda *, R. D. Veiga 4. J. C. N. Epiphanio ? 
?* EPAMIG/CTSM, CxP 76, 37200-000, Lavras, MG, Brasil - tatiana@ufla.br 
à Embrapa Café, CxP 176, Epamig/CTSM, 37200-000, Lavras, MG, Brasil - helena@ufla.br 
‘UnB, CxP 04508 - Asa Norte, 70910-970, Brasilia, DF, Brasil - marilusa@unb.br 
^ UFLA, CxP 37, 37200-000, Lavras, MG, Brasil - delly@ufla.br 
?* INPE/DSR, CxP 515, 12227-010, Sào José dos Campos, SP, Brasil - epiphani@dsr.inpe.br 
Session PS, WG VIU2 
KEY WORDS: Remote sensing, Agriculture, GIS, Satellite Images, Crop Mapping, Crop Parameters 
ABSTRACT: 
The work relates coffee plantation characteristics with their spectral responses in TM/Landsat images to obtain identification patterns 
to be used in the mapping and monitoring of coffee lands of the state of Minas Gerais using remote sensing. The fieldwork involved 
selection of representative areas from the main coffee production regions of the state, with definition of pilot areas from where the 
coffee parameters and environmental data were collected. Two pilot-areas representative of the physiographic regions Alto Paranaiba 
and Sul de Minas were selected for the study. The field data and TM/Landsat images were treated with the geographic information 
system SPRING. The reflectance data, as well as the remaining data collected in the field, were organized in a statistical programme 
for correlation studies. The statistical analysis showed that, among the fourteen variables evaluated, the highest correlation was 
observed between reflectance measured in the near infrared zone and the percentage of area covered by the plants canopies. This 
parameter reflects the effects of other crop variables, such as size, diameter, density, vegetative vigour and productivity. The results 
showed that, due to the great variability of the crop and the limitations imposed by TM/Landsat products, the definition of a pattern is 
unlikely. Nevertheless, for productive adult coffee plants in good vegetative state, the survey and monitoring of the crop can be 
carried out using TM/Landsat images, particularly in regions like Alto Paranaiba, where the landscape is mostly of gently undulating 
slopes and the coffee fields are more extensive and homogeneous. 
1l. INTRODUCTION 
Remote sensing can be defined as techniques that enable the 
extraction, at a distance, of information concerning targets of 
interest on the Earth's surface. This is done through detection, 
quantification and analysis of the electromagnetic energy that is 
reflected, absorbed, transmitted and/or emitted by these targets. 
Satellite images are tools that can be used to generate 
information that is obtained through the spectral behavior of the 
vegetation. The spectral differences are registered in the image 
as variations of tones, color or density. The remote sensors 
detect the differences in tone or color between an object and its 
environment (Valério Filho, 1996). 
According to Leonardi (1990), due to its multi-spectral 
character, its repetition in time and space and relative low cost, 
compared to aerial photographs, remote sensing is a potential 
method to obtain information on coffee production. However, 
coffee is a perennial crop with complex features. As observed 
by remote sensors, coffee is very heterogeneous, since it 
presents variations in all the parameters that influence. the 
spectral response (Epiphanio et al., 1994). 
Agricultural crops vary in spectral complexity and work has 
been carried out to evaluate the relation between agricultural 
variables and spectral responses. The spectral response of the 
canopy of a given crop may be influenced by many factors, such 
as: humidity, vegetative vigour, canopy architecture and size, 
soil type, topography, plan density and spacing, variety, age and 
crop consortiation among others. 
In perennial crops, the reflectance registered in a given pixel 
results not only from the plant, but also from factors related to 
the ground, to the plant architecture and shadowing geometry. 
The development of these plants is usually slower and other 
factors, such as soil, constant use of agricultural machines, 
shadowing in and between the plant rows and the crop's 
seasonal characteristics become important. 
Crops planted in rows form a complex field composed by 
vegetation and exposed soil, in different proportions that change 
during the year as the plants grow, which is captured by the 
sensor. In this case, the shadows of the plants in the rows, 
projected on the soil's surface or on other rows, become an 
important factor in the canopy radiation (Ranson e/ a/., 1984). 
Couvre (1980) studied the relation between some parameters of 
citrus plantations (coverage of the terrain by citrus trees, soil 
type, plant rows orientation, slope gradient, slope aspect, height 
of the trees, substratum and uniformity of the field), and the 
spectral reflectance obtained from the transformation of the gray 
levels of the TM/Landsat-5 image. The author explains that the 
incoherent relation between the percentage of the terrain 
covered by the trees and the reflectance in the TM4 band is due, 
mainly, to the increase of shadows and of the covering of the 
ground by the trees, due to the spacing and distribution of the 
plants. According to the author, the increase of shadows annuls 
the effect of the increase of green cover over the reflectance in 
this band. As for the parameter soil type, the author observed 
higher reflectance values in the Dark Red Latosol (LE) than in 
the Red Yellow Latosol (LV) in the TM3, TMS and TM7 bands. 
170 
Internatior 
pl a 
In this wo 
coffee croj 
coffee fiel 
TM/Lands: 
software fr 
the correla 
responses, 
crop in Lar 
Two 520 | 
Patrocinio, 
Paranaiba, 
the physio; 
two of the 
They are 
production 
manageme 
Alto Paran 
informatior 
was import 
areas were 
best repre: 
(Patrocinio 
characteriz 
1100 m, 1 
moderate w 
slopes, pre 
beverages 
region of S 
1260 m, f 
deficiency, 
soils with 
beverages, 
pilot-area c 
278 Km a 
encompassi 
topographic 
1:100.000. 
coordinates 
7.600 Km € 
IBGE (Braz: 
Machado ai 
TM/Landsa 
(Patrocinio) 
April-June/ 
when the fi 
0,69 um) is 
presents gr 
contrast be 
unoccupied 
presents go 
(grazing fi 
mapping th 
rivers’ cour 
(near-infrar 
uniform ve; 
very light 
spectral res, 
band preser 
the mappin; 
as well as 
geology anc 
also very