MONITORING THE RELATIONSHIPS BETWEEN ENVIRONMENT AND COFFEE PRODUCTION IN AGROECOSYSTEMS OF THE STATE OF MINAS GERAIS IN BRAZIL

ISPRS Technical Commission II Symposium, Vienna, 12 – 14 July 2006





131

MONITORING THE RELATIONSHIPS BETWEEN ENVIRONMENT AND
COFFEE PRODUCTION IN AGROECOSYSTEMS OF THE STATE OF
MINAS GERAIS IN BRAZIL


H.M.R. Alvesa, T.G.C.Vieirab, V.C.O. Souzab, M.A. Bertoldob, M.P.C. Lacerdac, H. Andraded, N. Bernardesd

a* EMBRAPA Café, GeoSolos, CxP 176, - Lavras, MG, 37200-000, Brasil - helena@ufla.br
b EPAMIG/CTSM, GeoSolos, CxP 176, Lavras, MG, 37200-000, Brasil - {tatiana, vanessa, matilde}@epamig.ufla.br
c FAVE/UnB, Campus Darcy Ribeiro, ICC Sul, 90910-970. CxP 4508, Brasília, DF, Brasil - marilusa@unb.br
d UFLA, Departamento de Ciência do Solo, CxP 3037, Lavras, MG, 37200-000, Brasil - {handrade, geosolos}@ufla.br

Technical Commission II


KEY WORDS: Coffee Environment, Remote Sensing, Land Use, Surveying, Geoprocessing, SPRING


ABSTRACT:

Satellite imagery has been acknowledged as the most promising way for detailed mapping and monitoring of land agricultural use,
although there are challenges that must be met in order to fulfil its potential. Coffee is one of Brazil’s most important cash crops and
the state of Minas Gerais responds for approximately half of the national production. Nevertheless, in spite of its importance, updated
detailed information about the areas occupied by the crop is scarce. In this work data obtained from secondary information,
interpretation of Landsat images of the years 2000 and 2003 and modelling, complemented by field surveys, were used to
characterize coffee lands of Minas Gerais and evaluate the relationships between the crop and the environment. Two of the most
important production regions where chosen, Sul de Minas and Alto Paranaíba, and 520 km2 representative study areas were selected.
Models relating geology, geomorphology and-pedology were used to map the main soil classes. The relationships between
environment and coffee production in the selected areas were assessed and quantified. The results showed the main soil, slope and
altitude classes in which the crop is cultivated, emphasizing the relations between management practices and limitations imposed by
the environment. The use of Landsat images in the mapping of coffee areas constituted an advance over the traditional
methodologies, especially for the gently undulating landscape of the western region of the state, where the coffee fields are more
extensive and homogeneous. The information obtained can subsidize better regional land use planning.


1. INTRODUCTION
Remote sensing is the combination of processes and techniques
used to measure a surface or object’s electromagnetic

properties, without physical contact between the object and the
When mankind explores natural resources, introducing in them
sensor equipment. In other words, it is the technology that
changes to suite its needs, the environmental impact is usually
gathers images and other types of data from earthly surfaces by
negative. An understanding of these changes is necessary to
registering the energy reflected or emitted by them (Moreira,
prevent the degradation of natural resources and should form
2001). It is an important relative of GIS and undoubtedly, the
the basis for any sustainable agricultural activity. Land use
largest primary source of digital data for use in GIS is created
planning based on sound studies of the physical environment
by remote sensing technology on board satellites and other
and its evolution dynamics becomes essential. These studies
aircraft. In fact, satellite imagery has been acknowledged as the
should be fundamental to a region’s development planning in
most promising way for detailed mapping and monitoring of
order to reduce socio-economic losses and make it a sustainable
land agricultural use, although there are challenges that must be
process through time (Assad et al., 1998).
met in order to fulfil its potential. It is therefore essential for

environmental applications that remote sensing and GIS be
Remote sensing, integrated to the Geographic Information
closely linked (Jones, 1997; Carvalho, 2001).
Systems (GIS), are useful tools in research applied to land use

planning. GIS are computerized systems used to store, analyze
The state of Minas Gerais has very different environments in
and manipulate geographic data, i.e. data that represent objects
terms of relief, geology, soils and climate. Detailed mapping in
and phenomena whose geographic location is a characteristic
the state is difficult due to its vast extension, the complexity of
inherent to their formation and therefore indispensable to their
the physical environment and the accelerated dynamics of its
analysis. They constitute a modern way of storing and
land use. Inserted in this context, coffee cropping is socially and
manipulating information of the physical environment, economically important to the state and the whole country.
integrating data from heterogeneous sources such as soils,
However, despite its importance, the coffee crop lacks precise
geology, topography, land use and land cover, which are
quantitative data, relating to the extension and distribution of
organized in spatially referenced digital databases in a the fields as well as to the characteristics of the environments in
transparent way (Câmara Neto et al., 1996). With this
which the crop is cultivated (MINAS GERAIS, 1995). The use
technology, spatial information for land evaluation and other
of Landsat images in the mapping of coffee areas, although
environmental studies can be generated and analyzed more
presenting problems to be solved, constitutes an advance over
efficiently.
traditional methodologies.


NON-REVIEWED PAPER

---

132
International Archives of Photogrammetry, Remote Sensing, and Spatial Information Sciences Vol. XXXVI - Part 2

This work is part of a project to characterize coffee types obtained during the field survey were used to classify the
agroecosystems of the state of Minas Gerais using images. The segmentation of the images was performed using a
geotechnologies to evaluate the relationships between the
region growing method and a supervised classification was
physical environment and local coffee production systems. The
carried out using the Maxver classifier (maximum likelihood
objective, in this case, was to use remote sensing and GIS
algorithm available in the SPRING) on band 4. This
technology to map and monitor the evolution of coffee areas in
classification was corrected by photointerpretation, checked in
relation to the environment in the regions of Patrocínio (within
the field and by comparison with the Ikonos image and edited to
the physiographic region of Alto do Paranaíba), São Sebastião
obtain the final map of coffee lands.
do Paraíso and Machado (both in the physiographic region of

Sul de Minas), between the years 2000 and 2003.
The maps generated and stored in the database were overlayed

through the computer language LEGAL (SPRING`s Spatial

Language for Algebric Processing). The relations between
2. MATERIAL METHODS
environment and the coffee crop for the years 2000 and 2003
were assessed and quantified. The results obtained for the two
The study areas were selected within two of the main coffee
years were compared and the changes observed in these
producing regions of the state of Minas Gerais. Three
relationships during this period were then evaluated.
representative study areas, in terms of characteristics of the crop

and its relation to the physical environment, were chosen, each

one with 520 km2: Machado, representative of the southern
3. RESULTS AND DISCUSSION
region of the state with hilly relief; São Sebastião do Paraíso,
also within the southern region of the state but with a different
Due to the alterations observed in the areas occupied by coffee,
environment, smother relief and soils rich in iron; and
a study of the dynamics of the crop in relation to the
Patrocínio, representative of the Alto Paranaíba region, with
environment was necessary. According to the results obtained
smoother landscapes within the cerrado (Brazilian savanna)
by overlaying the land use and environment maps, it was
environment.
observed that, in São Sebastião do Paraíso, the coffee areas

changed most in relation to the soils classes on which they are
To characterize each area all the secondary information
cultivated. In Patrocínio, the change occurred in relation to the
available on the natural resources and characteristics of the
classes of slope aspect and in Machado, in relation to slope
coffee crop of each region was gathered. Field surveys were
gradient classes.
carried out in each area to assess the relationships between

coffee and the environment and to collect referenced data from
The analysis carried out through geoprocessing techniques, the
coffee fields, soils and geomorphology in order to establish
generation and manipulation of thematic maps of distribution of
ground truth samples. Topographic maps from the Brazilian
natural resources, especially geology and geomorpholgy, allied
Geography and Statistic Institute (IBGE), scale 1:50,000, were
to the field observations, lead to a better understanding of soil
used as cartographic basis. For Machado, the sheets Machado
distribution in São Sebastião do Paraíso. The analysis also
and Campestre (UTM coordinates E 392 000 to 418 000 m and
established a correlation model between relief and geology,
N 7620 000 to 7600 000 m). For São Sebastião do Paraíso
which allowed the mapping of the main soil units. The soils
sheets São Sebastião do Paraíso and São Tomás de Aquino
map of the study area of São Sebastião do Paraíso was obtained
(UTM coordinates E 274 000 to 300 000 and N 7700 000 to
using the LEGAL program, through the overlaying of thematic
7680 000 m) were used. For the region of Patrocínio the maps
maps of slope classes and geologic domains, as presented in
of the Army Ministry, scale 1:100,000, sheets Patos de Minas
Table 1. This correlation model was developed and field
and Monte Carmelo (UTM coordinates E 278 000 to 304 000m
checked in the Ribeirão Fundo watershed, which is an
and N 7942 000 to 7922 000m) were used. To map land use the
environmental unit representative of the model of soil
TM-Landsat 7 images, from 2000 and 2003, corresponding to
distribution in the regional landscape. The soil profiles,
the regions of Machado (219 /75), São Sebastião do Paraíso
representatives of the main mapping units, were described and
(220/74) and Patrocínio (220/73), as well as the IKONOS Image
sampled for chemical and physical analysis according to the
of 11/08/2002 of São Sebastião do Paraíso, were used.
methodology suggested by EMBRAPA (1999) and Lemos

(1996). The main soil classes were defined and mapped
Through secondary information, field surveys and satellite
according to the Brazilian System of Soil Classification
image interpretation, data on soils, relief, hydrologic resources
(EMBRAPA, 1999). This model was applied to the study area
and land use were generated with emphasis on the coffee crop.
and in this way, by generalization, the map of soil classes of São
This information was incorporated through the geographic
Sebastião do Paraíso was created.
information system SPRING (SPRING, 2003) to generate a

digital database for each study area. From this database thematic
Analysing the distribution of coffee lands in relation to the soil
environment characterization maps were generated, among
map it was observed that, in the year 2000, 38% of the coffee
which are maps of land use, slope gradient, slope aspect and soil
fields were located on Ferric Red Latosols (LVf) and, in 2003,
classes. The soils maps were generated using geologic-
this area decreased to 32%. The coffee planted on Ferric Red
geomorphic-pedologic models (according to Andrade et al.,
Nitosols remained practically in the same proportion, 22% in
1998) and the relief maps were based on each region’s contours
2000 and 21% in 2003. In the Red Yellow Latosols there was an
and DTM obtained from the digitized topographic maps.
increase of 3%, as the coffee area increased from 33% in 2000

to 36% in 2003 (Figures 1 and 2).
The areas occupied by coffee were mapped using the Landsat

images treated in the SPRING’s IMAGE module. Only three
In Patrocínio the digitized contours and elevation points of the
spectral bands were used for the classification of the image, viz.
topographic maps were used to model and sample the terrain,
band 3 (red), band 4 (near infrared) and band 5 (mid infrared),
generating the digital terrain model (DTM) and to generate a
since these bands represent more than 80% of the spectral
thematic map of slope aspect. The slope aspect map was divided
information. Controlled samples of the main land cover/use
in the directions 1° to 45° (N-NE), 45° to 90° (NE-E), 90° to 135°
NON-REVIEWED PAPER

---

ISPRS Technical Commission II Symposium, Vienna, 12 – 14 July 2006





133

(E-SE), 135° to 180° (SE-S), 180° to 225° (S-SW), 225° to 270°
(SW-W), 270° to 315° (W-NW), 315° to 360° (NW-N) and flat
areas that have no angle inclination.

Coffee/LVf
Geological
38%
Slope Class
Soil Mapping Unit
22%
Coffee/NVf
Domain*
Association of Haplic
Coffee/LVAp
Qa
Gleisol (GX) + Fluvic
Coffee/PVA
Neosol (RU)**
22%
KJsg
Ferric Red Latosol (LVf)
Coffee/PVAa
4% 3%
11%
Association of Red-
Coffee/LVA
0-12%
Yellow Latosol (LVA) +
TQi, Kb, KJb
Sandy-Loamy Red-
Yellow Latosol (LVAp)

Association of Red-
Figure 1. Distribution of coffee lands of São Sebastião do
PCi
Yellow Latosol (LVA) +
Paraíso by soil class in the year 2000
Red Latosol (LV)

Association of Ferric Red
KJsg
Nitosol (NVf) + Haplic
Cambisol (CX)
Association of Red-
34%
Cofee/LVf
Yellow Argisol (PVA) +
20-45%
TQi, Kb, KJb
Sandy Red-Yellow
Cofee/NVf
Argisol (PVAa) + Haplic
Cof ee/LVAp
Cambisol (CX)
Association of Red-
25%
Coffe/PVA
PCi
Yellow Argisol (PVA) +
Cof e/PVAa
Red Argisol (PV)
4%
5%
11%
21%
Association of Haplic
Coffe/LVA
KJsg, TQi, Kb,
>45%
Cambisol (CX) + Litholic
KJsg, KJb, PCi
Neosol (RL)
Table
1.
Correlation model between slope gradient, Figure 2. Distribution of coffee lands of São Sebastião do
geological domain and soil classes of the Ribeirão
Paraíso by soil class in the year 2003
Fundo watershed used to map the soils of the pilot

area of São Sebstião do Paraíso

* Geological domains obtained from DNPM/CPRM (1979),
Coffee/N-NE
where:
37%
Coffee/NE-E
Qa: Quaternary sedimentary deposits, mainly aluvial.
Coffee/E-SE
TQi: Undifferentiated covers, involving Latosols with
9%
8%
Coffee/SE-S
paleopavings.
Coffee/S-SW
Kb: Bauru Formation – sandstones with medium granules,
clayey, reddish pink and whitish to reddish, quartzons, locally
11%
9%
Coffee/SW-W
with coarse sandstones, with crossed and plain stratification
6%
10%
6%
4%
Coffee/W-NW
having small to medium portment.
Coffee/NW-N
KJsg: São Bento Group - Serra Geral Formation – basaltic
Coffee/Level areas
flows with sandier lenses and layers (Botucatu sandstone type).
KJb: São Bento Group - Botucatu Formation – sandstones with

fine to medium granules, well selected, whitish to reddish,
Figure 3. Distribution of coffee lands of Patrocínio by slope
quartzons; locally with coarse sandstones beds, with tangential,
aspect in the year 2000
crossed stratification, having large portment at the base.

Pci: Tubarão Super Group – Undifferentiated Itararé Group –
was planted in the SE-S surfaces and 4% in the W-NW,
coarse to fine sandstones, yellowish to reddish color, with
showing that there was an inversion of the localization of coffee
development linked to red-brick color diamictites, grading to
areas in relation to the slope aspect (Figures 3 and 4). This has
sandier and silty pellitic material; present crossed and plain
implications in the physiology of the plants and also in the use
stratification having small to medium portment.
of satellite images to map and monitor the crop.
** This soil class was mapped by photo interpretation.


In Machado, the slope gradient map was generated using

SPRING modules, from the contours digitized from the
Overlaying the coffee lands map of the two years onto the slope
topographic maps of IBGE. Five slope classes were defined and
aspect thematic map it was observed that in the region of
related to various types of relief and soil classes according to
Patrocínio the crop is cultivated mainly on the flat surfaces.
Table 2: 0-3% – flat or level areas; 3-12% – Gently undulated
However, in 2000, 4% of the coffee was planted in the SE-S
surfaces; 12-24% – Rolling to Strongly undulated; 20-45% –
slopes (Southwest-South) and 11% was planted in the W-NW
Hilly to Steeply dissected surfaces; and >45% – Mountainous
slopes (West-Northeast). In 2003 however, 10% of the coffee
relief.

NON-REVIEWED PAPER

---

134
International Archives of Photogrammetry, Remote Sensing, and Spatial Information Sciences Vol. XXXVI - Part 2

smother areas to annual crops. This tendency must be better
Cof ee/N-NE
investigated through time and taking into account other socio-
38%
10%
Cof ee/NE-E
economic factors.

6%
Cof ee/E-SE
Cof ee/SE-S
Coffee/Level
4%
Cof ee/S-SW
7%
7%
10%
8%
10%
Cof ee/SW-W
16%
2%
16%
Coffee/Gently
Cof ee/W-NW
undulating
Cof ee/NW-N
Coffee/Rolling
Cof ee/Level areas
36%
Coffee/Hilly

Figure 4. Distribution of coffee lands of Patrocínio by slope
30%
aspect in the year 2003
Coffee/Mountainous

Slope Gradient (%) Slope Classes
Soil Class
0 – 3
Flat
Latosols

Figure 6. Distribution of coffee lands of Machado by relief
Gently
3 – 12
Latosols
classes in the year 2003
undulated

Rolling to
Argisols and

12 – 24
Strongly
Nitosols
4. CONCLUSIONS
undulated
Hilly to
The results showed that using the methodology applied in the
Argisols, Nitosols
24 – 45
Steeply
work it was possible to observe the distribution and the
and Cambisols
dissected
dynamics of coffee occupation in relation to the surrounding
Cambisols and
environment, as well as to monitoring these changes, providing
> 45
Mountainous
Neosols
relevant information for land use planning. The behavior of the
crop was different in each region studied and the changes are
Table 2. Correlation between slope gradient, relief type and
probably closely related to the local production systems
soil class used to model the landscape of Machado
characteristics.


Also using SPRING’s computer spatial language processing
Remote sensing and GIS were useful in the characterization and
LEGAL, the land use maps of the years 2000 and 2003 were
mapping of the coffee agroecosystems of the selected regions,
overlayed with the environment maps. Through these proving to be an efficient methodology in terms of both speed
overlayings the evolution of the coffee crop’s distribution in the
and resources. Through the geotechnologies used, spaced
study area and the changes in relation to its occupation over the
information on the coffee crop of each study area was generated
landscape were quantitatively evaluated.
faster, with lower costs and higher precision, subsidizing

rational management and planning of the crop in these regions.
In Machado it was observed that, in 2000, 34% of the coffee

was planted on gently undulating relief and 13% on hilly to

steep relief (Figure 5).
BIBLIOGRAPHY

Andrade, H.; Alves, H.M.R.; Vieira, T.G.C.; Resende, R.J.T. P.;
Esteves, D.R; Brasil, J.K. & Rosa, E.R., 1998. Diagnóstico
Coffee/Level
ambiental do município de Lavras com base em dados
13%
1%
16%
georreferenciados do meio físico: IV – Principais grupamentos
Coffee/Gently
de solos. In: Congresso Brasileiro de Engenharia Agrícola,
undulating
Poços de Caldas, Brasil, Vol. IV, pp. 442-443.
Coffee/Rolling
Assad M.L.L., Hamada. E., Cavaueiri, A., 1998. Sistema de
Coffee/Hilly
informações geográficas na avaliação de terras para
agricultura. In: Assad, E. D. & Sano, E. E. 2a ed. Sistema de
34%
36%
Coffee/Mountainous
Informações geográficas - Aplicações na agricultura. Brasília,
Embrapa-CPAC, pp. 191-232.
Câmara Neto, G., Casanova, M.A., Hemerly, A.S., Magalhães,

G.C., Medeiros, C.M.B., 1996. Anatomia dos Sistemas de
Figure 5. Distribution of coffee lands of Machado by relief
Informações Geográficas. UNICAMP: IV Escola de
classes in the year 2000
Computação, Campinas, 193 p.

In 2003, the area of coffee planted on gently undulating relief
Carvalho, L.M.T., 2001. Mapping and monitoring forest
decreased 4% and the area planted on hilly to steep relief
remnants: a multiscale analysis of spatio-temporal data.
increased 3%, showing that there was a decrease of the coffee
Wageningen, WUR. (PhD Thesis – University of Wageningen)
areas planted on the smoother areas occupied preferably by
DEPARTAMENTO NACIONAL DA PRODUÇÃO
Latosols (Figures 5 and 6). This could indicate that the farmers
MINERAL. COMPANHIA DE PESQUISA E RECURSOS
are moving the coffee to areas with richer soils and leaving the
NON-REVIEWED PAPER

---

ISPRS Technical Commission II Symposium, Vienna, 12 – 14 July 2006





135

MINERAIS-DNPM/CPRM., 1979. Projeto Sapucaí. Report
ACKNOWLEDGEMENTS
DNPM/CPRM, n.5. São Paulo, SP, Brasil. 299 p.

This research was funded by the CBP&D/Café – Consorcio
EMPRESA BRASILEIRA DE PESQUISA AGROPECUÁRIA-
Brasileiro de Pesquisa e Desenvolvimento do Café. We also
EMBRAPA. Centro Nacional de Pesquisas de Solos., 1999.
acknowledge the EMBRAPA CAFÉ and EPAMIG/CTSM for
Sistema Brasileiro de Classificação de Solos. Brasília: the support.
EMBRAPA-Produção de Informação, 1999. Brasília, DF,

Brasil. 412 p.
Jones, C.B., 1997. Geographical Information Systems and
Computer Cartography. Longman, Harlow, pp. 3-17.
Lemos, R.C.; Santos, R.D., 1996. Manual de descrição e coleta
de solo no campo. 3.ed. SBCS/CNPS, Campinas, SP, Brasil.
84 p.
MINAS GERAIS. Secretaria de Estado da Agricultura, Pecuária
e Abastecimento., 1995. Cenário Futuro do Negócio Agrícola
de Minas Gerais. Belo Horizonte, MG, Brasil. 49 p.
Moreira, M.A., 2001. Fundamentos do Sensoriamento Remoto e
Metodologias de Aplicação. São José dos Campos, Instituto
Nacional de Pesquisas Espaciais (INPE), 250 p.
SPRING – Sistema de Processamento de Informações
Georreferenciadas, 2003. São José dos Campos, Instituto
Nacional de Pesquisas Espaciais (INPE), V.4.0.


NON-REVIEWED PAPER

---

---