Chris Bellman
USE OF A LANDSAT-TM TIME-SERIES FOR MONITORING EROSION FEATURES IN AREAS
OF GOLD EXPLORATION, SERRA TEPEQUEM, RORAIMA STATE, BRAZILIAN AMAZON
Raimundo ALMEIDA-FILHO, Edileuza C. MELO, Yosio, E. SHIMABUKURO
National Institute for Space Research-INPE, Caixa Postal 515
Sáo José dos Campos-SP, Brazil
rai 9 Itid.inpe.br; yosio € Itid.inpe.br
Working Group VII/2
KEY WORDS: Remote Sensing, erosion features, renewable resources management
ABSTRACT
This paper discusses the use of a Landsat-TM time series for monitoring erosion features caused by the action of
independent miners (garimpeiros) in the search for gold. Five free-cloud Landsat-TM scenes acquired in March/87,
October/91, November/94, January/96, and January/99 were used to produce maps showing temporal evolution of
mined areas in the region of the Tepequém plateau. To produce detailed thematic maps of the degraded areas, images
were segmented using a region growing approach, followed by the application of a region-classifier algorithm based on
clustering techniques. According to these maps, eroded areas increased by 388% from 1987 to 1994, comprising 1070
ha, 2,550 ha, and 4,150 ha, in the 1987, 1991, and 1994 images, respectively. Due to a governmental action, combined
with the exhaustion of the mineralized alluvial deposits, garimpeiros abandoned the region around 1993/1994. As a
result, vegetation soon started recovering eroded areas. Images acquired in 1996 and 1999 showed that degraded areas
decreased to 3,660 ha and to 3,391 ha, respectively. This work shows that Landsat-TM images may be an efficient mean
for monitoring erosion features, even under the ecological characteristics of the Tepequém plateau, where eroded areas
have low contrast relative to surrounding open grass fields. Additionally, the image processing procedures permitted to
save time and to curtail inherent subjectivity involved in the visual interpretation, producing accurate thematic maps of
land use/cover change classes.
1. INTRODUCTION
During the late eighties and early nineties, Roraima State, and particularly the region of the Tepequém plateau, was
severely exploited by independent miners (garimpeiros) in the search for gold (Oliveira, 1995), resulting in several
degraded areas along streams. In the search for gold, garimpeiros use high-pressure water jets to wash mineralized
alluvial deposits. This practice causes removal of the vegetation cover along stream banks, so creating conditions to
start erosional process in mined areas. In spite of constituting a serious environmental problem, few attention has been
given to study the impact caused by informal gold mining activities (garimpos) in the Brazilian Amazon. This kind of
activity causes two major environmental problems. The first one is related to the pollution of rivers and streams due to
the use of mercury amalgamation for gold recovery. The second problem is the deforestation and erosion along mined
streams. This second mining-derived problem may be addressed directly by remote sensing techniques.
Using visual interpretation of Landsat-Thematic Mapper (TM) images, Melo & Almeida-Filho (1996) showed
preliminarily that mined degraded areas could be mapped under the physiographic conditions of the Tepequém plateau.
More recently, Almeida-Filho & Shimabukuro (1999) showed that the degraded areas in that region could also be
identified using orbital radar images acquired by the Japanese Earth Remote Sensing satellite (JERS-1). In this study we
use techniques of automatic classification applied to a Landsat-TM time series, aiming at to map and to monitor the
evolution of areas degraded by gold mine activities in the Tepequém plateau region, over a twelve-year period, from
1987 to 1999.
2. STUDY AREA
The study area covers a surface of approximately 150 square kilometers, located in the north-central part of the State of
Roraima, Brazilian Amazon, close to the border with Venezuela (Figure 1). It includes the Tepequém plateau, being
defined by the following geographic coordinates: 61° 48” 15°” - 61° 37° 53° west longitude and 3° 50’ 30°’ - 3’ 44°20”
66 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000.
north
to Se
highe
The :
(Borg
1,000
veget
Figur
3. L/
In th
acqui
coor
proce
refle:
attem
The
deve
digit:
perm
level
(a) s:
merg
COM]
(d) tl
Filhc
Lanc
betw
imag
of pi
À re
bed
class
class
class
take
and |
the 1