| 2004
—
h 80m
1sional
ystem.
DEM
Is. We
virtual
n was
nde an
plinary
various
nge for
three-
use to
canopy
al Park.
Process
alaysia,
ach for
/ GIS/
tion -
ocesses
alaysia,
BURNED AREA, RECURRENCE OF FIRES AND PERMANENCE OF BURNT SCARS IN
SELECTED AREAS OF THE BRAZILIAN CERRADO USING TM-LANDSAT IMAGERY
Krug, T.'?* Rivera-Lombardi, R.J.'?: Santos, J.R. dos?
'"Inter-American Institute for Global Change, IAI, Caixa Postal 515 - 12201-010 Säo José dos Campos, SP,
Brazil — thelma Gdir.iai int
?'Instituto Nacional de Pesquisas Espaciais, INPE, Caixa Postal 515 - 12201-010 Sáo José dos Campos, SP,
Brazil — (lombardi, jroberto) @ltid.inpe.br
Commission VII, PS WG VII/2
*
KEY WORDS: Forest Fire, Land Cover, GIS, Monitoring, Multitemporal, Spatial
ABSTRACT:
A study was conducted in two selected areas of the Brazilian cerrado using Landsat imagery during a five years period (1996-2000),
to address the following estimation issues: (1) burnt area using statistical sample techniques; (2) burnt scars permanence; and (3)
recurrence of biomass burning. Regarding the burnt area estimates, a decrease of approximately 6096 was observed during the period
and was consistent in both study areas. It was also observed that the physiognomies most affected (wooded savannah and shrub
savannah) remained the same throughout the period considered. A method was developed to allow a quantitative assessment to be
made for the mean time of permanence of the burnt scars. The results indicated that there exists a large annual variation that also
depends on the type of physiognomy affected. Regarding the recurrence (reincidence period), the most common pattern was
characterized by the occurrence of only two burns in the period, spaced by two years time.
1. INTRODUCTION
The Brazilian cerrado (savannah) is the second largest bioma in
South America, totaling approximately 3,000,000 km? (Eiten,
1994; Sato and Miranda, 1996). In Brazil, it covers almost 25%
of the national territory (1,800,000 km”), predominantly in the
center-west region (Mato Grosso, Goias, Tocantins States). The
cerrado vegetation is characterized by a gradient of woody
biomass distributed in trees of at most 10 metres high and in a
diversity of shrub formations. There are several definitions for
the diverse vegetation physiognomies in the cerrado. In this
paper, the definitions proposed by IBGE (1991) are adopted, as
follows: cerradäo (woodland), campo cerrado (wooded
savannah), parque de cerrado (tree and/or shrub savannah), and
campo (grassland savannah).
Biomass burning in the tropical savannah is a common practice,
not only for the expansion of the agriculture but also, amongst
others, to renew the vegetation in those areas which are used for
cattle raising. It is well known that biomass burning activities
are responsible for the emission of some greenhouses gases to
the atmosphere, in particular carbon dioxide (approximately
90%). Other non-greenhouse gases, such as carbon monoxide,
contributes to approximately 10%, and the others with roughly
| to 2% (Levine et al, 1991). Grassland physiognomies of the
cerrado are very resilient to fire and recover rapidly after
buming, through regeneration. According to Andreae (1991)
and lacobellis et al. (1994), the increase of the atmospheric CO,
concentration that results from the cerrado biomass burning is
not significant, considering its re-incorporation in the
vegetation, resulting from the regeneration process. However,
lhis re-incorporation of carbon may occur after two years in the
parque de cerrado, or even four years, in the campo cerrado
(Miranda and Miranda, 2000). On the other hand, this re-
übsortion does not apply to other greenhouse gases that result
from biomass burning, such as methane (CH,) and nitrous oxide
243
(NO). Hence, it is important to estimate the emissions of these
gases which have a warming potential many times larger than
CO».
The Intergovernmental Panel on Climate Change - IPCC
provides, in its 1996 Revised Guidelines for Greenhouse Gas
Inventory (IPCC, 1997), a methodology to estimate emissions
of some greenhouse gases that result from biomass burning.
However, one of the elements required is the total area affected
by biomass burning, by type of vegetation physiognomy, a task
which is not trivial for Brazil, in particular due to the following:
(1) an extensive vegetation area to be monitored; (2) a long dry
season (from June to October) proportional to fire occurrences;
and (3) coarse resolution of the Brazilian vegetation maps
(available only at regional scale (1:500 000). In addition, the
burning efficiency, another element in the IPCC methodology,
is a difficult parameter to assess.
Remote sensing systems are well known to provide systematic
and complete coverage of the territory, thus being adequate to
resolve issues (1) and (2) above, in a degree proportional to
their spatial and temporal resolutions. Despite the regular
acquisition of Landsat data in Brazil since mid 1970s, it is not
simple to carry out an analysis of all Landsat imagery that
covers the cerrado vegetation (estimated to be approximately
130), at every passage (16 days apart). In addition to the heavy
workload to interpret and analyze the data, many images are
affected (partially or totally) by clouds, thus impairing or
making it impossible to assess the conditions on the ground.
Krug and Santos (2001) have estimated the total area burnt in
the cerrado using a sample of Landsat images and a number of
different statistical methods. However, uncertainties associated
with these estimates exist due to the fact that not all images
acquired during the burning period are used for each path/row
sampled, due to the quality of the images.
