tc
al
od
N,
les
en
en
er
en
nd
on
ich
2p
tS,
the
is
the
ith
AF
ed
fic,
ges
ton
lies
(ter
far
hip
soil
ol
int,
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
that SOC estimation by orbital remote sensing has mainly to
deal with partly or fully plant covered areas.
2. MATERIAL AND METHODS
2.1 Characterization of the study site
The study sites were located in the Brazilian Cerrado, the
second largest biome in South America. This vegetation
comprises an intricate mosaic of land cover types, vertically
structured as grassland, shrubland, and woodland. The
studied pastures are situated in the county of Piracicaba (Sao
Paulo State). It will be referred to them by their farm names
(Barreiro Rico, Bondade, Descalvado and Monjelada). The
climate is humid mesotermic with relative cold and dry
winters (Cw according to Koppen), in which the monthly
average temperature isn’t above 18°C for all months (Cwa).
Meteorological records of Piracicaba county indicate an
annual middle temperature of 21,6°C with an annual middle
precipitation of 1,166mm for the studied year 2001
(ESALQ/USP, 2003). The studied soil in all four pastures is
Neossolo Quartzarénico (Brazilian soil classification), which
refers to Psamments (U.S. soil taxonomy) and Arenosols
(FAO classification).
2.2 Pasture selection
Four pastures with different productivity levels were visually
chosen (and afterwards as representative validated). All four
pastures feature a big enough extension (approximately two
hectares) to cover entirely pixels of the satellite image, to
ensure a spectral response that is exclusively related to
pasture without interferences of other land uses. The pixel of
a Landsat 7 image represents 30m of width and length, which
leads to minimum extension of 3x3 pixels (90x90m) to
ensure an exclusive pasture pixel in the middle of the kernel.
2.3 SOC of the pastures
2.3.1 Soil sampling The soil was sampled by cylinders (¢
10cm) at following soil depths:
0—5cm/5-10 cm/ 10 — 20 cm / 20 - 30 cm / 30 —40 em /
40-50 cm
The soil was horizontally collected for each layer and pasture
by six repetitions, except the 0-5cm-layer (8 repetitions), to
account for higher spatial variety of carbon in the uppermost
layer.
2.3.2 Carbon analysis The samples were dried three days at
60 C°. Before weighing the samples, the gross roots were
removed. The samples were sub-sampled by successive
diagonal halving, maintaining a representative sub-sample.
Fine roots were removed from the sub-sample, considering
the recent endeavors to produce more precisely SOC
measures. Therefore, a plastic ruler was electro statically
charged and held over the sub-sample. The electrostatics
removed exclusively the fine roots without soil particles. The
carbon content of the sub-samples was determined by dry
combustion in a carbon analyzer (LECO CN-2000).
2.3.3 SOC calculation The SOC stocks were calculated as
following:
SOC (mg nay = carbon (s: bulk density qum» ‘ depth of
soil layer tem) © 100 (conversion factor %) 100 (conversion factor
bla si
g/m into Mgha )
797
2.4 LAI of the pastures
The LAI-sampling was done in the transition period from wet
to dry season. In this time of year, the less productive
pastures start to suffer water stress, affecting the
photosynthetic activity of the pasture, which leads to higher
differences of LAI in between pastures. The sampling was
done by a ring, covering 0,25m?. The samples were randomly
taken in May 30/31 in 2001, with 8 repetitions for each
pasture. The litter inside of the ring was removed, followed
by an entire cut of the green biomass. The samples were
immediately weighted due to minimizing humidity losses;
hereupon approximately 20% of the entire sample was
extracted and immediately weighted, to constitute a
representative sub-sample. Subsequently, the sub-samples
were separated in two fractions: green leaves and remaining
parts. The green leaves were measured by a Leaf Area Meter
(LI-COR: LI-3100). The LAI of the entire sample was
calculated as following:
LAT entire sample ^ Fresh Weight cuir somple 7 Fresh Weight xp.
sample * LAI sub-sample
2.5 Spectral behavior of the pastures
2.5.1 Satellite Image The study used a satellite image of
Landsat 7 (World Reference System: 220/76), which
represents one of the global standards in remote sensing of
terrestrial resources (USGS, 1999). The sensor ETM+ of
Landsat 7 features following spectral characteristics:
pes Principal Characteristics of Landsat 7
Band 1] 2 3 4
Bandwidth (jm):
0.45-0.52 0.53-0.61 0.63-0.69 0.78 — 0.90
Spatial resolution:
30m 30m 30m 30m
Band 5 6 7 8
Bandwith (jum)
1.55—1.75 10.4—12.5 2.00—2.35 0.52— 0.90
Spatial resolution:
30m 30m 60m 15m
Table 1. Characteristics of the spectral bands of Landsat7 -
(USGS, 1999)
2.5.2 Image Processing Before processing, a sub-image
(~35x12km) was extracted from the entire scene to enable
more precise georeferencing and processing speed. This sub-
image was then geometrically and radiometrically corrected.
It was georeferenced by 12 Ground Control Points (GCP),
taken by a differential GPS system (GeoExplorer II /
Trimble); the precision error of the used system is under 1m
(Trimble, 1999), which is considered adequate for the study
task. The GCPs were taken in locations, which were easily
recognizable on the satellite image, to assure precise
georeferencing. The GCPs were imported into a GIS
(Geographic Information System), in which the
georeferencing was compiled, using an order 3 polynomial
model. The sub-image was atmospherically corrected by an
atmospheric correction software (ATCOR-2). The extensions
of the studied pastures were marked out by DGPS. The GCPs
of the pastures were imported into a GIS (TNTmips) and
transformed into polygons, to extract the pastures as raster
objects from the sub-image.
2.5.3 Conversion of DNs into reflectance The digital
numbers (DN) of the grayscale raster image were converted
into surface reflectance (in Vo). First, the DNs were converted
