n? size
lowing
cover
of all
ion and
‚000 m?
5 values
lometric
tatistical
now the
obtained
idy with
on of the
ry forest
rassland,
3).
Figure 3 - Typical aspects of the savanna area.
The mean biomass values obtained for primary and secondary
forest is around 194.71 ton/ha and 47.06 ton/ha (dry weight)
respectively (Figure 4). Areas of primary degraded forests, such
as those sections of selective logging for example, present a
significant decrease of species density and diversity, with a total
biomass below 80 ton/ha. Those areas of secondary succession
present 20 % of variation of biomass within this cover type,
which is related to the degree of vegetation development and to
the intensity to its’ previous use. Those areas of primary forests
. inventoried (¢ > 10 cm) have an average of 430 individuals/ha,
DBH of 23.60 cm and the dossel height 14.54 m (the highest
individuals are up to 38 m height). The areas of secondary
succession (¢ > 5 cm) present, in average, 1294 individuals/ha,
DBH of around 11.22 cm and 7.94 m height.
li
A
a
Biomass (Ton/ha)
a
0-499]9 |
|
l.l
|
0 T T T '
eo eo o © © © o o eo eo
98888888888888
+ “+ “+ "f "t "b +
x 3 3 FE TE FT ST I TI
1 1 L ' ' ' ' ‘ ‘ - - -— = v
o eo eo eo e e © © e : i 3 ! E
oN e ~ co o e © © © o
= 0° SS € Qq ® 8
Q c x
DBH Intervais
D Savanna B Forest |
Figure 4 - Distribution of aerial biomass by DBH intervals of
primary and secondary forest.
As for those areas covered with savanna formations, the
estimated biomass (dry weight) was: for savanna grassland 4.86
ton/ha, for parkland savanna, 7.28 ton/ha, and for savanna
woodland 20.15 ton/ha. The savanna grassland is made up
solely by an herbaceous stratum, formed by gramineas species
and cyperaceae, presenting in average 48.57 % soil coverage.
The parkland savanna presents 230 individuals/ha at the upper
stratum (bush and small trees) wich represent the 40 % of the
whole biomass. These values are different for savanna
woodland, where 82 % of the biomass belongs to the
arboreal/bush stratum, showing an average of 300
individuals/ha.
Based on these information from field survey of the typical
physiognomic structure of forest and savanna in this region, it is
possible to show the behavior of the backscatter values from
JERS-1 (Figure 5). In the models of simple regression, biomass
values were considered as dependent variable (y) and the
equation y — 1.7881 In(x) - 15.821 has a significant performance
(r^ value — 0.8746). The amplitude of backscatter values varies
from - 9.04 to - 5.77 dB for forestry formations and from - 15.07
to - 9.55 for savanna types. One can observe that this regression
function shows a high sensibility of the model for biomass
values until 100 ton/ha, the space of attributes where savannas,
secondary forest and primary degraded forests are located. From
this value on, there is a decrease of sensibility, and there are not
significant changes of o^ values (saturation zone) with the
increase of biomass from primary forest.
ort
dB Mean Values
y = 1,7881Ln(x) - 15,821 |
-14 = R? = 0,8746 |
Biomass (Ton/ha)
| $ Savanna WForest ad
Figure 5 - Diagram of JERS-1 backscatter values of forest and
savanna formations.
The components “vegetation”, “soil” and “shade”, included in
the resolution element of the orbital data, are those factors
which are responsible for this backscatter signal, associated with
the variability of the vegetation structure and soil moisture
content. This concept becomes more evident when one observes
the result of the linear spectral mixture model, applied to the
TM-Landsat image (Xaud, 1998), which presents the dispersion
of the proportions of each component for each of the classes
identified in this contact area forest/savanna (Figure 6).
Vegatation
^
5095 2. -.-----" (Qc 5096
O PRIMARY FOREST
O SECONDARY FOREST
® SAVANNA WOODLAND
® SAVANNA PARKLAND
O SAVANNA GRASSLAND
e : \
; e
Q
; a
ti “of o.
o 3 Ô
Oo Qv o6
5 N
£s
Shade 50% Soil
Figure 6 - Diagram of dispersion from components
“vegetation”, “soil” and “shade” of forest and savanna
formations.
Source: Xaud, 1998.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7. Budapest, 1998 525