types of
wior of
different
* «n
N uo
N N
< <
sion
on
primary
asons.
r values
red with
growth.
) have a
explains
a high
vis-à-vis
sts. The
utes the
spectral-
nes that
| such as
2), Thé
munities
) species
At secondary successions, the amplitude of seasonal changes of
backscatter increases with the successional stage, from the
initial growth to advanced stage (age > 15 years). Those areas at
initial growth (age <S years) showing a more homogeneous
canopy and with low species diversity, had a similar response at
both RADARSAT datatakes analyzed (Figure 4). It is important
to note that during the RADARSAT imaging of May 15th and
October 23rd ‘96, there was rainfall of around 25 mm/day,
which minimizes the effect of moisture, when comparing the
backscatter of both dates.
itepeum
m — ee meta es s + resides p
135
1
. v »
3 e |
9 s - . ®
: ww
9
1
0
$. > : ——
1C © lerth ) P Rofirie sp. 3 Bsa PRT Ep
(Tdscoss] 5 Bine aiboree Hub. & Acalyphe irri 7 Aca echneaiT laete! 8 Acieha ms soda 1
an (Tienen 1 Cabe paunte B Lent Gnbtraceml Inge sp. Manceacess) 1 Hub [Bacese) 14. a abo
Hh Buacess) 15. Bio Hub. Buscess] 16. Bas abore Hub. Boucese] 17. Inge ep. AE SST Bos
Figure 4. Physiognomic profile of an initial secondary
succession.
At Figure 5 one can observe the seasonal backscatter (y°) values
plotted against biomass for the vegetation types under study.
The diagrams of data dispersion, independently of the
RADARSAT imaging date, show that there is a very low
sensitivity of the models to the relation backscatter against
biomass.
0,35
8 , S .
B e e
S 025 ; 8 a:
Ed e e
$ 02 u e
z 0,15 + gB a
8 0,1 +
3 0,05 + y * 0,0093La(x) + 0,1932
R? = 0,0135
0 + + ; ;
0 50 100 150 200 250
Aboveground biomass (ton/ha) May, 15th
0,35
3 ' 2
03
3 [| a = ® hd en
0,25 4
= pete SR
e
$ x ae
€ ois m
B oi e
à 0,05 + y = 0,0262La(x) + 0,1314
R? = 0,0669
0 + + 4 1
0 50 100 150 200 250
A x
boveground biomass (ton/ha) Ociobe 298b
€ Primary Forest ® Secondary Forest
Figure 5. Diagrams of RADARSAT backscatter (1°) against
aboveground biomass values of forest and secondary
succession.
The backscatter mean values (y?) for areas of secondary
succession present a variation range from 0.1563 to 0.3309 and
for areas of primary forest of 0.2078 to 0.2991. Even when there
are distinct mean values of aboveground biomass for primary
forest (170.25 ton/ha + 49.28), primary forest with bamboo
(93.61 ton/ha + 18.32), initial secondary succession (35.66
ton/ha + 13.48), intermediate secondary succession (56.33
ton/ha + 11.02) and advanced secondary succession ( 85.62
ton/ha + 26.73) there is no significant correspondence. The
intensity of sampling (18 plots) was considered sufficient to
characterize the needed measurements, totaling 600 trees of
primary forest and 284 trees of secondary forest, measured at all
sample sections. On the forest with bamboo, where the field
measurements are difficult, specially due to the intense
interlacing of these bamboo in the under-growth, where all trees
and 508 thatches/0.2 ha were measured (DBH>2.5 cm).
The intermediate and advanced regrowth classes present a
significant amplitude of backscatter values, whose variation
cannot be fitted into a theoretical model (behavior), which could
express medium and high biomass values. A theory to explain
such configurations could be related to the type of management
(frequent burning, soil compaction, land use type and duration)
used in this area before the regeneration process. In practical
terms, the floristic and biomass recomposition of such areas is
different. Furthermore there are limitations of C-band for the
penetration beyond the canopy, and as a result there is a low
saturation level (sensitiveness) for higher biomass values.
The C-band images appear to be of little use to monitor biomass
density in tropical forest under regeneration (Luckman et al.,
1997), although the maximum retrievable biomass estimate,
suggested by the backscatter response is 38.6 ton/ha. According
to Imhoff (1995) the values for the broadleaf evergreen forest is
around 20 ton/ha. At this study, the distribution of y? values
against aboveground biomass allows to consider as adequate
this relation at the maximum limit of 50 ton/ha.
When we consider only the discrimination among those areas in
the process of initial regeneration (Figure 4) from those sections
that are structurally more developed or at climax (Figure 6) such
as the primary forest, one can perceive that they occupy a
distinct space of attributes at the backscatter.
in. Platycyamus ulei à (Fabacose) 2 Sloarea brevipes [Else
(Martius (Aracacoae) 5. Sor 'orocea sp (Moraceae]
! Tetragastris
!Sociatea exonhiza (Mart ) (^ ) 14. H Sra as ul Sup ([Euphorbi 15. Carapa guianensis Aublet (Meliaceae) 16. Metrodorea
flavida Krause (Rutacoae) 17. Caron us Moracase }1 à Ectecies ened 19 nla 20. Carapa guianensie Aublet
{Melacess) 21. Pseudolmedia laevis (R. ! PINS Mate) 2 tech = ca tapa guianensis Aublet (Melacose] 24.
{Taie sp. (Sapindaceas) 25 Tot etragastris altissima (Aublet) Swert wart (Burseraceae) 26. Apeiba sp. (Ti
Figure 6. Structural profile of the open tropical rainforest.
This can be simply explained by the uniformity of the canopy of
the initial natural regeneration, originated from only one stratum
International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 529
