1997) was used. A detailed description of the entire 
methodological procedure can be found in Xaud (1998), which 
will be briefly summarized below. 
The TM-Landsat scene was processed by the linear spectral 
mixture model, and it became the database to discriminate land 
cover classes and to define and locate the sample areas to 
collect information for field work. This image was merged to 
the JERS-1 scene, and when the error of fitting was less than 1 
pixel, it was considered adequate to extract o? values (dB) from 
coverage was performed in at least 5 sections with Im? size 
each. 
In those areas covered by forest (14 samples) the following 
measurements were made: DBH, height, crown cover 
percentage as well as the botanical identification of all 
individuals with DBH > 5 cm for secondary succession and 
DBH > 10 cm for primary forest, at sample areas of 1,000 m? 
and 2,500 m? respectively. The estimation of biomass values 
was modeled by dendrometric parameters into the allometric 
equations: 
the previously selected sample areas. The JERS-1 scene was Y = 0.044 . (DBH?. height)??? 
processed at the amplitude mode, in order to better visualize the 
area under study. After the sample areas were plotted on the 
amplitude image, it was transformed in intensity image, and the 
backscatter values were extracted, considering the following 
  
according to Brown et al. (1989) for primary forest, and 
  
InY = - 2.17 + 1.02 In(DBH)’ + 0.39 1n height 
  
  
equation (Rosenqvist, 1997): Figu 
according to Uhl et al. (1988) for secondary forest. Statistical The 
o° = 10 logo {(E DN?) / n}+CF tests, using simple regression models were used to know the fores 
relation between field data (biomass values) and those obtained 
where: from orbital images. : En: 
DN = digital number of an image pixel (16 bits) signi 
n = number of pixels of the sample area biom 
CF = offset calibration factor (-68.5 dB). RESULTS he 
During the field survey, two procedures for data collection were The joint analysis of landscape at the area under study with pis 
used, to know the physiognomic-structural characteristics of JERS-1 and TM-Landsat data, allowed the discrimination of the ; 
vegetation cover. In those areas covered by savanna, the following land cover classes: primary forest, secondary forest FR 
biomass estimation was made by cutting and weighting (intermediate secondary succession), savanna grassland, indiv 
herbaceous, bush and shrub vegetation. The sample areas (29) parkland savanna and savanna woodland (Figures 2 and 3). Cn 
| had a variable size of 200-500 m” each, according to the SU 
| physiognomy of the savanna; all individuals of bush and/or 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
    
  
  
  
arboreal features were identified botanically and weighted. For [= 
the herbaceous strata, sampling to collect biomass and % soil 
z 
= 
É 
0 
S 
£ 
9 
m 
Figure 
‘| As fo 
0} = - 150 | estima 
4. 2. Trema micrantha Blume. (Uimaceas) 3 indeterminate 4. Cecrcpia leucocoma Miquel. (Moraceae) 5. indeterminate 6. Trema micrantha Blume. (Uimaceas) 7. ton/ha 
& Trma miorantha Blume. (Ulmaceze) O. Inga ep. (Leguminosae mimcscideae) 1Q indeterminate 11. Ceorpia leucoooma Miquel (Moraceag) 12. Tema 
micrantha Blume. (Uimaoead) 13. Cecrcpia leucocoma Miquel. (Moracezs) 14. Trema micsantha Blume. (Uimanezs) 16 Ceorpia leuocooma Miquel. (Moraoeag) 10 woodl; 
17. 18. Trema mictantha Blume. (Uimeceas) 1Q 20 21. indeterminate 22. Inga sp. (Leguminosae mimceoideas) 23. Ceorcpia leuocooma Miquel. (Moraceae) 24. 
Inga sp. (Leguminceae mimoscidead) 25. Trema micrantha Blume. (Uimaceas) 288 Bagasea guisnensis Aubl (Moraceae) 27. Cearcpia leuoccoma Miquel. solely 
(Morzceas) 28 Trema micrantha Blume. (Uimaceas) 28. 30 Cecropia leucccoma Miquel. (Morzceas) 31. Trema micrantha Blure. (Ulmsceas) 3X2. Inga sp. and c 
(Leguminosae mimoscideae) 33 Trema micrantha Blume. (Ulmacezg) 34. Cordia ep. (Baaginaceas) 35 Tema mictantha Blume. (Uimaceas) 3B. Ceacpia J 
leucocma Miquel (Moraceas 7. 38 Ingasp. (Leguminceae mimcecideas) 3a 40. 41. Cecrcpia leuocooma Miquel. (Moranead) 42. 43 Trema micrantha Blume. The p: 
(Ulmaceag) 44. Cecsopia leucoooma Miquel. (Moraoeze) 45 48 Trema micrantha Blume. (Ulmacese) 47. 48 Cearcpia leuocooma Miquel. (Moraceae) 43 50 Inga 
ep. (Leguminceze mimcscideae) 51. Ceorcpia leuccooma Miquel. (Morzoeze) £. 53. 54. hgasp. (Leguminceze mimoeoideas) ^5. 68. Cecicpia leuccooma Miquel. stratun 
(Moraceag) 5. £B. Trema micrantha Blume. (Ulmaceze) SA Cecropia leucccoma Miquel. (Moraceas) CO Trema micsantha Blume. (Ulmeceag) O1. Inga ep. whole 
(Leguminoeae mimosoideag) €2. Trema micrantha Blume. (Ulmaceag) O3 64. Ingaep. (Legurninoeae mimoscideze) €5. Trema micrantha Blume. (Ulmaceag) Cà 
Inga «p. (Leguminceae mimcscideag) C7. Cecrcpia leucocoma Miqual. Moraceaa CB. 6. Trema micrantha Blume. (Ulmaceag) woodl: 
te arbore. 
individ 
Figure 2 - Example of typical physiognomic profile of a secondary forest. 
  
  
524 International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998