RS, Yol. XXXVIII, Part 7B 
In: Wagner W., Székely, В. (eds.): ISPRS ТС VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Yol. XXXVIII, Part 7B 
31 
64 
0.07 
0.32 
0.45 
66 
0.13 
0.33 
0.41 
42 
0.20 
0.44 
0.46 
63 
0.04* 
0.34 
0.40 
.48 
0.16 
0.20 
0.40 
.27 
0.45 
0.39 
0.45 
.09 0.05 0.41 
0.54 
r band 3 and 4 (adjusted R 2 
column shows the window 
;en pixels (d). Boxed values 
574 enti 35 -0.0055 asmo 
0.00064 con 135 (2) 
]4 В - 0.0662 R 
103 asm 90 
00246 con 45 (3) 
4. CONCLUSIONS AND FUTURE WORKS 
In this article two methodological approaches were used to map 
and model the structure of riparian vegetation in a Brazilian 
savanna region. For this, combining a high resolution image, the 
MAGIC segmentation/classification software and the use of 
texture measurements was evaluated. The results demonstrate a 
great capacity of the MAGIC program to identify regions of 
riparian forest without the need of field data. For this step, the 
best results were obtained by using all four spectral bands of the 
Ikonos image and a sufficient number of classes to account for the 
wide variety of land cover within the non-riparian class. 
Statistical analysis between the parameters obtained in the field 
and image processing results permitted the creation of explanatory 
vegetation structure models applicable regionally. The best 
models were obtained for the allometric variables basal area and 
volume (0.61 and 0.66), using window size of 11x11 pixels and 
distance analysis of four pixels. The direction did not appear to be 
critical but some texture parameters (ASM and Entropy) are more 
frequently chosen by the stepwise feature selection. Moreover, it 
is the diversity of measurements that appear to be most effective. 
Future work will include a much broader range of plots in 
different segments of the river in the hope of creating a more 
robust set of models. Texture feature will also be made direction- 
invariant. The approach taken here is comparable to an object- 
oriented approach that is much more appropriate for high 
resolution images. It will eventually be integrated into a single 
package. 
Haralick, R, 1979. Statiscal and structural approaches to 
texture. The Institute of Electrical and Electronics 
Engineers, Inc., 67 (5), pp 786-804. 
Jensen, J. R., 2007. Remote sensing of the environment: an 
earth resource perspective. Pearson Education, Inc. 
Prentice Hall, London, 598p. 
Johansen, K. and S. Phinn, 2006a. Linking riparian vegetation 
spatial structure in Australian tropical savannas to 
ecosystem health indicators: semi-variogram analysis of 
high spatial resolution satellite imagery. Canadian J. of 
Remote Sensing, 32(3), pp. 228-243. 
Johansen, K. and S. Phinn. 2006b. Mapping structural 
parameters and species composition of riparian vegetation 
using IKONOS and Landsat ETM+ data in Australian 
tropical savannahs. Photogrammetric Enginnering & 
Remote Sensing, 72(1), pp. 71-80. 
Lillesand, T.M., and R.W. Kiefer, 2000. Remote Sensing and 
Image Interpretation, 4th edition, John Wiley and Sons, 
Inc., New York, 724 p. 
LWRRDC, 1999a. Riparian land management technical 
guidelines: volume 1, Land and water resources research 
and development corporation, Canberra, 194p. 
Maillard, P., Alencar-Silva, T., and Clausi, D. A. 2008. An 
evaluation of Radarsat-1 and ASTER data for mapping 
veredas. Sensors (Basel), v. 8, p. 6055-6076. 
ismi35 + 0.000001 IR 
15 
- 0.00555 asíais 
+ 0.000137 C01145 
lo- 0.000531 ent 135 
■45 
14 ent 90 + 0.136 R 
0.143 asm 90 -0.101 G (7) 
573 con 0 - 0.00337 cor 0 
90 + 0.00354 cor 45 (8) 
ng factor in the models and 
antly. It is also difficult to 
neasurement that stands out. 
rs, the ASM seem to be 
py seems to come in second 
/ of measurements is the best 
nts for their strength. When 
together, it can be verified 
nd Entropy are predominant 
d volume). These models are 
>f the riparian forest and are 
in another region. However, 
ly to orientate the riparian 
rently being undertaken in 
Minas Gerais by the Forest 
(4) 
(5) 
(6) 
REFERENCES 
Alencar-Silva, T. and P. Maillard. 2009. Segmentaqao de imagem 
de alta resoluto utilizando o programa SMAGIC. 
Proceeding of XIVSBSR, Natal, Brazil, pp. 6743-6750. 
Barbosa, L, P. Maillard and T. Alencar-Silva. 2009. Mapping 
wetlands variation using high resolution image in the 
Pandeiros wildilife sanctuary, Brazil. Proceeding of 30 lh 
Canadian Symp. on Remote Sensing, Lethbridge, Canada. 
Biodiversitas, 2005. Biodiversidade em Minas Gerais: um atlas 
para sua conservado [Biodiversity in Minas Gerais: na atlas 
for its conservation] (Second Edition). Fundagao 
Biodiversitas, Belo Horizonte, Brazil, 242p. 
Clausi, D.A., K. Qin, M.S. Chowdhury, P. Yu, P. Maillard. 
Accepted April 2009 MAGIC: MAp-Guided Ice 
Classification, Canadian Journal of Remote Sensing. 
Congalton, R.G., K. Birch, R. Jones, and J. Schriever, 2002. 
Evaluating remotely sensed techniques for mapping riparian 
vegetation. Computers and Electronics in Agriculture, 37(1- 
3), pp. 113-126. 
CSIRO, 2003. Determination of sustainable river audit (SRA) 
habitat indicators by remote sensing, Technical Scoping 
Document, Technical Report 28/03, CSIRO Land and Water, 
Canberra, 105 p. 
Davis, P.A., M.I. Staid, J.B. Plescia and J.R. Johnson, 2002. 
Evaluation of airbone image data for mapping riparian 
vegetation within the grand canyon, Report 02-470, U. S. 
Geological Suvey, Flagstaff, Arizona, 65p. 
Muller, E., 1997. Mapping riparian vegetation along rivers: 
old concepts and new methods. Aquatic Botany, 58, pp. 
411-437 
Nagler, P.L., E.P. Glenn, and A.R. Huete, 2001. Assessment 
of spectral vegetation indices for riparian vegetation in the 
Colorado River delta, Mexico. Journal of Arid 
Environments, 49(1), pp. 91-110. 
Tso, B. and P. Mather. 2001. Classification Methods for 
Remotely Sensed Data (First Edition). Taylor and 
Francis, London. 242 p. 
Yu, Q. and D.A. Clausi 2008. IRGS: image segmentation 
using edge penalties and region growing. IEEE Trans. 
Pattern Analysis and Machine Intelligence, 30(12), pp. 
2126-2139. 
ACKNOWLEDGEMENTS 
The authors are thankful to the Forestry Institute of Minas 
Gerais for providing the Ikonos data and field support. For the 
MAGIC package we thank Dr. David Clausi. The authors are 
most grateful to the Laboratory of Ecology and Plant 
Propagation / UNIMONTES University for providing the 
field data.