the selection of enough CPs with good spatial 
distribution, being the minimal qual ity criterion 
that all points belong to region (i). 
The W3 data set corresponds to a Buritama (Sao 
Paulo State, Brazil) image. In this case the 
selection of the nine CPs was easier, rendering an 
spatial distribution better fitted by a "repulsive" 
model (Fig. 3a). The plot 3b shows that the spatial 
quality of these points is mostly "optimal", or 
"acceptable", 
Thus, the presented methodology allowed the 
assessing of the quality of the spatial 
distribution of these three data sets, Also, it 
suggests simple options to improve these qualities. 
APPENDIX: COMPUTATIONAL INFORMATION 
Hardware: SUN SPARK 1 STATION. 
Language/Compiler: SUN C++ Version 2.0. 
Special functions used: gsort provided in stdlib.h. 
Pseudorandom number generator: g9827 (Bustos, 
1990). 
CPU time used: about one second for every presented 
result. 
ACKNOWLEDGMENTS 
The authors are gratefull to Dr. Evlyn M.L. de 
Moraes Novo and to Ecologist Silvana Amaral for 
supplying the data sets of the Tucurui and Serra do 
Roncador areas, respectively. 
REFERENCES 
BUSTOS, O.H., 1990. A TURBO C implementation of a 
GFSR[0,1] pseudorandom number algorithm. Informes 
de Matematica B-62, IMPA, Rio de Janeiro, Brazil. 
DIGGLE, P.J., 1983. Statistical analysis of spatial 
point patterns. Academic Press, London. 
FORD, G.F.; ZANELLI, C.I., 1985. Analysis and 
Quantification of Errors in the Geometric 
Correction of Satellite Images. Photogrammetric 
Engineering and Remote Sensing, 51(11): 1725-1734. 
MASCARENHAS, N.D.A.; VELASCO, F.R.D., 1989. 
Processamento digital de imagens. Buenos Aires, 
Kapelusz, 1989. Quarta Escola Brasileiro-Argentina 
de Informatica, Termas de Rio Hondo, Argentina. 
ORTI, F., 1981. Optimal distribution of control 
points to minimize Landsat image registration 
errors. Photogrammetric Engineering and Remote 
Sensing, 47(1): 101-110. 
111