5 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004 
  
Class D: Represents 17.20% of the total area. This class is 
related with the surrounding relief to the Guadix basin. The 
class has a mean height value of 1272m and its variability is 
greater (coefficient of variation around 33%) due to the well 
developed hydrological network (gullies). Nevertheless, the 
directional variograms present a continuous and isotropic 
behaviour at distances lower than 1.22. The variability is 
slightly lower in E-W direction due to the foot of Sierra Nevada 
range. 
Class E: Represents 13.85% of the total area. This class is 
related with the hillside of the surroundings relief. The class has 
a mean height value of 1728m and present a considerable 
variability (coefficient of variation around 27%). The 
variograms present a small variability in NO°E and N45°E 
directions according to the north hillside of Sierra Nevada. 
Class F: Represents 8.52% of the total area. This class includes 
high-terrain areas of the Sierra Nevada and Sierra de Baza 
mountains. The class has a mean height value of 2018m and 
presents a coefficient of variation of 22%. The directional 
variograms presents a greater continuity in NO°E direction for 
distances greater than 1000m. 
Class G: Represent only 2.36% of the total area and it is related 
with the highest areas of the zone (Sierra Nevada and Sierra de 
Baza mountains peaks). The class has a mean height value of 
2175m and present a coefficient of variation of 21% (similar 
that the previous class). The directional variograms have a 
similar behaviour has the corresponding ones of the Class F. 
3.5 Geological interpretation 
Classes D, E and F (light green, yellow and red colors in Figure 
5) correspond with rocks which belong to the basement of the 
basin (Alboran Domain and Sudiberian Domain). In classes A 
and B (purple and blue colors) appear sedimentary rocks of the 
Guadix basin. The present fluvial pattern, corresponding to 
Guadalquivir river, can be identified in the Classes B and C 
(blue and green colors). Areas classified as Class A (purple 
colors) correspond to a glacis surface which represents the 
termination of the endorheic infilling. 
4. CONCLUSIONS 
The proposed method for the morphological terrain 
classification using statistical and geostatistical information 
provides valuable information about the terrain. This 
information is obtained from an objective procedure based in 
the ISODATA non-supervised classification using multivariable 
digital terrain model. The model is composed with a large 
number of variables that reflect different characteristics of the 
terrain (mean terrain height, variability, anisotropies, ...). 
The used information in the classification process provides a 
more robust classification that the usually applied based in a 
small number of variables. The classical procedures are based 
in a direct classification using in many times only one variable 
(for example, heights or slopes). 
The presented example shows a clear relationship between the 
established classes and lithological units of the area, providing 
additional statistical information very useful for the numerical 
terrain characterization. 
67 
REFERENCES 
Calvache, M.L. y Viseras, C. 1997. Long-term control 
mechanisms of stream piracy processes in southeast Spain. 
Earth Surface Processes and Landforms, 22(2), pp. 93-105. 
Fernández, J., Soria, J.M., Viseras, C., (1996). Stratigraphic 
architecture of Neogene basins in the central sector of the Betic 
Cordillera (Spain): tectonic control and base level changes. In: 
Friend, P.F., Dabrio, C.J., (eds.) Tertiary basins of Spain: the 
stratigraphic record of crustal kinematic, Cambridge University 
Press. Cambridge, pp. 353-365. 
Goovaerts, P. 1997. Geostatistics for natural resources 
characterization. Applied Geostatistics Series, Oxford 
University Press, New York, pp. 483. 
Leenaers, H.; Okx, J.P. and Burrough, P.A., 1990. Comparison 
of spatial prediction methods for mapping floodplain soil 
pollution. Catena, vol. 17, pp. 535-550. 
Moffat, A.J.; Catt, J.A.; Webster, R: and Brown, E.H., 1988. A 
re-examination of the evidence for a Plio-Pleistocene marine 
transgression on the Chitern Hills. I. Structures and surfaces. 
Earth Surface Processes and Landforms, vol. 11, pp. 95-106. 
Mulla, D.J., 1988. Using geostatistics and spectral analysis to 
study spatial patterns in the topography of southeastern 
Washington State, USA. Earth Surface Processes and 
Landforms, vol. 13, pp. 389-405. 
Soria, J.M., Viseras, C., Fernández, J., 1998. Late Miocene- 
Pleistocene tectono-sedimentary evolution and subsidence 
history of the central Betic Cordillera (Spain): a case study in 
the Guadix intramontane Basin, Geol. Mag., 135, pp. 565-574. 
Webster, R. and Oliver, M.A., 2001. Geostatistical for 
Environmental Scientists, Statistics in Practice Series Edition, 
John Wiley & Sons, Chichester, United Kingdom, pp. 271. 
Tou, J.T. and González, R.C., 1974. Pattern recognition 
principles, Addison-Wesley Publishing Company, Reading, 
Massachussets. 
ACKNOWLEDGEMENTS 
The present study has been sponsored by the grant REN2001- 
3378/RIES from the I--D-I program of the spanish Ministerio 
de Ciencia y Tecnología, partially financed by FEDER funds of 
the European Union. 
ea en SF 
UMSO 
pev 
iii A SESE 
n 
i 
| 
qi