The new concept of areal types was developed to overcome the limits of traditional land-use classification methods like the maximum-likelihood approach. Looking on urban structures the planners are dealing with, there is often poor correspondence between the pixel classes directly derived from the satellite and the urban planning units. Problems caused by mixed pixels, image distortions or surface structures representing different forms of land-use with similar electromagnetical properties, could be abolished or minimized to a high degree. This is particularly important for a highly differentiated urban region.

Areal types are complex spatial aggregates, usually consisting of several distinct pixel classes, which contribute to them in characteristic proportions depending on the areal type. Areal types represent physiognomic as well as functional units.

Following our intention to reach a high degree of automation, objectivity and transparency, we designed a rule-based classification scheme dealing with the areal percentages of the satellite-based pixel classes. The areal percentages of each pixel class have been calculated for every single grid element with a resolution of 100m and 25m for the whole test site. The automated classification resulted in 15 areal types. Most of the rural areal types like water or forestial areas could be delineated with simple majority rules for the dominating pixel class of this areal type. Other areal types, especially the ones describing urban, heterogeneous settlement structures necessitated complex rules combining the percentages of several pixel classes with 'AND' or 'OR' conditions.

Tab.: Areal Types applied in KABA and their spatial characteristics

For further, more detailed information you may click on Proceedings, see publication list or contact Dr. Dieter Scherer or Dr. Ute Fehrenbach.