Knowledge-based classification of land cover for the quality assessement of gis database

Valid classification of remotely sensed data is one of the most studied issues in the geoinformation science. In recent years knowledge-based approach to image analysis has been developing for assessment and improvement of traditional statistically-based image classification. Knowledge-based classification procedure integrates remote sensing imagery with ancillary geospatial information from GIS. Data about land cover stored in GIS database are usually subjected to an intensive change processes that diminish their relevance and include different types of discrepant information. Classification of land cover by up-to-date satellite imagery and automatic updating of GIS database allows revision of discrepant or erroroneous data. The knowledge-based classification doesn't require any assumptions regarding the data distribution and allows straightforward incorporation of ancillary data from GIS. Compared to traditional mapping approaches knowledge-based classification has the advantages of lower cost, area-wide coverage, and possibility to frequent updating. In perspective of large GIS maintenance the knowledge-based classification may contribute to detection of change and assist automatic updating of spatial databases. The objective of this study was to perform knowledge-based classification of land cover using satellite remote sensing data and GIS ancillary data. The selected target groups of land cover from the Israeli National GIS have been characterized spectrally by multispectral IKONOS data and geometrically by GIS data. The formalized knowledge about the target groups was incorporated into classification of remote sensing data. By means of classification results the discrepant land cover polygons have been detected and suggested for revision. Discussed are classification results and the analysis of detected discrepancies. The classification results have provided an indication of the utility of formalized knowledge for classification of land cover. The proposed method could be one possible approach to quality assessment and may contribute to automatic updating of existing spatial databases.