Mapping biotopes of river corridors using GIS, Remote Sensing, and Decision Tree Analysis for ecotechnological applications
2003 (English)In: Annals of Warsaw Agricultural University-SGGW, Land Reclamation, ISSN 0208-5771, no 34, 43-50 p.Article in journal (Refereed) Published
A biotope classification map was created for a 215 km stretch of the Ljusnan River, Sweden, by combining GIS, satellite remote sensing, and decision tree analysis. Traditional methods of biotope mapping are both labour intensive and slow. Decision tree algorithms have significant potential in this area, particularly in highly fragmented landscapes, which conventional satellite remote sensing techniques or GIS models are unsuited to. In this study, several Digital Elevation Model (DEM) derived thematic layers (Topographical Wetness Index, Slope, and Euclidean distance from river), Landsat-7 ETM derived data (NDVI and Principal Component Analysis 1, 2, 3, 4), and other ancillary GIS layers were combined and input into KnowledgeSTUDIOTM decision tree software. A decision tree was generated and was used to compile a set of classification rules. A biotope classification map (9 classes), was the final output. An overall accuracy of 50% was achieved in the accuracy assessment. The method is developed to be applied in studies, which predict the ecological consequences of nature development projects in riverine areas, and for calculations of river corridor functions.
Place, publisher, year, edition, pages
Warszawa: Warsaw Agricultural University Press , 2003. no 34, 43-50 p.
biotop map, decision support, ecotechnology, hydropower, riparian zones
IdentifiersURN: urn:nbn:se:kth:diva-83654OAI: oai:DiVA.org:kth-83654DiVA: diva2:498890
NR 201408052012-02-122012-02-12Bibliographically approved