Automatic extraction of the wadi channel network for topographic mapping Rafe Abu Ashour The UAE Military Survey, PO Box – 3947 Abu Dhabi, UAE Email: rashour@uaesurvey.ae Abstract Digital Elevation Models have been widely used for extracting significant topographic features like contour lines, spot heights, slope and aspect. Further processing using GIS raster processing techniques slope and aspect could be used to accurately define the catchement areas and the channel drainage networks. This paper presents the methodology of extracting the wadi channel network from DEM and conducts a comparison assessment with their homologue features on topographic maps. Introduction The use of Digital Elevation Models (DEM) for extracting morphologic indices like slope and aspect has found a broad range of applications in studies that require careful description of the topographic surface like hydrology, road construction and soil erosion. One important application that have been widely used in surface hydrology modeling is the automatic extraction of the channel drainage network and the delineation of the watershed boundaries. Besides being more efficient in terms of time required to produce those layers the automatic procedures are less subject to errors especially in flat areas where it is difficult to define accurately the channel network (Burrough,1990). This paper will describe a GIS model for the automatic extraction of the drainage network and examine the quality of the extracted features as compared to the 50k topographic maps blue lines. Description of the procedure The model described here is based on the procedure of automatic extraction of the channel drainage network described by (O’Callaghan and Mark in Beven,K. and Moore,I(1993) . The basic concept of the method relies on the assumption that water moves from the processing cell to one of its 8 neighbor cells following a unique path according to the steepest slope direction (figure 2). Although this may not be representative of the real water flow conditions, the single flow direction is an acceptable approximation of the reality and has the advantage of simplifying the complex flow dynamics processes. Initially, the DEM involved In the extraction process should be free of sinks. Sinks represent locations of internal drainage where no flow out of the cell can be found. Removing the DEM sinks requires filling each sink to its pour point which can be achieved by giving the sink the same height as that of the minimum elevation of the cells that constitute the watershed area contributing flow to it. Based on the flow direction raster the amount of flow that is accumulated by each cell in the DEM may be calculated. The resulting flow accumulation raster will form the basic layer from which the channel drainage network will be extracted Extracting the channel network The process of extracting the drainage channels from the DEM consist In selecting all the cells with flow accumulation exceeding a certain threshold value. The selected threshold value will determine the density of the extracted network and defines its resolution in terms of stream order. To serve the purpose of using the drainage network for topographic mapping the density of the extracted network should be comparable to that of the blue lines on the topographic maps. Figure (3) shows a comparison between the topographic maps blue lines and drainage channels extracted at threshold values of 1000,500 and 200. the results show that the density of the drainage network extracted at a threshold value of 200 (figure 3-c) is comparable to that of the topographic maps blue lines (figure 3-d). Assessment of the results Quick overlay comparison test between the two sets shows good match in hilly areas where flow pathways are well defined (figure 4). In the wadi bed areas the automatic extraction process tends to define the single flow pathways within the flat wadi beds (figure 5-a) whereas in the topographic maps wadis appear as polygons rather than flow lines (figure 5-b). The single flow path as defined by the extracted network is useful for hydrologic modeling purposes where a well defined flow pathway is required. an additional advantage of the automatically extracted drainage channel is the perfect match between the drainage network lines and the shapes of the contour lines that the cartographers used to spent significant time to check and adjust (figure 5). References Burrough,P. (1990) Principles of Geographic Information Systems for Land Resources Management. Calderon Press, Oxford Beven,K. and Moore,I. (1993) Terrain Analysis and Distributed Modelling in Hydrology. John Wiley & sons, Chichester Kvak,K. and Nachtenebel,H. (1993) Applications of GIS in Hydrology and Water Resources. IAHS publicacations