Paper No. 302-7
Presentation Time: 10:30 AM
DELINEATING CATCHMENT BASINS BY GRAPH TRIPLES
Catchment basin analysis is essential for levelling, interpreting, visualizing, and applying stream sediment geochemical surveys. Over 64,000 stream sediment samples have been collected covering about 80% of British Columbia. With so many sample sites extending across the diverse terrains of a large landmass, it is a challenge to efficiently delineate catchment basins with consistent results. A fully automated process has been developed to address this challenge based on graph theory and SQL recursive Common Table Expressions (CTE). A graph data-store was created to represent the stream network in the province, consisting of 6.5 million triples that describe the relationships between drainage edges. The drainage edges are topographically represented at a scale of 1:20,000 and are linked to heights-of-land or watershed units delineated from mass points and breaklines in the digital elevation model (DEM). For a given sample site, the CTE based query would traversal and retrieve the upstream edges that would flow through that sample location from graph data-store of triples. The upstream watershed units (polygons) for the drainage edges are then retrieved and dissolved to form the catchment basin polygon. The graph triples were generated and stored in a PostgreSQL/PostGIS database. The CTE was developed in SQL and watershed polygons were dissolved using the CascadedUnion in PostGIS. For a typical stream sediment sample site, it takes only a fraction of a second to generate a catchment basin polygon. It takes only two minutes to retrieve the upstream edges for the over 52,000 stream sediment sample sites in British Columbia, and another 30 minutes to dissolve the 2 million watershed polygons into catchment basin polygons. This process has been packaged as a complete solution, with interfaces to Web Services and Google Earth. Retrieved streams and catchment basins can be displayed in GIS or on Google Earth.A major limitation of this method is the dependency on a well-defined stream network and pre-delineated heights-of-land with consistent and correct definition of drainages and watersheds.