Fuzzy identification and manipulation of source channels in channel network derived from Digital Elevation Model
The accuracy of channel networks extracted automatically from a grid digital elevation model (DEM) generally relies on the selection of a Critical Source Area (CSA) threshold. Using an unique CSA within an entire DEM will lead to the assumption that all source (1st order) channels have an equal contributing area. Thus, overshoot and undershoot problems, excessive and deficient representation of source channels, appear frequently in the channel network, especially for a DEM of a hydrologically heterogeneous region. This research focused on these problems by introducing fuzzy set theory for source channel identification and manipulation. A comprehensive valley morphology membership function is employed to calculate the CSA for each individual source channel. The function depends on the relationship between the contributing area and valley morphology. A new approach, Fuzzy Identification Channel Network Model (FICNM), is proposed and developed in the research. The main procedure of FICNM includes: ( 1) generating an extremely dense channel network by DEDNM (Digital Elevation Drainage Network Model) to avoid undershoot problems; (2) collecting valley morphology information for each source channel; (3) calculating valley morphology membership grade and converting the grade to CSA for each source channel; and ( 4) eliminating some source channels partly or totally to solve the overshoot problems. Performance of FICNM has been tested in three DEMs. The results indicate that FICNM works well in heterogeneous regions and can set flexible CSA for individual source channels. Therefore, FICNM can enhance the potential reliability of channel network derived from DEM.