Abstract
This paper presents a fast converging iterative algorithm for implicit flood routing in channel networks of general configuration. Flood routing in a looped network can be simplified to the problem of dendritic networks by either dividing the original network into dendritic blocks or/and simply opening the loops. The dendritic subnetworks are then solved separately by the effective double sweep method assuming flow conditions (water levels) for the dividing nodes which are called interior boundaries. Water levels at these interior boundaries will be adjusted using a bisection-like iteration procedure until achieving the desired accuracy, i.e., flow discharges obtained for any pair of the interior boundaries which come from the same channel node become approximately equal. Test using data of a real river system proves the model's validity and its superiority in reducing computation time. The required number of iterations is independent from the network'scomplexity, and estimated at only 2 to 3 times more than that for a dendritic network of a similar size. Since each dendritic subnetwork is solved by the double sweep algorithm, the proposed model eliminates any difficultyconcerning the network size and computer capacity.
