A Fast Priority-Flood Algorithm with Pruning for Depression Filling in Hydrologic Analysis Ruchi Lawaniya, Manish Pandey Maulana Azad National Institute of Technology, Bhopal, 462003, India Abstract— The Digital elevation models are widely used spatial data source to incorporate the topographic information within geographical and hydrological applications. Depressions in DEM are lower areas surrounded by surface without any outlet. They interrupt or disconnect the flow path and create inaccurate drainage pattern. Subsequently, recognizing and removing the depression is a vital necessity for any hydrological study, which is commonly done prior to the use of DEM to conduct the hydrologic analysis. Usually, handling the depressions is a time consuming task for applications of huge terrain dataset with high resolution. This paper presents an improvement on priority-flood algorithm for recognizing and processing the depressions based on gridded digital elevation model in digital terrain analysis. The improvement on previous method is done by introducing a novel concept of pruning the dead cells from the priority queue. The priority queue cells that will never be used for further computation are considered as dead cells. Pruning of the dead cells can reduce the number of cells in the priority queue. Thus, the overall running time of Insertion and Deletion operations within the priority queue is asymptotically decreased. The proposed PriorityFloodPruning algorithm runs in     time, where K is the number of cells present in the priority queue after pruning. The proposed PriorityFloodPruning algorithm shows 1.13x to1.25x speedup. Keywords - Digital Elevation Model, depression filling, priority-flood, hydrologic analysis I. INTRODUCTION The DEMs are digital representation of altitude of continuous ground surface in raster form. They are stored as a rectangular array where value of each cell represents a landscape height regarding any reference datum. The DEMs are widely used to derive relief features such as gradient and slope direction, compute flow simulation path, drainage pattern, watershed features and sub-catchment areas in hydrology. The automatic derivation of drainage network over a DEM is an essential requirement of many applications of GIS such as hydrologic analysis, pollution diffusion analysis and land erosion etc. To automate extraction of the fully connected drainage pattern in the raster environment, the flow path from every cell must be directed to an outlet on the border of DEM. But, the presence of spurious depressions in DEM interrupts and disconnects the flow path. The depressions are the lower elevation areas without outlet, they consist one or more connected cells of equal height which are surrounded by higher elevation cells of topographic surface. In the hydrological and geomorphic applications, identification and removal of the depressions is completed prior to the use of the DEM. In the hydrology community, there has been a significant measure of innovative work that concentrates on in utilizing the raster environment for handling the depressions and extracting the flow path such as elevation- smoothing method [14], pit-filling algorithms [3], carving method [17], breaching method [11], [8] and hybrid algorithm combining procedures of carving and depression filling [18]. This paper proposed an improvement on the priority flood algorithm by introducing a new concept of pruning the dead cells from the priority queue. The DEM cells, whose neighboring cells are already traversed or processed, are considered as dead cells. As the computation of least-cost search progresses for flow path, the number of dead cells is increased. From these dead cells, new cells can’t be grown. So there is no longer need to remain in priority queue. The removal of these cells decreases the number of cells in priority queue. The memory consumption and computation time required for insert, find minimum, delete minimum operations of priority queue is decreases. Thus, the computation time of PriorityFloodPruning algorithm    , where K is the less than the total number of cells in DEM. International Journal of Computer Science and Information Security (IJCSIS), Vol. 14, No. 7, July 2016 694 https://sites.google.com/site/ijcsis/ ISSN 1947-5500