- Title
- A comparison of the performance of digital elevation model pit filling algorithms for hydrology
- Creator
- Senevirathne, N.; Willgoose, G.
- Relation
- 20th International Congress on Modelling and Simulation (MODSIM2013). Proceedings of the 20th International Congress on Modelling and Simulation (Adelaide, S.A. 2-6 December, 2013) p. 1624-1630
- Relation
- http://www.mssanz.org.au/modsim2013/
- Publisher
- Modelling and Simulation Society of Australia and New Zealand
- Resource Type
- conference paper
- Date
- 2013
- Description
- Digital elevation models (DEM) are widely used in hydrological applications for computing useful topographic parameters such as slope, flow direction, flow accumulation area and stream channel network. However, DEMs generally contain numerous topographic depressions which are real and/or artifactual. These depressions can take the form of single cells (pits) or contiguous areas in DEM (depressions). The problem with these pits and depressions is that they interrupt continuous flow paths in DEMs. To avoid these problems, all pits have to be rectified and create a depressionless DEM before calculating flow directions or any related topographic parameters. Agency provided DEMs may be pit filled, but pits can also be generated while interpolating DEMs for changing grid spacing (e.g. LiDAR data). Therefore pit filling is an essential requirement for any hydrological study. A number of algorithms have been developed over the past few decades to treat pits in DEMs. With the availability of high resolution data, DEMs typically contain millions of cells which increase file sizes and computational effort. Therefore, efficiency of pit filling algorithms has to be taken into account when using such DEMs in hydrological applications. In this paper, two of most widely used pit filling algorithms (Jenson and Domingue, 1988 and Planchon, 2001) are compared in terms of their performance and ability to extract topographical parameters. One arc second DEM derived from Shuttle Radar Topography Mission (SRTM) data was used in the study. Two study areas were used in the comparison. The first study area is comprised of three catchments located in Eastern Australia and they were used for evaluating topographic changes made by pit filling algorithms. The second study area is located in East coast and it was used to compare the performance of the two algorithms across a regional extent. According to the results, both algorithms behave similarly in modifying existing topography, but calculated flow accumulations and drainage networks were slightly different from each other. When the filled area was relatively small, both algorithms have resulted in similar flow paths. But in relatively large filled areas, they have resulted in unrealistic parallel flow paths significantly different from each other. Both algorithms were implemented in the Python programming language to provide a common platform for comparison. Python is an interpreted language and the Cython tool has been recently developed to convert Python code to C code and allow it to be compiled. Cython was used to convert Python code to C extensions and the performance of both Python and Cython versions were evaluated. Time taken to execute pit filling algorithms on different sizes of DEMs was measured. The execution time of the Planchon algorithm showed a linear relationship with the size of the DEM while execution time of Jenson algorithm increased exponentially. Moreover, performance of both algorithms was evaluated for different resolutions, on a constant grid extent. The execution time of the Jenson algorithm increased with DEM resolution and showed a direct relationship to the total number of pits. However, execution time of Planchon algorithm remained almost constant regardless of the total number of pits. In Cython, both Jenson and Planchon algorithms showed significant improvement in execution time, relative to implementation in Python.
- Subject
- pit filling; digital elevation model; python; cython
- Identifier
- http://hdl.handle.net/1959.13/1308698
- Identifier
- uon:21704
- Identifier
- ISBN:9780987214331
- Language
- eng
- Reviewed
- Hits: 1856
- Visitors: 1832
- Downloads: 0
Thumbnail | File | Description | Size | Format |
---|