Application of Linear Programming in DEM Pit Removal
Location
Space Dynamics Laboratory
Event Website
http://water.usu.edu/
Start Date
3-26-2004 9:30 AM
End Date
3-26-2004 9:45 AM
Description
Digital elevation models play a large role in the modeling of hydrological flow processes driven by topology. DEM's commonly contain spurious sinks that are artifacts of the data rather than true representations of the fluvial terrain. It is sometimes desirable to remove these prior to use in analysis of hydrological flow processes. The most common method of pit removal involves filling the pits to the elevation of the outlet or pour point of the bounded region containing the pit. A new method for pit removal is proposed that poses the solution as the local elevation changes necessary for pit removal that results in a minimum distortion of the original data. This problem is solved through linear programming. The dimensionality of the problem is limited by localizing the problem to the region draining to each pit. In order to ensure a unique local solution a carving solution is used to impose a flow field with a corresponding set of unique constraints for the linear program. The resulting changes to the elevation grid, though not guaranteed to be a global minimum, are smaller than those from traditional filling techniques or carving algorithms alone.
Application of Linear Programming in DEM Pit Removal
Space Dynamics Laboratory
Digital elevation models play a large role in the modeling of hydrological flow processes driven by topology. DEM's commonly contain spurious sinks that are artifacts of the data rather than true representations of the fluvial terrain. It is sometimes desirable to remove these prior to use in analysis of hydrological flow processes. The most common method of pit removal involves filling the pits to the elevation of the outlet or pour point of the bounded region containing the pit. A new method for pit removal is proposed that poses the solution as the local elevation changes necessary for pit removal that results in a minimum distortion of the original data. This problem is solved through linear programming. The dimensionality of the problem is limited by localizing the problem to the region draining to each pit. In order to ensure a unique local solution a carving solution is used to impose a flow field with a corresponding set of unique constraints for the linear program. The resulting changes to the elevation grid, though not guaranteed to be a global minimum, are smaller than those from traditional filling techniques or carving algorithms alone.
https://digitalcommons.usu.edu/runoff/2004/AllAbstracts/30