Authors: Leo Meirose*, University of South Florida, Christopher Brown, Worcester Polytechnic Institute, Worcester, MA, Barnali Dixon, University of South Florida, Saint Petersburg , Alvan Karlin, Southwest Florida Water Management District (SWFWMD)
Topics: Geographic Information Science and Systems, Environmental Science, Water Resources and Hydrology
Keywords: DEM resolution, resampling, stream delineation, fractal analyses, statistics, spatial location analysis
Session Type: Paper
Start / End Time: 5:00 PM / 6:40 PM
Room: Buchanan, Marriott, Mezzanine Level
Presentation File: No File Uploaded
Digital elevation model (DEM) resolution and type constitute a critical variable in hydrological modeling. Changes in DEM resolution and algorithm type will impact stream delineation, which will ultimately impact stream spatial location, fractal properties and statistical results. These changes may manifest differently in flat or mountainous terrains. Because of this, there is a need to analyze i.) how stream delineation changes based on DEM resolution, ii.) how stream delineation changes based on terrain and iii.) how the fractal geometry of the delineated streams change. Since the spatial location of streams matter (to answer the title question “Next to or Through Your House?”), it is necessary to evaluate the application of spatial location analysis of streams versus traditional statistical tests at different resolutions of DEMs. A comprehensive sensitivity analysis using an axiomatic classification to characterize geospatial and statistical changes in delineated streams can provide insights into effects of varying the resolution in stream delineation and its spatial and statistical properties (relative area, complexity and F-Test). Results of DEMs with original and resampled 1.5, 10, 30 and 90m resolutions were compared with one another to determine spatial and statistical accuracy. Results showed that in both flat and mountainous terrains, delineated streams of DEMs having identical cell size provided comparable statistical results, but displayed different spatial locations. This suggests that applying a statistical index of complexity (such as a fractal dimension) to compare spatial changes in DEM surfaces and their delineated streams may prove problematic because the statistical results may not translate spatially.