DEM Mastery

Project Overview

• Manipulate DEMs from various sources into usable formats

• Gain understanding of the methods required to create DEMs

• Display the same location using DEMs of multiple resolutions

The tutorial for this exercise from the Utah Geospatial Consortium can be found here. Data and sources are listed below.

1. 0.5-m lidar DEM (Utah Geospatial Research Center)

2. 5-m autocorrelated DEM (Utah Geospatial Research Center)

NOTE: 10-m and 30-m 3-DEP DEMs from the USGS National Map were also used as part of this exercise, but were not used to create any of the maps below.

All maps are displayed in NAD 1983 (2011) UTM Zone 12N. 

Types of DEMs

LiDAR

LiDAR DEMs are created by shining a laser at the Earth's surface over a geographic area of interest in a  pattern of evenly-spaced points. The light that is reflected back from the surface is recorded, and from that information elevation data can be derived. 

Autocorrelated

Autocorrelated photogrammetry DEMs are created using multiple aerial photographs taken from different vantage points relative to the Earth's surface. Elevation is estimated based on how the perspective of each photograph differs from the others. While this kind of DEM is relatively easy to create, the vertical accuracy tends to be low compared with other types of DEMs.   

3-DEP

The USGS 3-DEP data set is a compilation of various LiDAR DEMs that have been taken across the contiguous United States. The goal of the project is to provide multiple resolutions of DEMs that are as complete as possible for use by US government agencies and the general public. 

Comparative Map

A higher resolution of this map can be found here.

Conclusions

Figure 1 demonstrates that higher resolution DEMs will generally show more surface detail than lower resolution DEMs, especially on a small-scale. However, one thing that I took away from this exercise was that higher resolution does not necessarily mean "better" or "most accurate." Table 1 shows a comparison of elevations for Signal Mountain in UT according to each DEM. Although the 5m DEM had the second to highest resolution of all four DEMs, it was the farthest off from the elevation shown by the highest resolution DEM. Figure 1 also suggests that the 5m autocorrelated DEM has some serious accuracy issues since Thomson Creek is not even visible in the 5m DEM. Based on this information, if given the choice between using the 5m autocorrelated DEM and the 10m 3-DEP DEM, I would use the 10m 3-DEP DEM despite the loss in resolution.  Though it is generally good to have the highest resolution possible for an analysis using a DEM, it is important to also consider the method that was used to create the DEM and whether that method will have a large impact on the accuracy of that DEM. 

Mastery Observations

• To check that I ran the Mosaic to New Raster tool correctly, I made sure that the metadata of the new raster matched with a single tile of the original raster. I also made sure that the maximum and minimum elevations matched between the tiles and mosaic.

• Zoomed out, the USGS DEMs looked like they were the same resolution. To check that I had not accidentally added two of the same rasters to my map, I created a hillshade of each layer and compared cell sizes. The 10-m DEM clearly had smaller cells than the 30-m DEM when I zoomed in on the hillshade. Additionally, I checked the properties for each reprojected DEM to check the cell size and coordinate system were correct. 

• When I created my contour lines, I guessed that the 30m DEM contours would look blockier than the 10m contours because I assumed the lower resolution would aggressively smooth out details on the landscape. I decided to click through each contour layer to test my guess. The .5m DEM had the most representative contours and was the choppiest since it followed the surface the closest. The following contours "smoothed" out the surface more and more as the resolution decreased.