Lab 6: Working with DEMs

View from the mouth of Big Cottonwood Canyon looking upstream to the east. Click to enlarge

The purpose of Lab 6 was to give an introduction into building, manipulating and displaying Digital Elevation Models (DEMs), with the end goal of establishing proficiency when it comes to working with DEM datasets. To accomplish these goals, the lab gives a background into acquiring DEM data for different data sources, creating raster sets from multiple data types and editing those rasters. Working with multiple DEMs gave me an understanding of the advantages and disadvantages that come at different resolutions. This lab was insightful and informative, providing solid information, but a large part of it was a cartographic exercise relative to analysis, an outcome which frustrated me.

Context for Map Analysis and Products

This project (Note: a fictional project) was undertaken to help the Cottonwood Canyons Foundation with identifying areas of advocacy and public outreach. Big Cottonwood Canyon is one of several watersheds that feed into the SLC area, a map of them can be seen on Salt Lake City's Protected Watersheds page.

Task 1: Show All the DEMs

Click to enlarge

Map shows comparison of the 4 DEM datasets that were compared for use in Big Cottonwood Canyon. The coarseness of the 30 meter DEM is apparent even at the water shed scale. For watershed scale analysis, I would recommend the using the 10 meter DEM. The 2 meter could be used depending on the level of analysis and computational power. A 2 meter DEM may provide more accurate information, but it is a substantially large set of files. The 5 meter dataset looks to have false precision based upon a visual inspection of the image.

Table 1: comparison of DEM datasets

DEM Resolution Data Source Error/Uncertainty Download (For Utah only)
30m NED USGS National Elevation Dataset (NED) Uses the most current standards for geospatial data and conforms to standards of the USGS's National Mapping Program. 30 meter dataset is upscaled from the 10 NED rasters, so error in that dataset is propagated to 30 meter NED. 30 Meter Elevation Models
10m NED USGS National Elevation Dataset (NED) Derived from USGS 7.5" Topographic Map contour data. Since at its root, it comes from paper maps, errors in the paper map along with the multi-step process of digitizing maps, creating contours and converting contours to DEM can propagate to the final 10 meter DEM. Uses the most current standards for geospatial data and conforms to standards of the USGS's National Mapping Program. 10 Meter Elevation Models
5m Auto Correlated DEM Utah Automated Geographic Reference Center (AGRC) DEMs created from imagery collected during the 2006 NAIP and HRO aerial photography flights. Auto-correlation process is not as rigorous and anomalies are common within the dataset. In comparison to USGS NED data, this DEM provides a higher degree of resolution and horizontal accuracy. Utah AGRC 5 Meter Auto Correlated DEM
2m Lidar Utah Automated Geographic Reference Center (AGRC) Created from aerial Lidar data - specifications do not appear to adhere to USGS Lidar Base Specification. Uncertainty in this data comes from processing of data to remove vegetation and produce a 2D bare earth surface 2 Meter Lidar

Though I preferred the 10 meter DEM at the watershed scale, it may not be the same preference at a finer scale. For assess each DEM set at a closer in resolution, I looked at each DEM in the area of Solitude Mountain Resort.

Click to enlarge

The fineness of the 2 meter really shows up here in the close in map. Roads and a minor topographical changes can be easily seen. The 5 meter DEM still suffers from false precision and the 30 meter is too coarse to be useful. The 10 meter DEM, while blurry, may be more useful, since its resolution excludes roads and other man-made structures. The decision between the 2 and 10 meter sets comes down to complexity of analyses, computational power and time pressures. If none of those issues are major pressures, then the 2 meter data may be the superior choice. 

Task 2: Highlight Differences Between the DEMs

While I prefer the 10 meter DEM data for analysis, it may not be the most accurate data set relative to other data types. When you compare the contours extracted from DEMs to a USGS 7.5" topographic map, values for the two data types diverge. 

Click to enlarge

The 10 meter contours reflect the topo map most accurately, an expected outcome since both data types come from the same source data. The 5 meter contours appear to suffer from false accuracy due to anomalies in the DEM. Compared to the topo map, 2 meter data may be reflecting the "true slope" to a higher degree. If minor changes in slope are of the highest importance, 2 meter data may be the direction to go in. If comparison to older, national USGS topographic maps is important then the 10 meter DEM will better reflect national data sets.

One other issue to be aware in using DEMs of different resolution is how each set records and stores elevation. DEMs of larger resolution have a single elevation value per cell and when the cell covers a peak, a value other than the peak is often recorded. An example of this phenomenon is shown below.

Click to enlarge

The 2 and 10 meter sites report the same peak elevation, but the peak elevation for 10 meters is between 10-20 meters from the actual peak. In this situation the 5 meter data is pretty accurate, with a peak elevation 2 meters different from actual and a distance 5-10 meters from the actual peak. The unsuitability of the 30 meter data is further confirmed since it is neither accurate with location (20-50 meters off) and elevation (8 meters from actual peak elevation). 

Overall, after looking at all the available DEMs and assessing their usefulness, I would recommend that the 10 meter DEM be used. I feel that it accurately reflects the actual land surface, providing a balance between proper representation of the watershed and file size. 

Printable High Quality Maps (PDF file format)

Watershed scale DEM comparison

Close up DEM comparison

Contour Line Comparison

DEM Peak Elevation Comparison


Joe Wheaton & Shannon Belmont Lab 06 - Working with DEMs. 2015. Web. 26 February 2015.

Joe Wheaton & Shannon Belmont Lab 06 - Working with DEMs - Task 1. 2015. Web. 26 February 2015.

Joe Wheaton & Shannon Belmont Lab 06 - Working with DEMs - Task 2. 2015. Web. 26 February 2015.

NED Frequently Asked Questions. 2015. Web. 26 February 2015.

Standards used by the National Elevation Dataset. 2015. Web. 26 February 2015.

Eelvations and Terrain Datasets for Utah. 2015. Web. 26 February 2015.

The table in this assignment was created using the HTML Table Generator

The idea for formatting the watershed scale comparison map came from Sammy Lyster