Labs‎ > ‎

Lab 06: Working with DEMs

Introduction

Lab 06 in Advanced GIS and Spatial Analysis (WATS 6920) was an opportunity to download, construct, manipulate and display a type of raster data known as Digital Elevation Models (DEMs).  Additionally, we examined the trade-offs that occur with increasing resolution.

Objectives

As described by Dr. Wheaton, our objectives were to:

Have acquired and worked with DEMs in common formats and from typical data sources

Build an appreciation of the tradeoffs between DEMs at different resolutions and built from different data sources using different methods

Background

As an employee for the hypothetical Big Cheese Watershed Management consulting firm, I have been tasked with providing the Cottonwood Canyon Foundation with a graphical presentation of the cost-benefits associated with increasing resolution in DEM datasets so that they might decide which scale is most appropriate for their stewardship needs.  Below please find: 1) Context map displaying a 30 meter DEM of Cottonwood Canyon in its entirety relative to Salt Lake City, Utah; 2) Comparison of 5, 10 and 30 meter DEM maps at the watershed scale, and our recommendation; 3) Comparison of 2, 5, and 10 meter DEM maps at the 1:20000 meter scale, and our recommendation; 4) An image of an ArcScene Map with a 30 meter DEM; 5) Comparison of 2, 5, and 10 meter DEM maps relative to the highest known peak at Solitude Ski Resort, and our recommendation; 5) Comparison of 2, 5, 10 and 30 meter DEM maps displaying differences in 12 meter (40 ft) contours; 6) Discussions outlining the rationale in our recommendations. 

Task 1 Products

Context Map

Click on the image below to enlarge to full screen:


To download a PDF of this Context Map click here.

DEM Comparison: Watershed Scale

Click on the image below to enlarge to full screen:


To download a PDF of this Watershed Scale Comparison click here.

DEM Comparison: 1:20,000 Meter Scale

Click on the image below to enlarge to full screen:


To download a PDF of this 1:20,000 m Scale Comparison click here.

ArcScene Map

Click on the image below to enlarge to full screen:


Task 1 Discussion

Origin of Data:

2 meter LiDaR – Light Detection and Ranging is a remote sensing application that calculates distances by illuminating target with pulses of light (often laser).
5 meter auto-correlated – This data set is generated by approximating values to the 5 meter scale from nearby features or similar values within dataset.
10 and 30 meter DEMs – these datasets are open-source, provided by the National Elevation Dataset.

Resolution Differences:

As the name implies, 2 meter LiDaR, is composed of cells that equate to 2 square meters, our highest resolution in this comparison.  Our resolution becomes gradually coarser in these analyses, as 5, 10, and 30 meter DEMs contain pixels that are 5, 10, and 30 square meters, respectively.  Measurements assume a flat surface, i.e. they are not draped over the variable topography.

Error/Uncertainty:

With the exception of the now available 1 meter LiDaR (2011), these 2 meter DEMs obviously have some of the best reductions in error and uncertainty associated with their use.  One should be aware that water surfaces can refract light pulses, distorting distances and introducing error.  However these properties can be modeled and LiDar is also used in bathymetry.  An issue that we raised in our 5 meter DEM is the irregularity that is introduced by vegetation – upon close analysis, one can see how large trees along pixel edges distort the imagery.  At the 10 meter scale, interpolation (averaging across the 10 meter footprint) begins to become an issue, and the imagery takes on a pixelated look; this is especially pronounced at the 30 meter scale (which is really only appropriate at the regional scale because of the display resulting from interpolation).

Resolvable Surface Features:

The 2 meter LiDaR resolves amazingly minute features, small rivulets, shallow depressions, the edge of ridges.  The 5 meter DEM captures much of this as well, but in the area I focused on, the resolvable features so prevalent at the 2 meter scale were often obscured by the “lumpy” vegetation.  The 10 meter smooths the vegetation interference out, but many of the small-scale features now disappear – particularly the diminutive shaping of the land by water.

File Sizes:

Dataset

Coverage Area

File Size

2 meter LiDaR

2000 x 2000 meter block

~10 mb

5 meter DEM

20,000 x 20,000 meter block

~125 mb

10 meter DEM

County

~200 mb

30 meter DEM

County

~20 mb

Data Availability within Utah:

The 2 meter coverage is very limited in Utah, to primarily Salt Lake and surrounding areas.  However, the 5, 10 and 30 meter DEMs are readily available and open source at Utah AGRC.

Task 2 Products

DEM Comparison: Highest Elevation Cell

Click on the image below to enlarge to full screen:

To download a PDF of this DEM Peak Comparison click here.

DEM Comparison: Contour Lines

Click on the image below to enlarge to full screen:


To download a PDF of this DEM Contour Line Comparison click here.

Task 2 Discussion

We see a shift in the pixel with the highest elevation relative to the known peak coordinates as the resolution becomes coarser (see DEM Comparison: Highest Elevation Cell above) because the elevation represents a mean value across the entire pixel.  Moving away from the peak, and encompassing more lower elevation values within the increasing pixel sizes reduces and spatially shifts the cell with the highest elevation.  The contour lines simply run between cells that are within the same range of elevations (we set contours for 12.192 meters); again as we increase pixel size there will be slight shifts in cell values and the groupings of cells by contour lines.  This is especially evident with small peaks (see DEM Comparison: Differences in Contour Lines above) which disappear all together, as those cells are now lumped into group of cells below.

We hope this has provided some clarity for you to make your decision based upon your needs and objectives.

Comments