Spatial Technologies--Terrain Modeling in a 3D World1 Apr, 2006 By: James L. Sipes
Using DEMs and DTMs to Represent 3D Landforms and Surfaces.
The ability to represent landforms in both 2D and 3D is beneficial for any professional who works with the land. 3D terrain modeling depicts landforms not only on the different continents but also ocean floors and the surfaces of other planets.
DTM & DEM
DTMs (digital terrain models), also called DEMs (digital elevation models), are computer models of the surface of the Earth. DTM and DEM typically are used interchangeably. Technically, a DTM refers specifically to a terrain model, and DEM refers to the surface of water, vegetation or another feature.
Figure 1. A DEM file is a gridded pattern of cells, with each cell be assigned a specific elevation.
DEMs include spatial elevation data used to depict 3D landforms. A DEM file consists of a gridded pattern of cells with each cell assigned a specific elevation. The finer the grid, the more detailed the resulting landform model. A grid system is the most common way to represent surfaces because it is the simplest way to store elevation points. As a result, a grid format can be used in many different applications.
DEMs are available from government agencies such as USGS (U.S. Geologic Survey), EDC (USGS EROS data center), NIMA (National Imagery and Mapping Agency) and NGDC (National Geophysical Data Center; figure 1). Most states have some type of GIS clearinghouse, which typically includes digital terrain data. One source I visit frequently is GIS Data Depot (www.geocomm.com), a commercial site that offers a lot of data for free, including DEM models.
A common format for DEM data is the NED (national elevation dataset). NED is a seamless coverage of DEM data with a cell size of 30 meters, and it includes all of the continental United States. Users can access the data on the USGS seamless data viewer for free, as long as they limit their downloads to less than 10MB.
Other sources focus on specific types of DEM data. Bathymetry is the mapping of undersea terrain. GEBCO (general bathymetric chart of the oceans), which provides bathymetry data sets for the world's oceans, is available on CD-ROM from the NGDC. Unlike the terrain of dry land, undersea areas typically aren't mapped using regular grid points but instead have more points near land masses and fewer points farther away (figure 2).
Figure 2. A representation of the Earth made with the global IHO/IOC GEBCO One-Minute Grid. The 3D visualization was made by Martin Jakobsson at the Center for Coastal and Ocean Mapping/Joint Hydrographic Center, University of New Hampshire. Image courtesy NGDC.
TINS and Traditional Models
DEMs are not the only way to represent 3D terrains. A TIN (triangular irregular network) analyzes elevation points and connects them to create a series of triangular facets. These facets can define a particular surface. This approach produces a more accurate representation of a terrain surface than other commonly used techniques. A problem with TINs is that they can oversimplify the shape of the land and indicate that areas are flat when they really aren't. The best way to address this deficiency is to add breaklines and spot elevations that provide an additional level of detail and control. TINs are more difficult to work with, and some GIS programs cannot work with a TIN.
Most CAD programs and 3D visualization programs let users create terrain models from contours, but some do it better than others. For example, I am a big fan of SketchUp, but the program does not work very well with large terrain files. Autodesk's Map 3D and Autodesk Land Desktop, on the other hand, handle virtually any task that involves creating or modifying landforms.
Figure 3. Plan view of a pseudo-color map of Lake Tahoe, California. Image courtesy of Satellite Imaging Corp. (SIC).
Uses of DEMs
DEMs are helpful in a wide range of applications. Terrain modeling has become integral to hydrology, tectonics, oceanography, climatology and geohazard assessment. Resource managers are concerned about how water moves through a watershed. The flow of water is determined by comparing the elevation of one cell in a DEM grid with that of neighboring cells. Programs such as RiverCAD XP use DEMs to create a 3D watershed model that can be used for this type of analysis.
Terrain modeling is used extensively in many design, planning and engineering applications. Regional planners need to understand landforms to comprehend potential growth patterns. Roads traditionally follow ridges and valleys, and transportation engineers continue that practice because it is the easiest way to build roads that are not too steep.
DEMs are useful for making geospatial information more readable. Hill shading, for example, makes terrain information much easier to understand than simply using contours.
In ArcGIS's Spatial Analyst, the default sun azimuth is 315° northwest. This setting produces a shadow pattern similar to what you often get with overhead lights, and the resulting map visually matches what a user would expect. The basic idea is to produce a readable map, not a hillshade pattern, similar to what you would find in real life. In virtually all plan and map production, shades and shadows are on the bottom because the resulting images are easy to read.
The lighting used in shaded relief models darkens one side of terrain features, and the darker the shading, the steeper the slope for a particular area. Shaded relief models frequently are rendered using a color ramp with cool colors for lower elevations and warm colors for higher elevations (figure 3). I seldom use the default color ramps offered by GIS programs, though, because they do not always result in the kind of graphic representation I'm seeking. For example, one color ramp I frequently use works well in Colorado, but applying it to a terrain model in Alabama mistakenly gives the impression that all of the peaks in the state are covered with snow, because white is used for the highest elevations.
Tools for DEMs
Many tools exist for working with DEMs. Virtually every GIS program lets users import and view DEM data. I work most frequently with ESRI's Spatial Analyst and 3D Analyst.
For users with limited budgets who want to access DEM files, 3DEM and MICRODEM are two programs worth noting. 3DEM is a freeware program from Microcomputer Topography (www.visualizationsoftware.com/3dem.html) for viewing and converting DEM files. MICRODEM is a microcomputer mapping program developed by the U.S. Naval Academy that displays and merges digital elevation models, satellite imagery, scanned maps, vector map data and GIS databases. The program is simple, easy to use and offered as freeware (www.usna.edu/Users/oceano/pguth/website/microdem.htm).
Figure 4. 3D terrain model of Mt. St. Helens in Washington state. Image courtesy of the Jet Propulsion Laboratory.
Programs such as MultiGen-Paradigm's SiteBuilder 3D, a powerful GIS-based modeling tool, integrate DEMs with other 3D models. This software also is used in CommunityViz, a popular program used by community planners. Specialty programs such as Bentley's GEOPAK Road, GEOPAK Site and GeoTerrain are particularly good at shaping and editing DEMs and terrain for highway design and other transportation projects. All three are add-ons to MicroStation.
Landscape visualization programs such as Bryce, WorldBuilder and World Construction Set use DEMs to create dazzling representations of the natural environment.
Traditional 3D modeling software tools such as 3D Studio MAX and Maya are very good at visualizing terrains, but they typically use TINs and the resulting file size can be large and cumbersome. These programs also lack the analytical capability of GIS programs.
Some programs are geared specifically to model caves and overhangs (figure 4). As you can imagine, these features are much more difficult to model than traditional landforms. A couple of years ago, I worked on creating 3D models for a mining company in Washington State. Our objective was to create models of the existing tunnels and then explore alternatives for creating new tunnels to connect with the existing ones. The only tools available for this type of project were traditional 3D modeling tools, so we used 3D Studio Max.
Figure 5. 3D surface of the Big Foot Cave produced with WinKarst software. Image courtesy of Resurgent Software.
Modeling a cave is difficult because spaces must be defined as volumes instead of the surfaces used for traditional landforms. Another problem is that caves can have deep shafts, irregular passageways, stalactites, stalagmites and tight spaces. WinKarst cave mapping software (www.resurgentsoftware.com/winkarst.html) is shareware for Windows that is designed to ease the visualization of cave passages (figure 5). WinKarst supports the whole process of surveying caves—from sketching and describing to digitizing maps to creating 3D visualizations of caves. One cool capability of WinKarst is that it generates Track and Waypoint export files of cave surveys that can be uploaded to a GPS unit. This ability helps spelunkers find their way in even the deepest, darkest caves.
Concerns when Working with DEMs
Using DEM data certainly has limitations. Errors can occur when terrain models are created from sources such as aerial photographs, contour maps and field surveys. Even small errors in height can cause significant errors when analyzing the surface of a terrain. Pits and depressions can result from data error, and small stream channels or abrupt changes in landforms sometimes don't appear because the DEM grid is too large.
In this article
The files users download from many sites, including USGS sites, typically are not projected. Instead, they use longitude and latitude for the x and y coordinates. Many GIS programs can modify the projections for data on the fly, but they don't always do it well. I always try to change the projections of the DEM before actually using the data. This process takes a little more time up front, but it minimizes potential problems and helps ensure that my data fits together the way it should. If you do not project your data, your final products may have errors or discrepancies. For example, hillshade maps may have high contrast shadows or lines of noise might appear across the map.
One mistake that many people make is to overestimate the quality and accuracy of DEM models. Accuracy when modeling terrain is more critical for some applications than others, and not all DEMs are created the same. Users can check metadata to determine how a DTM was created and the level of accuracy it should have.
But as long as users check the accuracy of the DEM data they use and make sure it is projected to match other data being used for a particular project, DEMs will continue to be among the most widely used types of geospatial data. In this 3D world of ours, that is probably the way it should be.
James L. Sipes is the founding principal of Sand County Studios in Seattle, Washington, and senior associate with EDAW in Atlanta, Georgia. E-mail him at firstname.lastname@example.org.