The Perils of 3D Modeling

Know what to watch out for -- and when to stop -- when using these powerful tools.

Over the past few months, this column has examined various new technologies for civil engineering and geographic information system (GIS) applications. This month, I'd like to step back and take a different perspective -- one that identifies some pitfalls of applying today’s high-powered technology. In particular, I'll focus on digital 3D modeling, a powerful tool for civil engineering, surveying, and land planning professionals, but one that can create problems when misunderstood or misused.

Digital 3D modeling has vastly expanded the capabilities of land and transportation designers in recent years, enabling faster design, visualization, and data retrieval, plus numerous other benefits. With an accurate digital terrain model (DTM) users can readily determine x, y, and z coordinates virtually anywhere on a surface, whether it's an existing ground surface or a proposed grading surface. But based on what I've seen in numerous e-mails, posted on discussion groups, and through personal experience, 3D modeling is no panacea. Some of the common problems include inaccurate modeling, excessive modeling, and inappropriate combination of multiple data sources. There are probably others, but let’s start with these.

Inaccurate Models
While a perfect DTM is often not attainable, the accuracy of the model must be properly matched with the task at hand. For example, GIS-level topographic mapping may be totally adequate for identifying overall drainage patterns in a community, but inappropriate for final design of a roadway. End users must know the source and accuracy of their data.

Inaccurate DTMs can also be built unwittingly, even with the best intentions. An existing ground model may be developed for final design purposes, but insufficient data point density may preclude the model from accurately depicting the ground surface. This can occur with both ground- and aerial-based surveys. Accuracy of existing-ground models can also be diminished by improperly identified breaklines -- the boundaries that delineate ridges, valleys, and other areas where surfaces change markedly. Even with high data point density, inadequate breaklines will generally cause modeling software to interpolate across those boundaries and misrepresent actual conditions.

Without proper breaklines, DTM software will interpolate across features such as curb lines.

Proposed DTMs can suffer some of the same ills as existing DTMs, although designers, rather than surveying/mapping professionals, are typically in control. Accounting for the design software capabilities, the designer must apply sound judgment in deciding how to build the model, based on how it will be used. Is it just for generating construction plan data (i.e., alignments, profiles, and cross-sections), or will it be used to develop 3D renderings and/or animations? Will the project be staked using conventional ground surveys, or will machine-control techniques (see "GPS-Guided Earthmoving Gains Ground," Jan. 2, 2007) be used? As the potential uses of design data grow beyond plan preparation, the accuracy of modeling techniques must generally increase accordingly.

A slight variation of inaccurate modeling dilemma is the over-reliance on DTM data during design. For example, consider a street reconstruction project with an existing ground DTM based on field survey shots collected every 50 feet. Now assume the reconstructed roadway meets the existing roadway within a relatively sharp vertical curve. If a survey shot was not taken at the exact tie-in point and the designer establishes a starting elevation based strictly on a DTM, the proposed elevation could miss the mark by several inches. A more thorough approach would examine the existing roadway surface in a profile view and establish a starting elevation based on a series of surrounding data points. And ideally, a field survey shot should be taken at the tie-in point.

Click on image to enlarge. Profile tie-ins can be inaccurate if based strictly on DTMs.

Overmodeling
Surprisingly to some, problems also can occur with excessive modeling. For example, if half of a project budget is drained by creation of an existing ground DTM, a design team would be hard pressed to meet the overall project budget. Also, sometimes an excess of data points can bog down the modeling process, although most modeling software allows for “weeding out” excess points.

On the flip side, a designer might use all the bells and whistles in the latest software to create complex surfaces for modeling proposed grading schemes, when in reality, simple profiles at key locations might suffice, or even be more effective in certain situations. For example, design software might enable display of contours at a 0.1-ft interval in a street intersection and imply adequate slopes for drainage, but if the actual slope along a curb return is less than 0.5%, water will likely pond in the gutter. Similarly, building complex models for a parking lot sloping generally in one direction can consume significant time of designers, surveyors, and constructors. As one designer emailed me recently: “We are grading the earth, not creating a Rolex watch.”

Multiple Data Sources and Disciplines
Another problematic scenario occurs when data are provided from multiple sources. A planning-level topographic map supplemented with ground survey shots might work fine in some cases, but not in others, because the accuracy of the planning level map is generally lower than that of the survey shots. Worse yet, aerial and ground surveys might be based on different control points, leading to confusion and discrepancies.

Yet another issue occurs when data sets are based on different coordinate systems. Planimetric data could be based on state-plane coordinates, while topographic data could be based on another system. Each might have its own conversion factor to account for curvature of the Earth. Entire columns could be devoted to these scenarios, and most surveying textbooks provide an overview of this topic.

Power in the Right Hands
I write this not to disparage digital 3D modeling. It’s generally a wonderful tool that produces valuable results. Most digital mapping obtained from reputable mapping firms is accurate within the stated limits and is generally free of errors. And knowledgeable users of today’s civil/land software can produce fine DTMs of both existing and proposed surfaces. But, as with any tool, users need to be cognizant of software limitations and their own skill sets and consult experts when needed. Engineers, surveyors, and GIS professionals all bring different knowledge to the table, and ideally all should be involved in planning and development of DTMs. I purposely did not identify specific software products in this column because many products are fully capable of producing high-quality models. The main burden is on us -- end users -- to properly understand how the tools work and properly apply them.

I’m interested in hearing more experiences with 3D modeling and other technology. If you have a success story or have learned valuable lessons through problems and challenges, please drop me an email.