3D Geospatial Data

The usage of three dimensional data in the geospatial industry is in its infancy. It makes sense to me. Sometimes, it’s hard enough for folks to obtain and maintain accurate two dimensional data, not to mention elevation! However, as geospatial technology continues to evolve, the availability of 3D geospatial data will evolve. I’m pretty sure that in ten years we will look back and be amazed at how little we used 3D geospatial data.

But for now, what the heck are Mean Sea Level, ellipsoidal height, orthometric height, geoid height?

Sources of accurate elevation data are difficult to find. Typically, you’re going to find elevation data from aerial photogrammetry projects, LiDAR missions or from GPS data collection projects. Since availability of this sort of data on the world-wide web isn’t as prevalent as 2D geospatial data, 3D geospatial data utilization isn’t main stream yet.

There’s also the issue of the definition of elevation. Yes, just like there are differential horizontal datums, there are a variety of elevation datums. On legacy paper maps, elevations are typically displayed with respect to Mean Sea Level (MSL). MSL is an the elevation reference for local areas, but the Earth is not like a bathtub where gravity has an equal impact on the water in the bathtub that forms a smooth surface. MSL around the world varies tremendously. 2 meters MSL in New York is orders of magnitude different than 2 meters MSL in Hong Kong.

MSL is a complicated subject in itself. Check out this web page on the National Geodetic Survey’s web site that provides definitions related to MSL. The Earth is not a perfect sphere and gravity influences vary by region. For centuries until recently, elevations were stated with respect to sea level because that was the most reliable and widely known reference.

GPS has changed that. GPS uses an elevation model called the geoid which was intended to somewhat approximates MSL. There are a couple of good references that provide much more detail. They are worth reading. One is from ESRI written back in 2003 by Witold Fraczek. The other is from an NGS presentation given in 2007 by Daniel Roman.

In fact, following are a couple of graphics from the NGS as well as one from Dr. Roman’s presentation that draws a clear picture of how GPS heights are related to MSL.

H = Orthometric height (Mean Sea Level), h = Ellipsoidal height, N = Geoid height

Note that the height determined by GPS is the ellipsoidal height, not Mean Sea level. The difference between the two can be tens of meters.

Most GPS receivers have a rough model of the Geoid height built into it. However, it’s very rough and can be a few meters in error. To resolve this, significant efforts have been made in the two decades to create high resolution geoid models. Creating a high resolution geoid model (for a country) is a relatively large effort that requires very skilled people and specific equipment.

Following is a similar graphic illustrating North American Datum of 1983 and GEOID03, which was the most recent geoid model of the United States (GEOID09 was just released).

Finally, following is a graphic from Dr. Fraczek that depicts the relationship between the ellipsoid, MSL and the Earth’s surface. You can see here that at some points, the ellipsoid is actually above the geoid and at some points, it’s below the geoid.

The purpose of this column is to point out that when you receive 3D geospatial data, you should inquire what about the elevation data is referenced to. Are they ellipsoidal elevations? Are they MSL elevations? If MSL, what was the resolution of the geoid model used?

Flushing out horizontal datum inconsistencies in your GIS is, for the most part, pretty straight-forward. The 2D view is the norm and once you bring data into your GIS, you can compare the imported features to the existing features and identify fairly quickly if there’s a problem with the 2D data. The problem is that most GIS folks aren’t used to working in a 3D world. I speculate that most people figure that if the 2D data is reasonable, then the elevation (if it exists in the database at all) must be accurate. It would be interesting to hear from folks who are making a concerted effort in quality checking the heights used in their GIS.

Even though GIS horizontal data is still far from perfect with respect to accuracy, at least I can see the road to success. The quality of horizontal data in the past ten years has improved significantly thanks to widespread availability of data collected via remote sensing and GPS data. I think that trend will continue as the widespread availability of accurate horizontal data continues to improve. The roadmap for 3D data isn’t so clear. Not only is there a lack of accurate 3D data, but also the models (eg. geoid model) for generating accurate 3D data continue to evolve.

Applications for 3D data are expanding and are going to continue to expand. People, both inside the geospatial industry as well as the general public, still have a hard time visualizing 3D data. For example, a land development plan for a site can be communicated much more effectively if there’s a 3D visualization (either still image or animated video) that accompanies the engineering drawings. Following is a visualization of a particular golf course hole where the architect was trying to convey the design change to the golf course owner. The image on the top is the existing golf course. The image on the bottom is the proposed design. The data used to create the terrain model in these images was high quality 3D geospatial data.

Thanks and see you next week.