Every so often I get a question from someone that makes me realize I haven’t done a good enough job of describing what’s going on in the lidar distribution system of Digital Coast and why we do certain things that might appear crazy. The most recent one involved a data set that is about five years old and a user was trying to fill some gaps in their copies of the original data. When they downloaded the data from Digital Coast, they were showing just over half a foot vertical difference from the originals, though the points were horizontally right on.
As you might expect, this sent us into a flurry of activity trying to figure out what we might have done wrong and how many people we would need to notify. In the end, it all came down to the geoid. When we get a lidar data set, one of the first things we do for storage and distribution is to put it into ellipsoid heights. We do this because the original measurements will have been referenced to the ellipsoid – or at least directly relatable to it mathematically. Those orthometric heights (e.g. NAVD88) that people like to actually do work in are generated from the ellipsoid heights by using a geoid model. And that’s the problem. It’s a model. Models aren’t perfect and we like to update models to make them better as we learn of their imperfections. [For more information on some of these terms, NGS has some good slide shows on their workshop program page, particularly the GPS-Derived Heights]
In this case, the original data flown in 2006 had used the GEOID03 model released in 2003 to transform from NAD83 ellipsoid heights to NAVD88 orthometric heights. We did the reverse operation to put it back to ellipsoid. However, when Digital Coast processes a lidar data set, most people want it in NAVD88 again and we apply the latest geoid model. Currently, that’s GEOID09 released in 2009. For this data set in Texas, the difference between those models is just over half a foot. If, as expected, GEOID12 is released sometime in the next year or so, our data will be provide slightly different answers
That might seem a little crazy. But I think it’s the right way to go for a couple reasons. First, storing in the ellipsoid will do a better job of preserving the original measurements now that we are in a GPS age instead of a leveling age for surveying. Second, it allows users to do change comparisons without having to wonder whether the changes they are seeing between two data sets are real or just a difference in processing. On the more selfish side of things, it means I don’t have to transform tens of terabytes of data each time a new geoid model comes out, I just need to put in the new transform grids and let it happen on the fly. You can still call me crazy if you want though.
If you’re interested in more about upcoming changes to the national spatial reference system, particularly with regard to elevation, the NOAA NOS Diving Deeper and Making Waves podcasts include one on building a better geoid and one on height modernization.