Saturday, January 31, 2009

Voxels and Tetrahedral Networks

One of the more interesting capabilities of 3D GIS is the ability to represent 3D objects to more closely match reality. Right now, however, you only have 2.5D surfaces with ESRI technology. There are TINs, raster surfaces and ESRI’s relatively new type of surface called a Terrain but none of these are 3D, meaning they can contain multiple Zs at the same X,Y. In a recent dialog I had with a leading ESRI developer, I found out that ESRI is still in a research stage when it comes to using Voxels and Tetrahedral Networks (TEN). This is pretty astounding considering that GRASS has supported Voxels for nearly 3 years now. What gives? Both of these data models are sorely needed in the earth sciences for a variety of applications but ESRI is saying that they are finding it difficult to find the users and requirements for them. But, all you have to do is a basic Google search and you will see that these models are already used especially with geology, archaeology, oceanography, soils, etc. Therefore, is this a problem of ESRI looking for people that want these kinds of models but users aren’t requesting them or is it that users are just going out and using other tools and not even asking ESRI? Also, which is the best approach for representing true 3D surfaces? It’s probably not one of them but both of them and it’s highly dependent of the data and the modeling that is required. Another important component of this is whether ESRI should proceed forward from the perspective of OpenGIS specifications for these data models or developer another proprietary format. I would hope the former. As a geographer, I’ve always found ESRI’s technology sufficient for most applications, but as GIS becomes more integrated into science these new data models are sorely needed. What’s your take on this? Is this something you need?


  1. I have no experience with voxels or TENs but I would love to see more intuitive editing of multipatch features, and generally editing in a 3D environment.

  2. The problem may be that many users feel like why bother complaining or asking for things that will take such a long time to come to fruition. I’m not accusing ESRI of being unresponsive but they are a pretty big machine and it takes time to them moving. Maybe part of their marketing strategy could be to entice users to provide input and participate in more surveys. A piece of candy works as well as a cattle-prod.


  3. GMS ( has been doing this since the at least 1996. You can have a TIN with several versions of Z values, e.g., from different interpolation algorithms, multiple strata surfaces or whatever.

    In addition to multiple Z, the time dimension is also supported in GMS (for groundwater levels that change with time).