What really matters to GIS professionals

Last week I attended a workshop sponsored by the Oregon GPS User’s Group (soon to be Oregon GNSS User’s Group). OGUG invited Michael Dennis, RLS, PE, current Ph.D. geomatics student, former full-time National Geodetic Survey (NGS) employee, all-around smart guy and entertaining speaker to present an all-day workshop entitled “Space Time and Datum Forensics – A Geodetic Workshop.” Let me tell you, its 260 slides of stuff that matters in GIS, surveying and GNSS if you’re working with data at the sub-meter level and better.

The audience was largely surveyors, and that’s a problem. I’d go as far as saying that it’s significantly more important for GIS professionals to understand this topic than surveyors. The reason is because surveyors are project-oriented. For example, Joe Surveyor is hired to complete a boundary and topo survey for a new commercial real estate development project. He does the research, does the field work, completes the deliverables, issues an invoice, and places the project file into storage. Joe might look at the file again in six months when construction begins and may never look at it again after that.

Surveyors are short-term, project-based data generators. On the other hand, GIS professionals are long-term data managers. Therefore, for surveyors, their data doesn’t require accuracy, it requires precision. On the other hand, GIS professionals value accuracy much more, or at least they should.

The reason is because all the data layers in their GIS need to play together. GIS layers need to be spatially consistent. Managing a spatial and tabular-robust GIS database is a substantially more complex task than the typical surveyor encounters. Perhaps that complexity is one of the reasons that the spatial geodesy of a GIS database largely falls below the noise floor. In other words, there are much larger problems to tackle in a substantial GIS database other than geodesy.

How many surveyors have ever had to deal with SAP databases? How about an SDE (how many of you had to Google the acronym)? How about writing a script that queries a MySQL database to extract features with particular attributes? That’s just the beginning.

Before a surveyor criticizes a GIS for its accuracy, or lack thereof, that person should spend some time dealing with some of the data-management issues encountered by a GIS professional. There are entire conferences focused on only this subject. That’s what GIS is all about: data management, long-term data management.

A GIS doesn’t get filed after every project is completed; it gets added to the last project, and with each project, the database grows larger, more unwieldy, and likely more difficult to manage. And then, someone or some company throws a curveball at them, a new schema, a new operating system, or a new enterprise GIS software version that impacts the entire database. The IT department gets involved, and then procurement gets involved. Before you know it, it’s meetings after meetings. You get the picture.

Among all of the complex GIS database management issues, the geodesy of GIS has stayed below the noise floor. In other words, it’s been largely ignored. But as I’ve written in the past for GPS World magazine and this publication, GNSS, remote sensing and other sensors are becoming cheaper, faster and more precise. Therefore, data being appended to GIS databases are becoming more precise.

This creates challenges by highlighting the legacy inaccurate or imprecise data in the GIS database, which in turn creates the necessity for another decision to be made: what should we do about it? The answer to that question is for another article, or three.

With that, there are a few slides from Michael’s total of 260 slides in the workshop that I would like to highlight.

His second slide is one my favorites. It’s a bit rhetorical in that there is no answer, but succinctly states the problem. The value of a GIS database is the relationship of spatial data amongst its elements. How close is the gas pipeline to the nearest home? Where’s the shut-off valve for main water line on First Street? Which homes will be at risk of flooding during a storm surge in Galveston, Texas? How fast will the latest virus likely spread if the first case is discovered in Atlanta vs. Nowhere, USA? GIS can answer these questions, but its answers are only as good as the data in the GIS. Good ol’ garbage-in, garbage-out.

Before we get into the weeds, this is another setup slide that succinctly frames the challenge.

To be clear, a coordinate system always includes a datum (a.k.a. geographic coordinate system, geometric reference system/frame), and it may or may not also include a map projection. Examples of projected coordinate systems include UTM (Universal Transverse Mercator), US SPC (State Plane Coordinates), Web Mercator (think Google Earth), Lambert Conformal Conic, and Gauss-Kruger for my European brethren. These systems must always include a specific datum. Common geodetic datums are ITRF08, IGS08, NAD83, NAD27, ED50, and WGS-84. You may have different map projections for each datum. For example, UTM or SPC can be referenced to NAD83. It’s a straight-forward mathematical operation to change the projection if the underlying datum is the same. However, UTM coordinates referenced to NAD83 or WGS-84 is a different subject altogether. Going to/from UTM/NAD83 to UTM/WGS-84 is far from being a straight-forward mathematical operation.

The next feature slide gets into the weeds a bit. This is a subject I’ve written about for a few years and was somewhat highlighted in two articles I wrote (and a webinar I moderated) called “Nightmare on GIS Street.” How many of you think you use data referenced to WGS-84?

WGS-84 referenced data is probably the most widely mis-used. As you can see from the above slide, the definition of WGS-84 has changed over time. You’ll never find a survey mark on the ground with coordinates referenced to WGS-84. If you do, it’s wrong. This is because it’s a military thing. WGS-84 is managed by the US Department of Defense. More specifically, the US National Geospatial Intelligence Agency (formerly NIMA, formerly DMA). Fortunately, in recent years, the Department of Defense has aligned WGS-84 with ITRF (International Terrestrial Reference Frame) — most recently to ITRF08 — and ITRF/IGS coordinates are publically available. For example, IGS08 (International GNSS Service of 2008) coordinates are published for NGS CORS and available in NGS OPUS solutions (for the purpose of this discussion we can consider ITRF and IGS the same). However,  there is a caveat: ITRF08/IGS08/WGS-84 coordinates are referenced to specific dates (epochs).

WGS-84 was aligned with ITRF08 at epoch 2005.00, meaning that the WGS-84 coordinates were defined for the date of January 1, 2005. NGS publishes IGS08 coordinates at epoch 2005.00 for CORS. But OPUS solutions give IGS08 coordinates at the date of the GPS data file, and both autonomous and WAAS-corrected GPS gives positions at the mid-year epoch of the current year (i.e., positions are now at epoch 2016.5). This matters because stuff moves, including the ground you are standing on. Some places move more than others. California moves more than Missouri. Chile moves more than Germany. January 1, 2005 is 11+ years ago. If the ground is moving 3cm/yr, that’s 33cm over 11+ years. If you’re counting centimeters, that’s quite a few of them.

Software vendors have a hard time keeping up with modern datum transformations, and this next slide is a perfect example of that. It’s not just one vendor. Nearly all software vendors “aren’t doing it right.” In this slide, this software vendor displays 10 different transformations from “WGS84” to “NAD83”. Which one do you use? None of them get it right.

The most accurate one is still 20 cm in error. The worst is more than a meter in error. It makes you wonder why you spent $8,000 on that sub-foot GPS handheld when your GIS software may be introducing three feet of error.

Finally, should you be concerned about this stuff?

If you expect some of your data layers to be accurate to less than three meters, the answer is “yes.”

I’ll likely continue this discussion next month or in the coming months,and also provide a link to Michael’s 260-page slideshow.

Thanks, and see you next month.

Follow me on Twitter at GPSGIS_Eric