Geospatial Data Accuracy – Better and Better

Circa. 1995. I walked into the GIS office of a major forest products company in northern Arkansas (or was it northern Louisiana…not sure). At the time, I was a product manager for a GPS company and field testing one of our newer GPS mapping hand-held products.

We decided to go out and map the perimeter of a timber tract they owned to compare the area (acres) that the GPS calculated vs. what was in their GIS. This was nothing new as I’d done it many times for other companies. We went to the tract (adjacent to a road) and walked the boundary (~40 acres). Afterwards, we returned to their office and I post-processed the GPS data.

I forget what the final area calculation was, but it was believable and the company didn’t challenge the result. Also, the shape of the polygon seemed reasonable. However, when the GIS manager inserted the GPS data into his GIS, it was offset a significant amount. I forget exactly, but something on the order of 50-100 feet. I immediately began considering if my data was the problem, but concluded the chance was low. For GPS post-processing, I’d tied into a local US Forest Service GPS base station so my GPS data was referenced to NAD83/86 (if I recall correctly). In the end, we agreed that my data was most likely positioned correctly.

“So what?”, he said, “Do you think I’m going to adjust my entire GIS because it doesn’t agree with your GPS?” (I’m paraphrasing based on my recollection). I understood that I had won the battle, but lost the war. It didn’t matter that I was right, at least at that moment in time. However, he did agree with me that eventually he was going to have to reconcile the difference because GPS was destined to be the technology that defined the national spatial framework.

I had many more experiences similar to the above during the mid-90’s. People would swear by the accuracy of USGS 1:24,000 quad sheets because that’s what they were used to. If the GPS data didn’t agree with the quad sheet, they’d dismiss the accuracy of GPS because it didn’t fit. This was particular true with utility companies too, that were some of the early adopters of CAD for mapping.

Fifteen years later, this problem is not going away. The accuracy of Geospatial data continues to get better and better. Think back fifteen years and ask yourself about the quality/availability/price of orthophotography back then. I remember we were ecstatic to have access to free 1-meter, black/white DOQQs. Today, I can easily find 1’ pixel resolution orthophotography, commonly find 6” and occasionally run into 3” pixel resolution orthophotos free of charge. This allows one to digitize manholes and other infrastructure without leaving the seat at your GIS workstation.

During the same period, the cost of accurate GPS measurements has reduced considerably. Whereas fifteen years ago, achieving sub-meter accuracy with a $12,000 mapping receiver was on the bleeding edge of technology. Today, a $2,000 mapping receiver can give you sub-meter results and a $6,000 receiver can achieve sub-foot accuracy. Looking way forward, the cost and availability of GPS accuracy is going to change significantly in the next 10 years. Obtaining one foot accuracy will be achievable with a very inexpensive GPS receiver.

We all know that data drives a GIS. The better quality data we have, the more accurately and precisely and completely the GIS can answer our queries. Along these lines, I think it’s worth mentioning again the outcome of the litigation in California involving Santa Clara County and the ownership of GIS data.

Santa Clara County GIS lawsuit

Santa Clara County (California) was charging significant fees (potentially several hundred thousand dollars) to organizations who wanted to utilitize its full suite of GIS data including orthophotography, parcel, planning, streets, boundary, etc. A lawsuit was filed in 2006 by the First Amendment Coalition arguing that the GIS data should be released under the California Public Records Act. Santa Clara County argued that the GIS data was sensitive enough to be excluded due to homeland security issues (eg. making known the locations of critical infrastructure such as utilities). Santa Clara County lost the argument and was ordered by the court to hand over the GIS data. The 6th District Court of Appeal gave the final word last February.

The outcome of the court case establishes a significant precedent in the geospatial industry. For as long as I can remember, this issue has been solidly ambiguous among state and local governments. One entity would email (or make available via FTP) GIS data at a moment’s notice. Another entity would have you sign away your first-born child. Even another would not entertain the thought of releasing “our data” to anyone. I think the attorney for Santa Clara County was accurate in stating “It was one of those cases that needed to be tried and for which we needed guidance from the court”. Normally, I have an anti-litigious attitude, but I’m happy to see a precedent has been established and publicized.

Go on…be a TIGER

After last week’s column about Google’s step forward in using their own base map for Google Maps/Earth in the US, I was admonished by a reader, and rightfully so, in not mentioning the value of Census data as an important part of the history of base map evolution in the US.

Jon Sperling, Ph.D., GISP wrote:

“It is quite disconcerting, from an historical and current perspective, that your article made no mention of the “pre-internet” Census TIGER database, the first topologically integrated national digital street centerline for the US or even the newly updated and positionally accurate TIGER files (with an associated but confidential file of GPS address points collected for every housing unit in the Nation). These files, newly updated for the 2010 Census, are still a major source for accurate and easily accessible public domain street level data for every community in the US, including Puerto Rico, the Virgin Islands, and other territories. More importantly, it was the innovation that spurred the GIS revolution across government, private industry, and academia by enabling every local agency, entity and person in the United States to build their own geographic information system by combining census data with TIGER. Prior to this development, the Census was also a leader in the development of the GBF/DIME Files which enabled address geocoding, a capability that led to the later success of Mapquest, Google, and others. Like the development of the internet by DARPA and the Global Positioning System also by the federal government, the widespread development of intelligent national street level mapping was also led by the government.  GDT/TeleAtlas began as a company that offered “enhanced” TIGER files and Navteq often used TIGER for the more rural areas. Not only has the Census pioneered but it remains a key catalyst and building block for delivering a cost-effective and truly integrated national spatial data infrastructure.”

Attached is an article I wrote back in 1992 on the history/development of TIGER as well as a 2002 proposal for creating a shared national road network (geometry and basic attributes such as address range to enable consistent and shareable geocoding across
domains). OpenStreetMap is a nice expression of a way to move forward.”

A copy of Jon Sperling’s 2002 proposal can be read here (scroll down to Page 16).

Thanks and see you next week.