Conventional wisdom has promoted the myth that high-quality asset inventories can be completed using entry-level personnel. This is categorically incorrect. Experience is the more cost-effective option — people that know their data can do the work more accurately, much faster, and (in terms of quality and productivity) way cheaper.
Recently, I took over an asset inventory project that had been implemented using summer interns. I applaud efforts to pinch pennies in these tight economic times, but I would never want to manage a project like this without the involvement of experienced land surveyors.
In April 2008, three college interns and one brand-new city employee were tasked to locate, identify, and evaluate the condition of all surface utility structures within the more than 400 miles of dedicated bike lanes and major arterials of Seattle. None were familiar with the features they were inventorying, the bicycle network and city neighborhoods in which they worked, or the GPS equipment they used.
Although GPS gear is easy to use, getting location coordinates is only one of the attributes collected for an asset inventory. Finding, identifying, and characterizing surface utility features — manholes, inlets, gate valves, etc. — in a complete, correct, and consistent fashion takes training and experience. Several months and $25,000 later, only a fraction of the city's utilities within the bike lanes and arterials were inventoried, and the resulting data was in shambles.
It was at this point that the project came to me, with only seven weeks left to complete it. I was assigned two inexperienced engineering technicians who would work as one full-time equivalent, and two experienced in-house surveyors. I also hired two private-sector surveyors — though it took three weeks to navigate the paperwork to get them contracted and on the job.
To properly identify and describe the state of utility structures I first needed to create a data schema — or data dictionary — which defines the features to be located, the attributes used to describe each feature, and the valid values (a range of characters or numbers indicating type, condition, etc.) allowed for each attribute. Once completed, the schema was loaded into a Trimble handheld GeoXH GIS data collection unit.
I also supplied each field person with a users manual explaining the operation of the unit, the five attributes to be captured, and the valid values associated with each attribute. None of the staff were familiar with data schemas; however, the land surveyors were accustomed to using descriptor codes to describe features.
After mobilizing the equipment and crews, the next challenge was to review the interns' data to figure out what areas had been completed.
I was given a file that was supposed to contain all data collected in the 2008 inventory. After more than 60 hours of trying to normalize the data into a coherent schema I realized that a lot of data was missing. I combed through all of the files I could find, scattered on three servers, and came up with nearly twice the data I had been given originally. After formatting the new data to match my schema, I found that several data sets had been saved under different names and therefore had been entered twice. Sometimes the duplicate data sets represented the same data, and sometimes the information was different. I had to make a lot of assumptions to decide which data to keep.
I used the two engineering technicians as my control group to evaluate how their work compared to the 2008 interns' work, as well as the 2009 land surveyors' work. The following elements were reviewed: