One Digit, Half a State

Four codes, one digit apart

Indiana splits into two state-plane zones, East and West, running down the middle of the state. Each zone exists in more than one datum realization. Four codes cover the common cases in US survey feet:

• EPSG 2965 – NAD83 / Indiana East (ftUS)
• EPSG 2966 – NAD83 / Indiana West (ftUS)
• EPSG 2967 – NAD83(HARN) / Indiana East (ftUS)
• EPSG 2968 – NAD83(HARN) / Indiana West (ftUS)

Adjacent numbers, radically different meanings. 2965 to 2966 moves you half a state sideways. 2966 to 2968 keeps you in the same zone but changes which realization of NAD83 your coordinates are referenced to. A PNEZD file carries none of this information in its rows – the numbers are just numbers – so the label travels in a project note, a CAD setting, or somebody’s memory. One digit of drift in that label and the file means something else.

The loud failure: wrong zone

Take a file of East-zone coordinates and declare it West zone. Nothing subtle happens. The two zones have different central meridians and very different false origins, so reprojected positions land somewhere absurd – the parcel renders in the wrong part of the state, or the eastings are obviously the wrong magnitude for the county you are in.

That sounds safe, because it is unmissable. The real danger is the second move: an office in a hurry sees data that “came from the surveyor” landing in the wrong place and reaches for a shift. Translate it until it sits over the aerial, rubber-sheet it in the GIS, call it done. Now the file looks right, checks against nothing, and carries an undocumented transformation that someone will rediscover the hard way in five years. The loud failure is only safe if the response is “fix the label,” never “move the points.”

The quiet failure: wrong realization

Now the nastier sibling. NAD83 and NAD83(HARN) are two realizations of the same datum – same zone definitions, same units, coordinates that agree to within tenths of a foot. Label a HARN file as plain NAD83 and everything passes the eyeball test: the parcel sits on the aerial, the eastings are plausible, the map looks perfect.

The offset is still there. It is systematic, it is real money at boundary tolerances, and it surfaces at the worst possible moment – when a corner will not check against a control point, or two adjacent jobs disagree by a stubborn few tenths that no re-measurement explains. Nobody suspects the label, because the label has been “working” for years. If the loud failure wastes an afternoon, the quiet one wastes a week – most of it spent doubting the field work instead of the metadata.

Zones are assigned by county – look it up

One more trap in the family: guessing the zone from geography. Indiana’s zone boundary follows county lines, not a line of longitude you can eyeball. A corridor that crosses a county line may stay entirely in one zone – or split – and the only authority is the registry: every EPSG zone publishes an area-of-use that names its counties. Ten seconds on epsg.org settles what an hour of squinting at a map will not.

The habit that catches all of this is cheap: when a coordinate file enters your world – from a subconsultant, a county, an old archive – check one known point before you build on it. A section corner, a published benchmark, a monument your firm has shot before. If it checks to survey tolerance, the label is probably right. If it is off by tenths of a foot systematically, suspect the realization. If it is off by miles, fix the label and touch nothing else.

The correction: we got one of these wrong

This post has a postscript with our own name on it. Our May piece on modified state-plane archives cited “EPSG 2966, NAD83(HARN) Indiana East” as its example code. Wrong on both counts of the family: 2966 is NAD83 / Indiana West. HARN Indiana East is 2967. A reader working in these zones – exactly the reader that post was for – deserved better, and the original post is now fixed.

We are correcting it in public rather than quietly because the lesson is the post: CRS identity lives in labels, labels drift, and the fix is verification against an authority, not confidence. That applies to a firm’s forty-year archive, and it applies to a blog post about a firm’s forty-year archive.