Geospatial metadata and data standards for agency GIS

Reference article · 5 min read

A GIS layer is only as useful as what a stranger can learn about it without calling the person who made it. Coordinate reference system, feature definitions, attribute meanings, collection method, accuracy, and currency separate a layer a partner agency can build on from one that gets quietly set aside. Most agencies hold both kinds, and the undocumented ones surface when a state program opens a submission window or a records request produces a file nobody can explain.

Standards are revised and program requirements change, so confirm the current edition with the issuing body before writing any of this into a solicitation.

Spatial reference: name it precisely

Most trust problems trace back to an ambiguous coordinate reference system.

Name the realization, not just the datum. "NAD 83" alone is ambiguous. NAD 83 (2011) epoch 2010.00 is not. WGS 84 has likewise had multiple realizations that differ from one another.
Record an EPSG identifier. Registry codes travel between software far better than prose. If a projection has no registry code, ship its full parameter set, units and datum included.
Treat vertical separately. Elevations need their own datum, commonly NAVD 88, plus the geoid model used to derive orthometric heights.
Watch units. The US survey foot was deprecated in favor of the international foot at the end of 2022. The difference is tiny as a ratio and large enough to matter at State Plane magnitudes.
Expect the frame to change. The National Geodetic Survey has announced a modernized National Spatial Reference System replacing NAD 83 and NAVD 88. The schedule has shifted more than once, so confirm timing with NGS.

Schema and attribute domains

Document the schema as a data dictionary that travels with the data.

Coded value domains. Every categorical field needs an enumerated list of permitted values and the meaning of each. "PARCEL_TYPE = 3" is not data until someone writes down what 3 means.
Null handling. Publish whether unknown means null, empty string, zero, or a sentinel such as -9999. Mixed conventions in one layer cause silent analytical errors.
Persistent identifiers. Features need a stable key that survives edits, splits, merges, and reloads. A key reassigned on each publish makes downstream change detection impossible.
Dates and geometry. Use ISO 8601 with a stated time zone, and distinguish capture date from edit date. State the geometry type, topology rules enforced, and minimum capture scale.

What a minimally useful metadata record contains

Agencies are usually asked to meet one of two families: the FGDC Content Standard for Digital Geospatial Metadata, still common in state programs, or the ISO 19115 family, which FGDC endorsed as the path forward and which has companion XML encoding specifications. Open data catalogs often want a parallel DCAT style record. The profile matters less than the content. A trustworthy record answers:

What this is, in plain language, and what it was made for
Who produced it, who maintains it, and who to contact
Spatial reference, spatial extent, and temporal extent
Lineage: source materials, capture method, processing steps
Positional and attribute accuracy, including known gaps
Entity and attribute definitions, including every domain
Use and access constraints, license, and any redaction applied
Maintenance frequency and when the metadata was last reviewed
Distribution formats and how to obtain the current version

A record that passes a validator but says "see attached" in the abstract meets the letter of the standard and fails its purpose.

Update cadence and versioning

ISO metadata carries a maintenance frequency element with values such as continual, monthly, annually, as needed, and irregular. Choose one deliberately and hold to it.

Give each published extract a vintage label, carried inside the data as well as the metadata, and keep prior vintages retrievable so older analysis can be reproduced.
Version the schema separately from the data. A new field is additive. A renamed field, changed domain, or changed identifier scheme is a breaking change that deserves notice before it ships.
Keep a plain change log so consumers do not have to reverse engineer what changed.

Parcels, addresses, and utility networks

Parcels. The FGDC cadastral data content standard is the long-standing national reference, but the binding requirement is usually a state or regional aggregation program with its own submission schema and validation rules. Get that specification before building.
Addresses. The FGDC United States Thoroughfare, Landmark, and Postal Address Data Standard defines address components and how to parse them. For public safety, the NENA GIS data model for Next Generation 9-1-1 governs address points, road centerlines, and emergency service and provisioning boundaries, with its own validation rules.
Utility networks. There is no single national content standard. ASCE standards cover the collection and depiction of existing subsurface utility data, including quality levels describing how a location was determined. Record that quality level: a line traced from a scanned as-built and a line from surveyed potholing are not the same fact.

What to require in a solicitation or contract

Name the standard, profile, and edition, and make a passing validation report an acceptance condition rather than a best effort.
Require a written data dictionary covering every field and domain, delivered with the data, and the coordinate reference system declared by registry code in every deliverable.
Require an early sample deliverable, accepted before bulk production, so schema disputes surface in week three and not month nine.
Require at least one open delivery format, and state that the agency owns the data and metadata.
Make correcting nonconforming deliverables the vendor's obligation at no added cost, within a defined cure period.

HQT works with state and local agencies on data engineering and documentation for geospatial systems, including schema design and metadata conformance for program submissions.

Frequently asked questions

What is the difference between FGDC CSDGM and ISO 19115?

CSDGM is the older Federal Geographic Data Committee content standard for digital geospatial metadata, and it remains in use across many state and local programs. ISO 19115 is the international metadata standard family that FGDC endorsed as the path forward, with companion specifications defining how records are encoded in XML. Both cover the same underlying questions about identification, lineage, quality, spatial reference, and distribution. Which one you must produce depends on the program requesting the submission, so ask for its current specification rather than assuming.

What is the minimum metadata a local government GIS layer should have?

At minimum, a reader should be able to learn what the layer represents, who maintains it and how to reach them, the coordinate reference system, the spatial and temporal extent, how it was collected and processed, its known accuracy and gaps, the definition of every field and coded value, any use or access constraints, how often it is updated, and where to get the current version. That content is what makes a layer reusable, regardless of which standard it is encoded in. A record that validates cleanly but leaves a reader guessing has met the format and missed the point.

Which coordinate system should we publish GIS data in?

Publish in the system your primary users and partners already work in, which for most local agencies is the applicable State Plane zone or a documented local projection, and be explicit about the datum realization and units. Whatever you choose, record an EPSG code so software can interpret it without guesswork, and note the vertical datum and geoid model separately if the data carries elevations. If you also serve data on the web, a reprojected copy in a common web mapping projection is fine as long as the authoritative version and its reference system are clearly identified.

What standards apply to parcel and address data submissions?

For parcels, the FGDC cadastral data content standard is the national reference, but the requirement that actually binds you is usually your state or regional parcel aggregation program, which publishes its own schema, field names, and validation rules. For addresses, the FGDC United States Thoroughfare, Landmark, and Postal Address Data Standard defines address components and parsing, and public safety address and boundary data is typically governed by the NENA GIS data model for Next Generation 9-1-1. Request the current submission specification from the program office before you build, since these documents are revised.

How often should GIS data and metadata be updated?

Set a maintenance frequency you can actually sustain, publish it in the metadata, and review the record itself on that same schedule so the stated currency stays true. Parcel and address layers usually warrant frequent cycles because they change constantly, while base layers such as jurisdictional boundaries may change rarely. Label each published extract with a vintage, keep prior vintages retrievable, and maintain a short change log so downstream users know what moved and when.