County Lines

Nobody shops for land by county. You shop by region — the Blue Ridge foothills, the Shenandoah Valley, the mountains west of Asheville. You have a sense of the area you want to be in, and the county boundaries running through it are administrative lines that have nothing to do with the terrain, the forest, or the value of the land.

But every data source that describes rural land is organized by county. Tax records. GIS layers. Parcel boundaries. Assessment data. Ownership records. Each county maintains its own database, in its own format, with its own schema and its own quirks.

If you’re a buyer looking at properties across Albemarle and Nelson counties in Virginia, you’re searching two completely separate systems. Different column names. Different data formats. Different web interfaces. The parcel that’s perfect for you might be half a mile across the county line from where you’ve been looking, and you’d never find it because you searched the wrong county’s GIS site.

The Data Problem

TALON started with Albemarle County, Virginia — 47,000 parcels and a weekend. Getting that data into a searchable format meant building an ETL pipeline — extract from the county’s published data, transform it into a normalized schema, load it into a database that supports spatial queries.

When I started adding Buncombe County, North Carolina, I expected to reuse most of that pipeline. Different state, different county, but how different can parcel data really be?

The answer: very.

Column names don’t match. Albemarle stores acreage in one field; Buncombe splits it across multiple fields with different names. Date formats differ. Coordinate systems differ — Virginia uses state plane coordinates in feet, North Carolina uses a different state plane zone. Owner name formatting is inconsistent. Assessment breakdowns use different categories. Zoning codes are county-specific and don’t map to each other cleanly.

The land itself is continuous. The data describing it is fragmented by jurisdiction.

One Schema, Multiple Sources

The solution is a normalized schema — a single data model that every county’s data gets translated into, regardless of how it was originally formatted. Parcel boundary geometry. Total acreage. Assessed land value. Assessed improvement value. Owner name and address. Transfer date. Zoning classification. Building count.

Each county gets its own ETL pipeline that handles the translation. The pipeline knows how Buncombe formats dates, how Albemarle encodes zoning, how each county structures its owner address fields. The output is the same: a row in the database that looks identical regardless of which county it came from.

This means terrain analysis, ownership classification, timber estimates, and every other computation TALON performs can run against the normalized data without knowing or caring which county sourced it. The slope distribution of a parcel in Buncombe County is computed the same way as a parcel in Albemarle County, from the same USGS elevation data, using the same algorithms.

Once multiple counties share a schema, search works the way buyers actually think.

“Show me parcels between 15 and 40 acres, with south-facing gentle slopes, zoned for rural or agricultural use, within 30 minutes of Charlottesville.” That query might span three counties. With TALON, it’s one search. The results show up on one map. The terrain analysis is comparable across county lines because it comes from the same national dataset.

This is the kind of search that’s completely impossible on any existing platform. County GIS sites don’t talk to each other. Listing sites only show properties that are actively for sale. The data that would let you search a region — all parcels, all attributes, all terrain metrics — has been locked inside county silos.

What’s Next

Every county TALON adds requires a custom ETL pipeline. The terrain analysis and LiDAR processing are county-agnostic — they run against national datasets. But parcel data, ownership records, and assessment information are county-specific, and each one needs its own translation layer.

That’s the work. It’s not glamorous. But every county that comes online extends the searchable area and gets closer to the way people actually look for land — by region, by terrain, by what the property is and what it’s worth, regardless of which side of the county line it falls on.

The land doesn’t know about county lines. The tool shouldn’t either.