CSV vs layer attribute: which rooftop height format fits your workflow

A industrial building's gabled facade with large windows under a clear blue sky.

You've narrowed a search ring to forty candidate rooftops and now someone has to decide what format the height data comes in. It sounds like a formatting detail. It isn't. The format changes how the data behaves six months from now, when the ring moves or a building gets renovated and your numbers need to update without you rebuilding the whole join.

Here's what actually differs between a CSV and a layer attribute, and which one fits a given stage of candidate screening.

What each format actually gives you

A CSV is a flat table: building ID, address or centroid, estimated height, estimated storey count, maybe a capture date. Nothing spatial about it until you join it to something. You open it in Excel, you filter it, you email it to a colleague who doesn't have GIS installed. That's the whole appeal.

A layer attribute lives inside a shapefile or geopackage, attached to the building footprint polygon itself. Open the layer in QGIS or ArcGIS and the height is already sitting there on the feature, no join required. Select a building on the map, the storey estimate shows up in the attribute table next to it.

Same underlying number. Different place it lives.

CSV: good for screening, bad for staying in sync

CSV wins when you're doing the first pass on a candidate list, the kind of work where you're pulling fifty addresses out of a market study and just need a height column next to each one. A spreadsheet you can sort by storey count and hand to the RF team with a one-line email gets the job done without opening a map.

The catch: a CSV's building ID has to match something on your side, usually a parcel ID or a footprint you already hold. If your footprint layer gets a re-run, or the vendor refreshes the dataset and IDs shift slightly, the join breaks: rows drop out of the match with no error message telling you which ones. The height column and the parcel column stop lining up, and the first sign of it is usually a height number that looks wrong three weeks later. For a one-off screening pass that's a manageable risk. For something you're going to re-run every cycle, it's a maintenance job you've signed up for without meaning to.

Layer attribute: good for staying current, heavier to set up

A layer attribute skips the join problem because the height rides along with the footprint geometry as one object. Refresh the layer and the attribute refreshes with it, same feature, same ID, no separate table to reconcile. If your site-selection workflow already runs on a footprint layer in QGIS, pulling height straight into that layer's attribute table is less work long-term than maintaining a side CSV and a join script.

The tradeoff is that it only helps if you're already working in a GIS. If the person who needs the number is a leasing agent with Excel and no GIS seat, a layer attribute does them no good until someone exports it back out to a table anyway. And a layer attribute is tied to one footprint source. If you change your base footprint provider, the attribute doesn't travel with you the way a CSV row with an address does.

Picking one for your workflow

If the height data is feeding a quick candidate ranking that leaves the GIS and goes to people who don't open shapefiles, CSV is the right call. If it's feeding a footprint layer you already maintain and query day to day, ask for the layer attribute and save yourself the join.

A fair number of planners end up wanting both: a layer attribute for the working map and a CSV export pulled from the same layer for whoever needs the flat list. That's a reasonable ask if the vendor can deliver per-building height either way, from the same underlying run, so the two never drift apart.

Storey Estimation delivers rooftop height and storey count in both formats from the same annual VHR pass, so the layer and the spreadsheet always match without a separate reconciliation step. If you're deciding how height data should land in your stack before your next candidate round, that's worth a look.

Get started

← Back to the blog