Greenness Scoring

Per-parcel or per-area score: which grain fits a listings page

Open grassy field bordered by a line of trees under a blue sky.

A location-quality field gets argued over more than it should, mostly because nobody agrees on what "location" means at the pixel level. Street noise, school catchment, walk time to transit, tree cover. Each of those can be scored at the parcel or at the area, and the choice changes what the number on the listing page tells a buyer.

What a per-parcel score buys you

A per-parcel score answers a narrow question: how green is this specific lot and its immediate surroundings. Useful when a listing sits right against a park on one side and a strip mall on the other, because an area average would smear those two realities into one mediocre number. Parcel-level scoring also matters for anything tied to the building footprint itself, like setback or lot coverage.

The cost is noise. Canopy cover on a single parcel can swing from one imagery pass to the next because a tree came down, a driveway got repaved, or the cloud mask clipped a corner. At parcel grain, that swing shows up as a visible score change on a specific listing, and someone has to explain it to an agent or a homeowner. It also means a nine-unit condo tower and the pocket park next to it get scored as two unrelated parcels, when a buyer looking at either listing cares about the same four blocks.

Why most listings pages end up at per-area

A property portal doesn't usually need to know that one rooftop has more canopy than the rooftop next door. It needs a number that's stable enough to show on thousands of listings without a support ticket every time a tree gets trimmed, and comparable enough that "72" on one neighborhood means roughly the same thing as "72" somewhere across town. That's a per-area location score: canopy, open space and impervious surface aggregated across a listing area, drawn at whatever boundary the portal already uses for its neighborhood pages, ZIP-level search, or custom polygons.

Aggregating greenness by listing area also smooths the exact noise that makes parcel scores jumpy. One tree removal barely moves a few-hundred-parcel area average. It would swing a single-parcel score by a visible amount. For a display field that sits next to price and square footage, stability matters more than precision down to the lot line.

There's a practical delivery reason too. A per-parcel score for every lot in a metro is a much bigger file and a much bigger refresh job than one score per listing area, and most of that resolution gets thrown away the moment it's rendered as a single badge on a card. If the boundary is small enough already, say a few blocks per polygon, the area score carries almost all the signal a buyer needs without the parcel-level overhead.

A middle path worth knowing about

The honest answer for a lot of portals is that neither pure parcel nor pure area is quite right. Picking a listing-area boundary small enough that the aggregate behaves like a neighborhood description, rather than a city-wide blur, solves that without adding a third scoring tier. A boundary drawn at the block-group or micro-neighborhood level gets most of the stability of area scoring and most of the specificity people want from a parcel score, without doubling the dataset.

That's the grain Greenness Scoring is built around: canopy cover, open space and impervious ratio aggregated per listing area from high-res multispectral imagery, refreshed annually, delivered as a score plus its underlying components so a portal can decide how much weight to give each one. See how the scoring works if a per-area number is what the listings page needs.

Score the parcel when the question is about that one lot; score the area when the question is about the neighborhood a buyer is choosing between. Most listings pages are asking the second question, whether they've said so or not.

Start a pilot

← Back to the blog