Editorial & Data Standards

Every number on this site traces to a named source. We compute against real measured data wherever it exists; where we fall back to an estimate — like our typical growing-degree-day and chill-hour bands, which are estimated from a state’s USDA hardiness zone — we label it as an estimate, never dressed up as a measured fact. Here is what each layer is built from and what it tells you:

• USDA SSURGOfield-mapped soil — texture, pH, drainage class, and depth under your parcel.
• NOAAfirst and last frost dates and climate normals that set your growing window.
• USDA PHZMthe USDA hardiness zone for your exact point, by spatial join.
• USGS 3DEP · EPQSelevation and terrain, read from the national digital elevation model.
• USGS NAWQA EPestcounty-level estimated agricultural pesticide use by chemical.
• EPAdocumented contamination records — Superfund, USTs, brownfields, TRI, PFAS, and more.
• FEMAflood-zone designations from the National Flood Hazard Layer.
• USGS 3DEP · NAIP-CHM · pvlib · Konarska 2013sun hours from our sun-cast scene decomposition — solar geometry ray-marched over a leaf-on canopy-height scene, with NREL NSRDB as the fallback.

The full scoring math lives on our Methodology page, and the complete source catalog — every dataset, its agency, and its license — is on Our Data.

Our plant catalog is published as a citable, versioned open dataset with a per-fact provenance ledger:

• Mechanical parsing, no invention. Values are read straight from the source datasets. Nothing on a data surface is written by a language model — the numbers come from the federal record, not from a paraphrase of it.
• We show gaps, we don’t fill them. When a dataset doesn’t cover your parcel, we say so — “not surveyed,” “not mapped” — rather than substitute a guess. A visible gap is more useful than a confident wrong answer.
• Cross-source validation. Layers that should agree are checked against each other, and our plant scoring is version-controlled with a public history you can read in Methodology — every change to how a score is computed is dated and recorded.

Our contamination and risk layers are a screening estimate, not a soil test. They tell you what is documented in the public record near an address — the sites on file and how close they are — measured against the thresholds set by government agencies. They do not measure what is actually in the soil on your land.

So we frame the result the way the data supports it: a nearby documented site is a reason it may be worth testing — not proof that anything is dangerous. The two questions are different, and only a physical soil test answers the second one.

When a screening result suggests it, the constructive next step is a lab soil test — inexpensive, and the one thing that turns a proximity estimate into a real measurement of your ground.

Found something off? Tell us and we’ll look into it and date-stamp the fix. Reach us through our contact page or at [email protected]. Corrections to the underlying federal datasets go to the agency that publishes them; we’ll point you to the right one.

Growable Ground is built and maintained by its founder, Michael Kirchick. As the platform grows we are working to add a named, credentialed reviewer to this page — until one is retained, we leave that line honest and empty rather than fill it.

A Growable Ground report is a starting point, not the last word. It doesn’t replace a physical soil test, a professional site survey, or the local knowledge of your Cooperative Extension office — it points you to the questions worth asking and scores 1,086 plants against the public record for your land so you know where to look first.