Fifty-four parcels left.
Fourteen pass the first screen.

Everything that's left

Green parcels passed the stated screen — a contiguous conceptual pad (16.3 or 7.4 acres) fits with at least 70% of it outside mapped wetland/stream buffers and steep ground. Brown parcels did not fit one under those assumptions. The extension south changed the picture: the rolling Hopkinton end is where dirt costs the most, and the flatter Bellingham–Franklin end is where the lower modeled earthwork results appear — the top five are all south of Milford.

The lidar that ranked them

The table is the result. This is the ground. The same 2021 MassGIS bare-earth hillshade used for the pad search, for all fifty-four parcels — the fourteen that passed in green, and the forty that did not in brown. Drag the line to compare current aerial imagery with the lidar surface. Click a parcel, a ranking card, or a brown site in the list to stand on that terrain.

Jump to a parcel that did not pass Choose a brown parcel…

All 54 screened parcels on 2021 lidar hillshade. Green passed; brown did not.

Four parcels that did not pass

The brown rows are easier to believe once you see the ground. Some are almost entirely steep or inside mapped buffers. Some sit against I-495. One Franklin parcel is only 14% constrained on paper and still could not fit a contiguous 7.4-acre pad — the lidar shows why a table cell cannot.

The numbers are checkable

1. Parcels — MassGIS L3 tax parcels joined to assessor records for the corridor towns; kept parcels ≥10 acres, classified vacant / forest / developable land, with under $50k of building value, within 1.5 km of the MassDOT I-495 centerline. Nine towns: Hopkinton, Holliston, Milford, Hopedale, Upton, Mendon, Medway, Bellingham, Franklin.
2. Terrain — the state's 2021 bare-earth lidar DEM, resampled to 1 m. On Projects 01–03 my point-cloud models and the state's DEM—derived from the same flights—differed by 2.0–7.6 cm RMSE. That supports processing consistency, not independent field verification of absolute accuracy. Wetlands and streams: DEP layers with 100 ft / 200 ft screening buffers.
3. The pad search — the same optimizer from Project 03, run inside every parcel: every possible pad position tested, largest pad kept where at least 70% of the window passes the terrain/mapped-buffer screen and 95% sits inside the parcel.
4. What screening doesn't do — no canopy metrics, no wetness-discrepancy or rock screens, no access routes, no zoning. Those belong to the per-parcel study (that's Projects 01–03, run on whichever of these parcels matters to you). Costs use the same $20/yd³ basis with the same caveats as the other projects.

Screening-level, from public data. This does not review current title, ownership, zoning, frontage, legal access, utilities, field wetlands, soils, endangered species, cultural resources, or permitting. A shortlist this small makes a deeper study practical for the parcels that matter. Also run per project: terrain-wetness discrepancy, vernal-pool candidates, surface roughness. Scripts and pipeline files available on request. Read the full assumptions, sources, and limitations.