Lidar terrain analysis · Hopkinton–Milford line, Massachusetts

Same warehouse.
Three spots.
Three price tags.

Rolling glacial forest along I-495, the corridor where Greater Boston's logistics buildings actually get built. On land like this the question isn't just "can you build" — it's "where does the dirt math hurt least." I screened every possible 16-acre pad position in this study area and priced the grading for the three best candidates.

Site location in Massachusetts

If you're looking at wooded land in Massachusetts, this is the kind of homework you can do before an offer — happy to walk through any of it. mhowe.gis@gmail.com

Bring us a parcelSee assumptions

Public-data demonstration. A client engagement would confirm current data, scope, control, and licensed-professional needs before relying on quantities.

$2.5M is the spread in earthwork cost between the three best pad locations — same building, same parcel-sized search area, different dirt.
316,993 yd³
planning-level earthwork at the selected conceptual pad
1,578,538 yd³
apparent earthwork if graded to the treetop surface
7.6 cm
how closely my model matches the state's published DEM
41.8M
laser points processed for this site
Start here

Drag the line

Left of the line is a treetop surface, similar to what photogrammetry commonly reconstructs when dense leaves hide the ground. Right is a bare-ground model, built from laser returns that reached openings in the canopy. Photography remains valuable for current color imagery and inspection; lidar adds the ground shape needed for this terrain screen. The yellow rectangle is the conceptual pad analyzed below.

Bare-earth lidar hillshade
Canopy surface hillshade
TREETOP SURFACE (CAMERA-STYLE)
BARE EARTH (LIDAR)
USGS/MassGIS lidar, flown spring 2021 · 1.5 × 1.2 km area · 0.5 m resolution
What it costs to get this wrong

One pad, three surfaces

The selected conceptual pad (P1 below) is 16.3 acres, 93% forested, chosen by the same constraint search that priced its rivals. It sits on the cleanest ground in the search area — 76% free of wetlands, setbacks and steep slopes — and still needs real dirt work, because on glacial till everything does. Grading it to a level 140.29 m requires:

Modeled quantities for one conceptual level pad. These are screening estimates, not construction quantities or bids.
Surface used for the estimateCutFillTotal movedIllustrative cost at $20 / yd³
Lidar bare earth158,621158,372316,993 yd³$6.34M
Treetop surface (canopy counted as ground)1,558,07320,4651,578,538 yd³$31.57M
10 m national DEM, 2019 archive (pre-lidar)184,217117,330301,547 yd³$6.03M
Canopy error hiding in the treetop surface1,261,545 yd³≈ $25.23M

Rolling terrain is where coarse old data fails hardest. The 2019 national DEM runs 77 cm high on average inside this pad (over a meter RMSE), and while its total looks close, it splits that total wrong: it predicts a 66,900 yd³ surplus to haul away from a site that actually balances to within 250 yd³. At $20 a yard, that's a $1.3M phantom line item — and the error grows with every hill the parcel has.

A simple ±10 cm data-sensitivity test changes the modeled volume by about ±8,628 yd³, or ±$173k at the illustrative unit rate. This shows how vertical uncertainty can move a screening result; it is not a construction contingency and does not cover design changes, soil conditions, haul, mobilization, rock, dewatering, escalation, or contractor pricing.

Cut and fill map of the pad
Where the dirt moves: cut in red, fill in blue, against the balanced design grade.
Site analysis from the same data

More answers from the same flight

The earthwork number is the headline, but the same dataset answers a bunch of other early-stage questions:

Three candidate pads with grading costs
Where should the pad go? Every possible position was screened; the three best are priced. The cheap-dirt option (P3, $3.96M) fails half its constraint screen — cheap dirt is usually cheap for a reason: the flattest ground here sits next to the wetlands.
Rock and till risk indicator
Is the cheap dirt hiding rock? Micro-relief under the canopy flags boulder fields and shallow ledge — a screen that can raise the estimate. The answer here: all three pads are clean at the surface (under 1% flagged; the open-field control reads 1 cm). Subsurface ledge stays invisible to lidar, so test pits remain on the diligence list.
Buildable land screen
Where can you build? Slope and wetland/stream setback screens on mapped ground: 176 acres pass the stated terrain and mapped-buffer screen out of 443. That is a shortlist, not a buildability opinion; zoning, access, utilities, field wetlands, soils, title, and permitting remain outside this screen.
Access road grade profile
Can you get a road in? Not cheaply from the east: the direct route climbs 15.9 m in 361 m and holds 11% grade for a stretch — over the usual 8% limit. The driveway needs a longer contour route, and that's a cost the flat parcels never mention.
Stormwater flow accumulation
Where does water go? Runoff concentration on the real terrain. On rolling till, every pad decision moves water somewhere new — this is the map the drainage engineer starts from.
Canopy height model
What does clearing cost? Tree heights for every cell. The selected conceptual pad carries 15.1 acres of forest — roughly $45–91k to clear at typical rates. Stumpage revenue can offset part of that — species- and market-dependent (UMass/DCR Southern New England Stumpage Price Survey); not netted here.
Viewshed before and after clearing
Who will see it? From the pad today: 0.0% of the surrounding area is visible through the forest. After clearing: 17.8% — including homes across I-495 that currently can't see the site at all. A retained tree buffer along the west edge would soften most of it — easier to plan for now than to argue about later.
Why this parcel

The corridor where the buildings actually land

The I-495 belt between Hopkinton and Milford holds the state's densest concentration of industrial space, and the next building goes on land like this: rolling, wooded, highway-adjacent. On flat ground, earthwork is a line item; on glacial till it's a siting decision worth millions, and it interacts with everything else — the flattest dirt here sits closest to the wetlands. This is also the site where the building-vs-tree problem got real: canopy height alone can't tell a subdivision from a forest, so the pad search uses the lidar's building classification to stay out of people's backyards.

Aerial imagery of the corridor today
The same ground today: subdivision west of I-495, the wooded block east of it where the candidate pads sit. Esri World Imagery; provider credits and source links are listed in the methodology.
The bottom line

What this means for the deal

Selected screening outputs. Costs are illustrative and exclude professional design, contractor pricing, and the items noted below.
Screening itemPlanning-level result
Modeled earthwork at the conceptual pad, graded to balance$6.34M; ±$173k data-sensitivity band
Clearing, 15.1 forested acres$45–91k
Access361 m route climbs 16 m, peaks at 11% — contour or regrade
Stormwater basin land take (pre-design screening)a natural low 300 m east stores the first-flush volume at 1 m stage — ~2.3 ac consumed
Terrain-screening subtotal≈ $6.41M + access work

Not included: survey/control, engineering, geotechnical work, rock excavation, unsuitable soil, drainage design, erosion control, utilities, pavement, retaining walls, permits, mitigation, mobilization, haul/disposal, escalation, or contractor markup.

On rolling till, earthwork is the biggest lever in the deal. Things a purchase agreement would want to cover:

  1. Siting flexibility — the grading bill swings $2.5M across the three viable pad positions; locking the building location before running the dirt math could leave a lot of that on the table.
  2. Geotech before reliance — the surface rock screen is clean, but subsurface ledge is invisible to lidar, so budget for test pits.
  3. Vernal pools — depression screen found no candidates in or near any pad; 3 state-mapped potential pools elsewhere in the study area. "Screened, clear" with field confirmation at permitting. Basin sizing above is screening-level — design belongs to the civil engineer.
Methodology

The numbers are checkable

  1. Data — USGS 3DEP lidar (2021 Central-Eastern Massachusetts, block 1, published accuracy 10 cm RMSE). 41.8 million points for this site. Wetlands and streams: MassGIS DEP layers with 100 ft / 200 ft buffers.
  2. Processing — same PDAL pipeline as Projects 01–02, third run: ground model, treetop model, canopy heights, plus the lidar's building classification as an exclusion mask for the pad search. Same scripts, different coordinates — which is most of what I wanted to prove.
  3. Processing consistency — my ground model and the state's DEM, both derived from the same flight, differ by 3 mm on average and 7.6 cm RMSE over open ground. That agreement helps check processing, but it is not independent field verification of absolute accuracy. Pre-2021 comparison: the archived December 2019 national DEM, not the current one, which was later rebuilt from this lidar.
  4. Volumes — cell-by-cell raster math, cross-checked against an independent QGIS implementation: cut agrees at 158,620.7 vs 158,621 yd³, fill at 158,372.0 vs 158,372 — six significant figures. Cost basis is an illustrative $20/yd³ screening assumption, with $10–$40 sensitivity; a real project would require current contractor or estimator input.

Public lidar ships with published accuracy specifications. A client scope would determine whether professional survey control and licensed oversight are needed before quantities are relied on. Screens run for every project: terrain-wetness discrepancy, closed-depression / vernal-pool candidates, surface roughness. Also available: intensity analysis, contours to CAD, per-tree inventory. Scripts and pipeline files available on request. Read the full assumptions, sources, and limitations.