Negative grading around a house: what it is, and how it is measured

The definition, stated tightly

Grade is a slope, expressed as a percentage: the vertical change divided by the horizontal distance. Ground that drops 15 cm over 3 m has a 5% fall.

Take the wall as the origin and measure outward. If elevation decreases as you move away, the grade is positive in the sense a builder means — water runs away from the house. If elevation increases as you move away, that is negative grading, and water runs at the house.

Sign conventions differ between trades, which is why on Measure's own output the convention is stated rather than assumed: positive means the ground rises away from the building and water runs at the wall.

Where "5% over 3 meters" comes from

It is a rule of thumb that has become close to universal in residential guidance: get surface water at least three meters from the wall before you stop caring about it, and give it enough fall that it moves rather than sits.

Three meters is roughly the distance over which the backfill trench against a foundation was disturbed during construction, so it is the ground most likely to have settled and the ground where water is most likely to find a path down against the wall. Five percent is enough fall to keep water moving over grass and mulch without being so steep that the lawn erodes or becomes unusable.

It is a target, not a code citation. Local requirements vary, and a site with heavy clay, a high water table or a hard surface abutting the wall may want more. Measure reports slope against this target because it is the number most contractors already carry in their heads.

Why it is a per-wall quantity, not a per-house one

A single "grade" for a property is close to meaningless. Every wall faces a different direction and sits on different ground. A house can be textbook on three elevations and a bowl on the fourth, and the fourth is the one that matters.

So the measurement is done per wall, on that wall's own outward normal — the direction it actually faces — rather than by taking one slope for the parcel.

How the slope is defined

The input is a bare-earth surface: a 3D model of the ground with the trees and buildings stripped out, leaving the shape of the ground underneath. Then, per wall:

• The full first three meters. Slope is measured from 0.75 m to 3.25 m out from the wall — deliberately spanning the same three meters the rule of thumb is about — rather than read off any single pair of points.
• No single reading decides a wall. The measurement is repeated along the wall and combined, so a wheelie bin, a shrub or one bad reading cannot carry it.
• Nothing measured nearer than 0.5 m to the wall. Inside a footprint, a bare-earth surface is not ground — it is the interpolation across a building that was removed, and treating it as ground is one of the commonest silent errors in this field.

One subtlety worth naming, because it is where remote measurement usually oversells itself: nearby readings are not independent observations. Oversampling buys smoothness, which is real, and buys no extra confidence — so the stated uncertainty is computed on the information actually present, not on the count of samples taken.

Before any of it: the data has to be good enough

Not all elevation data deserves to be measured at this scale. The surface is tested first — the median step between adjacent elevations must clear 0.004 m. Below that, the surface is a coarse source stretched onto a fine grid, and a grade computed from it would be interpolation presented as measurement.

Where that gate fails, the honest output is a refusal with the measured step attached, not a confidence downgrade. An address whose ground does not clear the gate is declined outright rather than measured anyway.

The uncertainty nobody else prints

Here is the problem that makes a single decimal figure dishonest. A published building outline is a trace, and it can sit a meter from the wall it claims to be.

Shifting a real Pittsburgh footprint north in 0.2 m steps moved one wall from +5.4% to −23.5%. It crossed zero inside 1.4 m — less than the outline error the data's own limits concede. A reading printed to one decimal place from a single footprint position would have been confidently wrong.

The fix is to move the house. The whole footprint is displaced one meter in each of sixteen directions and every wall is measured again — seventeen positions in all — and a direction is published only when all seventeen agree on the sign.

It does not demand that the magnitudes agree. A meter of movement on steep ground can legitimately swing a grade from −52% to −13%; that says the hillside is not a plane, and nothing about which way the water goes.

Walls that fail this test ship as too close to call, with which way they lean and how many of the sixteen agreed. They are reported as undecided rather than dressed up as a finding.

What negative grading does not explain

Slope is one of four conditions that put water at a wall. The others are how much ground upstream drains to that spot, whether the flow concentrates there, and whether the water has anywhere to spill. A wall with mild adverse grade collecting a hillside is a bigger situation than a steeply adverse wall collecting nine square meters.

And several common causes are not terrain at all: downpipes discharging at the wall, a settled backfill trench, a patio pitched the wrong way, or subsurface conditions no remote measurement can see.