Case study · The Channon, New South Wales
We traced 40 hectares of the world's best-known permaculture demonstration site and read it from satellite and airborne data. No site visit, no inside knowledge, no connection to the farm.
Then we did it twice more, on worse elevation data. This page is the comparison: the same boundary, the same code, read at 30 m, 5 m and 1 m, so you can see what the data quality is worth. The complete read of the farm is the full Zaytuna dossier at 1 m.
Zaytuna has been shaped deliberately for decades, with dams, swales and contour terracing. None of that is in the data we start from.
At 1 metre we can spot them. The dams read as flat pads with a hard rim where the wall drops away. The contour work on the western slope reads as a series of benches. At 5 metres they are faint. At 30 metres the hillside is one smooth shape and the earthworks do not exist.
There is no pattern-matching here. We measure ground, and this ground has been worked.
The same corner of the farm at each resolution. Click any image to open it full size.
Five scores, one for each question a buyer asks. The number is an absolute reading against fixed bands, so it means the same thing on any parcel; where one stands out against every parcel we have read, the assessment says where it stands.
Climate is the measured strength here: 82/100 places this property in the top 6% of all parcels read, carried by 365 frost-free days and 3,699 growing degree days. More surprising is what sits above the boundary. Some 14,605 hectares drain through these 40, roughly 365 times the parcel's own area, and the site sits in the lower third of a basin with 833 m of relief. Water is nonetheless the binding constraint: reference evapotranspiration runs 1,348 mm against 1,062 mm of rain, a 286 mm shortfall spread over seven consecutive deficit months from July. Where should storage sit to hold that inflow before it leaves?
Opening of the written assessment, generated from the computed data
Curious what your own parcel scores?
Trace your parcel · Free Terrain Check →The computed numbers rest on one thing: how good the elevation model is. And to highlight what that means in action, we ran the same boundary three times, same code, same day, changing only the input.
| What was measured | 30 m | 5 m | 1 m |
|---|---|---|---|
| Largest unbroken patch under 7° | 17.33 ha | 5.01 ha | 2.02 ha |
| Steepest point on the gentlest way in | 1.9% | 14.0% | 14.9% |
| Ground under 12°, still workable | 38.72 ha | 28.89 ha | 23.83 ha |
| Reads high on the wetness index | 26.4% | 8.1% | 7.2% |
| Upstream catchment | 14,771 ha | 14,604 ha | 14,605 ha |
At 30 metres the model says 17.3 hectares of flat building ground on a 1.9% approach. At 1 metre it says 2.0 hectares and 14.9%. Above roughly 12% you are paying for an engineered driveway rather than a track.
It is not blur. A 30 metre cell covers 900 square metres, so a gully four metres deep and ten across cannot fit inside one. The model merges separate flat areas across ravines it cannot see. That is where 17.3 hectares comes from.
Coarse elevation does not make land look uncertain. It makes it look better.
Green is gentle, red is steep. The figure is the largest single unbroken patch under seven degrees.
At 30 metres a quarter of the farm reads wet, as one soft blue wash. At 1 metre it resolves into channels with dry ground between them. Those are two different design problems: one says rethink the site, the other says here is where the swale goes.
The 1 metre plate also draws the earthworks themselves, in three grades of certainty. Filled outlines are water the national survey classified as water — six of them here. Dashed outlines are closed hollows we measured in the raw elevation that would hold water whether or not they hold any today; four, the deepest 4.09 metres. Dotted lines are cuts running along the slope. From elevation alone those could be swales, terraces or farm tracks, so the map says all three rather than guessing — three of them, at 1, 14 and 31 degrees off the contour. At 5 metres none of the thirteen appear. At 30 metres, none.
Four more hollows on this parcel are the creek itself, with 121 to 181 hectares draining through them. They are drawn as watercourse, not as storage.
This is a topographic wetness index: a slope-and-flow calculation from the terrain, with no soil moisture measurement and no rainfall in it.
Roughly 146 square kilometres of hill country drains through this farm. The parcel is about 0.3% of its own watershed. Water arrives here whether anyone plans for it or not.
This is the one measurement all three elevation models agree on, within about one percent. A watershed divide is a ridgeline, and a ridgeline is still a ridgeline at 30 metres.
Coarse elevation is fine for how much water arrives. It is unreliable for where you can put a house, a track, or a garden.
The whole basin above the farm, delineated at each of the three resolutions.
The same basin zoomed to the property, showing how the drainage arrives.
Where upslope water crosses the boundary, and how much land feeds each entry.
Aspect is which way each slope faces, and it decides what you can grow where. In the southern hemisphere the productive arc runs west through north through east, and 43.3% of this parcel falls inside it.
Coarse data gives the farm a handful of broad faces. At 1 metre every terrace and gully carries its own orientation, which is the difference between "this hillside faces north" and knowing which bench to plant.
Which way each slope faces, and so how much sun it receives.
The sector plate: summer and winter sun arcs, the SSE prevailing wind at 3.9 m/s, and the fire approach. It reads the terrain for shading and horizon, so it sharpens with the elevation model even though the sun and wind data behind it never change.
Most people read landform from hillshade. Contours are the first thing an earthworks contractor asks for, and altitude banding is how you find the gravity-fed line for a tank. All of them are drawn from the elevation model, so all of them move with it.
Even the satellite plate differs, though the imagery is identical in all three: the contours drawn over it come from the elevation model, and so does the frame.
The climate numbers are identical in all three runs. Nothing in them touches the elevation model.
Ten years of reanalysis, resolved to the parcel.
| Measure | Value | What it means |
|---|---|---|
| Annual rainfall | 1,062 mm | Wettest in March at 195 mm |
| Frost-free days | 365 | No winter dormancy |
| Growing degree days | 3,699 | Subtropical perennials, citrus, macadamia |
| Warmest / coolest month | 24.6 / 13.7 °C | An arc of under 11 °C |
| Solar | 3.0–6.6 kWh/m²/day | Winter trough in June |
| Evapotranspiration | 1,348 mm | Against 1,062 mm of rain |
| Water balance | −286 mm | Seven consecutive deficit months from July |
| Prevailing wind | SSE, 3.9 m/s | Closer to a sea breeze than a wind problem |
Rainfall is generous, but evapotranspiration exceeds it by 286 mm across the year. On a property receiving 365 times its own area in drainage, the task is holding water through the dry months.
The climate plate at all three resolutions; only the terrain inset and the frame differ.
This is terrain read from above. It says nothing about soil chemistry below the top horizon, water rights, easements, covenants, planning permission, the condition of the public road, or anything else that needs a person standing on the ground.
The soil read here is a national 90 m grid, good for texture and pH direction and not a substitute for a pit. The flood read leans on a global 1 km model because no parcel-scale flood register covers this location.
An on-site inspection is recommended before purchase or major development.
Draw any rural parcel and the free Terrain Check reads it the same way: the five scores, one satellite map, and the verdict’s opening. No account. Every map and the complete written assessment come in the full dossier, at €249.
Trace your parcel · Free Terrain Check →