# Methodology

**GB Land Use Framework — an interactive bivariate map of land use potential
across Great Britain.**

This document sets out how every figure in the tool is produced: what is
measured, from which source, how each measurement is banded, how land area is
counted, and where the method is weaker than it looks. It is written to be read
on its own. Where a choice was mine rather than a published standard, it says so.

**This is an independent tool. It is not published by Defra, and it is not
endorsed by Defra.** It reconstructs, from open data, the kind of bivariate
land-use comparison set out in Defra's Land Use Framework analytical annex. That
annex publishes figures but not a method: it states only that suitability maps
were overlaid and pairs of land uses plotted, without a grid size, a
normalisation rule or a classification. Every specification below is therefore
this project's, and should be attributed here rather than to government.

---

## Contents

1. [The hexagonal grid](#1-the-hexagonal-grid)
2. [Indicators and sources](#2-indicators-and-sources)
3. [Farm productivity](#3-farm-productivity)
4. [Prime farmland](#4-prime-farmland)
5. [Farmland extent and its components](#5-farmland-extent-and-its-components)
6. [Woodland creation potential](#6-woodland-creation-potential)
7. [Habitat creation potential and core habitat restoration](#7-habitat-creation-potential-and-core-habitat-restoration)
8. [Peatland extent](#8-peatland-extent)
9. [Solar yield potential](#9-solar-yield-potential)
10. [Banding](#10-banding)
11. [Scoring and weighting](#11-scoring-and-weighting)
12. [Land area and hectare reporting](#12-land-area-and-hectare-reporting)
13. [Cross-border comparability](#13-cross-border-comparability)
14. [Unmeasured ground](#14-unmeasured-ground)
15. [Local planning authority](#15-local-planning-authority)
16. [The coastline outline](#16-the-coastline-outline)
17. [Internal consistency checks](#17-internal-consistency-checks)
18. [Licensing and attribution](#18-licensing-and-attribution)

---

## 1. The hexagonal grid

The unit of analysis is a **5 km flat-topped hexagon on the British National
Grid (EPSG:27700)**, 21.65 km² in nominal area. **11,593 hexagons** contain land
in Great Britain.

Hexagons are generated across the national extent, intersected with the
full-resolution ONS coastline, and retained where at least 5% of the hexagon is
land; the threshold removes ragged marine slivers without discarding small
islands. Each hexagon is attributed to the country covering the largest share of
it. The intersection also yields `land_frac`, the proportion of the hexagon that
is land, which is retained and used throughout for area reporting (§12).

**Why 5 km.** The June Agricultural Survey, the highest-resolution national
source of agricultural activity data, is released aggregated to 5 km to preserve
respondent confidentiality. Any grid finer than that could not be populated from
it consistently, so 5 km is the binding resolution constraint on this class of
analysis. Derived gridded agricultural products are issued at 10, 5 and 2 km,
and 5 km is the finest of those that is unrestricted.

**Resolution caveat.** A 5 km hexagon is a coarse unit for land-use decisions. It
is appropriate for national comparison and strategic pattern-finding. It is not
appropriate for site assessment, and no figure here should be read as a statement
about a particular field or holding.

---

## 2. Indicators and sources

Eleven indicators are published. Each is computed independently and then banded
(§10).

| Indicator | Data column | Unit | Source |
|---|---|---|---|
| Farm productivity | `food_score` | score 0–1 | Provisional ALC (England); Predictive ALC Map 2 (Wales); Hutton Land Capability for Agriculture (Scotland) |
| Prime farmland | `pct_bmv` | % of farmland | as above |
| Farmland extent | `pct_agri` | % of land area | UKCEH Land Cover Map 2024 |
| Arable and horticulture | `pct_arable` | % of land area | UKCEH Land Cover Map 2024 |
| Improved grassland | `pct_improved` | % of land area | UKCEH Land Cover Map 2024 |
| Rough grazing | `pct_rough` | % of land area | UKCEH Land Cover Map 2024 |
| Woodland creation potential | `tree_potential` | % of land area | Woodland Creation Sensitivity Map v4 (England); Small Grants woodland creation area (Wales); Hutton Land Capability for Forestry (Scotland) |
| Habitat creation potential | `habitat_potential` | % of land area | Habitat Networks Combined (England); Priority Ecological Networks 2023 (Wales) |
| Core habitat restoration | `habitat_core` | % of land area | Habitat Networks Combined (England) |
| Peatland extent | `peat_extent` | % of land area | England Peat Map 2025; Carbon and Peatland 2016 (Scotland); Unified Peat Map 2019 (Wales) |
| Solar yield potential | `solar_yield` | kWh/kWp/yr | PVGIS 5.3 / SARAH-3, EC Joint Research Centre |

**Two indicators are single-product and therefore internally consistent across
Great Britain**: farmland extent with its three components, and solar yield. Every
other indicator is assembled from separate national surveys that do not share a
definition. §13 sets out what that permits and forbids.

---

## 3. Farm productivity

**What it measures.** The agricultural quality of the farmland within a hexagon,
expressed on a 0–1 scale. It is a capability measure, not an output measure: it
answers whether land could produce food well, not whether it currently does.

**Construction.** Every parcel of farmland in Great Britain carries an official
capability grade. England and Wales use the Agricultural Land Classification
(ALC, MAFF 1988), with six effective grades: 1, 2, 3a, 3b, 4, 5. Scotland uses
the Land Capability for Agriculture classification (LCA, Macaulay/Hutton 1981),
with seven classes and divisions within several of them.

Each grade is assigned a value between 0 and 1, and each hexagon takes the
area-weighted mean of those values across its farmland. Land that is not
agricultural — settlement, water, woodland, quarries — is excluded from the mean
rather than scored zero, so the indicator reports the quality of the farmland
present and not how much of it there is. Hexagons under 5% agricultural land have
the indicator nulled rather than published as noise.

**The grade-to-value scale is deliberately uneven.** The agronomic distance
between grades 1 and 2 is smaller than that between 3b and 4, and an evenly
spaced scale would misstate it.

| Value | England and Wales | Scotland |
|---|---|---|
| 1.00 | Grade 1 | Class 1 |
| 0.80 | Grade 2 | Class 2 |
| 0.60 | Grade 3a | Class 3.1 |
| 0.40 | Grade 3b | Class 3.2 |
| 0.20 | Grade 4 | Class 4.2 |
| 0.05 | Grade 5 | Class 6.1 |

**The ALC–LCA crosswalk is an interpretive judgement, not an official
conversion.** No official mapping between the two classifications exists, and
both the Scottish and Welsh government reviews state that direct comparison
between them is not meaningful. The crosswalk here anchors on the single point
where the two systems' policy definitions coincide: "best and most versatile"
land in England and Wales is ALC grades 1, 2 and 3a, and "prime agricultural
land" in Scotland is LCA classes 1, 2 and 3.1. The two scales are pinned to each
other there and interpolated within.

**Where the crosswalk strains.** LCA devotes classes 5, 6 and 7 to grades of
rough grazing, because so much of Scotland is rough grazing; ALC compresses all
of that into grade 5. Scottish variation in poor land is therefore flattened.
Cross-border comparison on this indicator is sound at the top of the range and
progressively less meaningful towards the bottom.

**National results.** 50.9% of England's agricultural land is best and most
versatile (49,265 km² of 96,872 km²). Wales returns 13.1% against the Welsh
Government's stated range of about 10–15%. Scotland returns 6.3% prime land
against an approximate published 7%.

The England figure carries a caveat. Earlier Natural England guidance implied
grades 1 and 2 at about 21% of farmland and 3a at about 21%, giving roughly 42%.
This analysis returns 23.2% for grades 1 and 2, which is close, and 27.6% for
3a, which is well above. That is consistent with the documented revision in the
Provisional map — Natural England state that about 20% of land improved by one
grade — but no national breakdown has been published for the revised map, so
50.9% is unverified and should not be cited as authoritative.

---

## 4. Prime farmland

**What it measures.** The share of a hexagon's farmland carrying a statutory
quality designation: "best and most versatile" (ALC grades 1, 2 and 3a) in
England and Wales, or "prime agricultural land" (LCA classes 1, 2 and 3.1) in
Scotland.

It is a percentage of the hexagon's **farmland**, not of its total area. A
hexagon that is 90% urban and 10% grade 1 farmland returns 100%.

**Policy relevance.** Both designations attract a presumption against
development. In England, a planning authority must consult Natural England on
non-agricultural development resulting in the loss of more than 20 hectares of
best and most versatile land where that land is not allocated in the development
plan. In Scotland, National Planning Framework 4 Policy 5 supports development on
prime land only for essential infrastructure with a specific locational need,
small-scale development tied to a rural business or croft, processing facilities
for the land's own produce, or renewable energy generation and mineral extraction
with secure provision for restoration. That last exception is material to this
tool, because it means a hexagon returning high prime farmland is not thereby
closed to the solar development the solar indicator scores.

**Limitation.** The indicator carries the combined share of the three qualifying
grades and does not distinguish grades 1 and 2 from grade 3a, or classes 1 and 2
from class 3.1. It reports how much designated land is present, not which grade
it is.

---

## 5. Farmland extent and its components

**Source.** UKCEH Land Cover Map 2024: satellite imagery classified to 21 UK BAP
Broad Habitat types at 25 m resolution, covering all of Great Britain on one
consistent method.

Three groups are extracted, each published as an indicator in its own right,
because which kind of farming a place supports is what determines what else the
land could do:

| Component | LCM classes | GB area |
|---|---|---|
| Arable and horticulture | 3 | 49,544 km² |
| Improved grassland | 4 | 71,121 km² |
| Rough grazing | 5, 6, 7, 9, 10, 11 | 53,677 km² |

Rough grazing aggregates acid, calcareous and neutral grassland, heather, heather
grassland and bog — six semi-natural classes that carry extensive livestock.

**Farmland extent is a weighted composite.** Arable, horticulture and improved
grassland count in full; rough grazing counts at half. Coastal water and the
foreshore are excluded from the denominator; inland water counts as land that is
not farmed. §11 sets out the weighting and its consequences.

**Overlap with peatland.** Rough grazing correlates **+0.74** with peatland
extent, because heather, moor and bog largely sit on peaty soil. Bog alone
returns +0.57, and removing bog from the group only brings the correlation down
to +0.56, because heather and acid grassland are also on peat. A bivariate
comparison of farmland against peatland is therefore partly measuring the same
ground twice, and is stated as such in the tool.

---

## 6. Woodland creation potential

**What it measures.** The share of a hexagon assessed as suitable for new
woodland planting.

**This is the least internally consistent indicator in the tool.** The three
national surveys behind it do not measure the same quantity:

| | Source | What it measures |
|---|---|---|
| England | Woodland Creation Sensitivity Map v4 | Absence of constraint |
| Wales | Small Grants woodland creation eligibility area | Absence of constraint, scoped to one grant scheme |
| Scotland | Hutton Land Capability for Forestry | How well trees would grow |

England and Wales map where planting faces fewest obstacles. Scotland maps
capability. These are different questions, and absolute values must not be
compared across the border.

**Weighting.** Land facing fewest obstacles counts in full; land where planting
is possible but more constrained counts at half. In England that is the low and
medium sensitivity classes; in Scotland, classes F1–F4 and class F5.

| Scotland, LCF class | Description | km² | Weight |
|---|---|---|---|
| F1 | Excellent flexibility for trees | 767 | 1.0 |
| F2 | Very good flexibility | 4,012 | 1.0 |
| F3 | Good flexibility | 6,229 | 1.0 |
| F4 | Moderate flexibility | 10,323 | 1.0 |
| F5 | Limited flexibility | 13,659 | 0.5 |
| F6 | Very limited flexibility | 19,152 | 0 |
| F7 | Unsuitable for trees | 21,589 | 0 |

England's sensitivity map carries four classes, of which medium is larger than
low:

| Class | Polygons | Area |
|---|---|---|
| Unsuitable | 310,562 | 3,314,664 ha |
| Low | 203,645 | 2,854,532 ha |
| Medium | 205,019 | 3,780,625 ha |
| High | 993,486 | 3,096,414 ha |

Medium is land where planting meets more constraints but remains possible.
Counting only low would state that such land cannot be planted, which is not
what the survey says; counting it in full would rank constrained land alongside
unconstrained land. Half weight is the compromise, and it is a judgement rather
than a measurement. The two classes are disjoint subsets of one survey, and the
pipeline asserts that low plus medium never exceeds 100% of any hexagon.

**Wales has no middle band, because Wales publishes none.** The Small Grants
layer carries geometry and an object identifier and no class field, and Natural
Resources Wales publishes no medium-sensitivity counterpart. Wales is therefore
the only country whose figure is a single-band share, which adds a second
asymmetry to the England–Wales comparison on top of the definitional one.

| | median | mean | hexagons at zero |
|---|---|---|---|
| England (low + half medium) | 33.4% | 32.1% | 632 of 6,293 |
| Wales (low sensitivity only) | 43.1% | 40.2% | 51 of 1,051 |
| Scotland (F1–F4 + half F5) | 14.9% | 28.4% | 1,161 of 4,249 |

**Circularity check.** England's sensitivity map is a composite of roughly 23
constraint datasets, several of which are also inputs to other indicators here —
including best and most versatile agricultural land, and the priority habitat and
protected-site designations behind the habitat indicators. Any trade-off plotted
against this axis might therefore be constructed rather than observed. Measured
across 6,258 English hexagons, the effect runs in the opposite direction to
circularity:

| Pair | Pearson | Spearman |
|---|---|---|
| `pct_bmv` × `tree_potential` | +0.193 | +0.232 |
| `food_score` × `tree_potential` | +0.306 | +0.260 |
| `habitat_potential` × `tree_potential` | −0.067 | −0.071 |
| `habitat_core` × `tree_potential` | −0.220 | −0.208 |
| `pct_bmv` × `habitat_potential` | −0.234 | −0.245 |
| `food_score` × `habitat_core` | −0.327 | −0.304 |

Circularity would show good farmland depressing the woodland score. It raises it.
The mechanism is visible when the two sensitivity classes are separated:

| Pair | Pearson | Spearman |
|---|---|---|
| `pct_bmv` × low sensitivity | −0.014 | +0.135 |
| `pct_bmv` × medium sensitivity | **+0.389** | **+0.413** |
| `food_score` × low sensitivity | +0.079 | +0.175 |
| `food_score` × medium sensitivity | **+0.432** | **+0.450** |

The medium band is disproportionately good farmland, which is not an artefact:
land that is easiest to plant other than for the constraint of being good
farmland is, necessarily, good farmland. The consequence for interpretation is
that the food–woodland comparison shows a sharper trade-off than the low-only
measure would, and should not be described as showing independence.

The residual circularity risk is small: agricultural land classification is one
input among roughly 23, the composite is dominated by designations, a 2 ha
minimum block size disperses it further, and the farm productivity indicator
measures quality while the woodland indicator measures area, so the two do not
share a denominator.

**The genuine observed trade-off** in this dataset is farm productivity against
core habitat restoration, at −0.30 to −0.33 across England. That is a
relationship between separately constructed datasets. The broader habitat
networks indicator correlates with farm productivity at only −0.17 across England
and Wales, and the two figures should not be substituted for one another.

---

## 7. Habitat creation potential and core habitat restoration

**Source.** Natural England's Habitat Networks (Combined) for England, and Natural
Resources Wales' Priority Ecological Networks 2023 for Wales. The English dataset
carries 31 classes; those describing where habitat could be created or restored
are retained, and those recording habitat already present — named priority
habitats, SSSI — are dropped, since the indicator measures opportunity rather
than existing stock.

Retained classes: Network Enhancement Zone 1, Network Enhancement Zone 2, Network
Expansion Zone, Restorable Habitat, Habitat Restoration-Creation, Other Priority
Habitats – NEZ, Other Priority Habitats – NEX. This is 635,192 of 813,420
polygons. Classes overlap, so coverage is measured by burning them to a 25 m grid,
which unions them, rather than by dissolving the polygon set.

- **Habitat creation potential** is the share of a hexagon under any
  creation or restoration class.
- **Core habitat restoration** is the share under Network Enhancement Zone 1,
  Restorable Habitat and Habitat Restoration-Creation only — a conservative
  subset comprising land immediately adjoining existing habitat and degraded
  habitat capable of restoration.

**Both indicators measure opportunity, not delivery.** Land inside an opportunity
zone is land where creation or restoration would extend or connect existing
habitat. It is not land under conservation management. A high reading is a
statement about headroom, not about achievement, and the two must not be
conflated in reporting.

**Scotland is not covered, and no substitute is used.** The national Habitat Map
of Scotland records existing habitat rather than restoration potential. Two
candidate substitutes were assessed and rejected on evidence:

*CSGN Habitat Networks and Opportunity Areas* (NatureScot) is structurally the
closest equivalent to the English dataset and asks the right question, but covers
only the Central Scotland Green Network. Of 4,249 Scottish hexagons, 1,120 fall
within its extent and 659 carry any network cover — 15% of Scotland. Populating
the central belt and leaving the Highlands empty would present an unmapped area
as an absence of opportunity.

*The Big Biodiversity Layer One* (NatureScot) is national, at 73,947 1 km pixels,
and returns a value for 4,243 of 4,249 hexagons. It is a composite heatmap of
biodiversity value. Because CSGN is regionally valid, it can test the national
layer where the two overlap:

| Big Biodiversity Layer against | Pearson | Spearman | n |
|---|---|---|---|
| CSGN opportunity areas | +0.091 | +0.047 | 1,120 |
| CSGN habitat networks | +0.197 | +0.146 | 659 |

Mean CSGN network cover by quartile of the national layer runs 28.5%, 27.4%,
27.4%, 38.2%: the bottom three quartiles are indistinguishable. For scale, an
independently constructed relationship in this dataset — peatland against
woodland creation in Scotland — returns −0.527. The national layer measures
something real, but not where habitat could be created.

NatureScot's Nature Networks Tool contains Wildlife Connectivity Maps and
Opportunity Maps with national coverage, which would complete the indicator. They
are distributed to local authorities as a tool rather than published as data.

**England and Wales are not the same measurement either.** England's graded
enhancement, expansion and restoration zones cover substantially more ground than
Wales' protected-site connectivity networks, and the difference is largely one of
construction rather than of ecology.

| | median | mean |
|---|---|---|
| England (Habitat Networks) | 45.6% | 44.7% |
| Wales (Priority Ecological Networks 2023) | 18.6% | 27.0% |

---

## 8. Peatland extent

**What it measures.** The share of a hexagon's land area mapped as peaty soil.

| Country | Source | Licence | This analysis | Published |
|---|---|---|---|---|
| England | England Peat Map: Peaty Soil Extent (2025) | OGL v3 | 8.2% | ~8.5% |
| Scotland | Carbon and Peatland 2016, classes 1–3 | OGL v3 | 25.6% | ~22.9% |
| Wales | Unified Peat Map 2019 | PSGA, non-commercial | 4.0% | ~4.3% |

All three fall within a few points of their country's published national total,
which is the strongest external validation any indicator here carries.

Scotland's source is a soil map carrying a peatland importance class rather than
a peat extent layer: classes 1 and 2 are nationally important carbon-rich soils
and deep peat, class 3 is carbon-rich soil and deep peat without the priority
habitat, and classes 4 and 5 are mineral soils and are excluded.

**Extent is not depth, and depth is what regulation turns on.** The regulated
category in England is deep peat, defined at 40 cm and reduced to 30 cm by the
Heather and Grass etc. Burning (England) (Amendment) Regulations 2025 (SI
2025/1000), in force from 30 September 2025. The England Peat Map's peaty soil
layer is defined by organic content of at least 20%, not by depth. This indicator
therefore cannot be read against the regulatory definition, and any peatland
percentage should be quoted with its definition attached.

**Independent validation of the woodland indicator.** Planting on deep peat
releases more carbon from the drained ground than the trees sequester, so a
credible woodland opportunity surface should avoid peat. All three do, and all
three were built by different bodies using different methods from the peat maps
used here:

| | peatland × woodland creation |
|---|---|
| England | −0.441 |
| Scotland | −0.527 |
| Wales | −0.405 |
| Great Britain | **−0.480** |

The Scottish woodland source measures growth capability rather than constraint
and still avoids peat as strongly as the other two. This is the best available
evidence that the woodland indicator, weak as it is across borders, measures
something real within each country.

Peatland also runs strongly against farm productivity (−0.556) and solar yield
(−0.502), both of which follow from peat being upland and northern.

---

## 9. Solar yield potential

**What it measures.** Modelled annual electricity generation from an optimally
tilted, free-standing 1 kWp photovoltaic system, in kWh/kWp/yr, from PVGIS 5.3
using the SARAH-3 irradiance dataset. Horizon shading from the model's own
digital elevation model is enabled, so deep valleys and north-facing slopes are
penalised.

**This maps the resource. It does not map where a solar farm could be built.**
The binding constraint on deployment in Great Britain is distribution network
capacity, for which no consistent national open dataset exists — the operators'
capacity heatmaps are separately defined and separately updated. Planning
designations, land ownership and grant eligibility are likewise absent. Nothing
in this indicator knows where the grid has headroom.

**Sampling.** PVGIS is a point API, so this indicator is a **sample rather than a
hexagon mean**. That is defensible here and would not be for the other
indicators: irradiance varies by roughly 1 kWh/kWp per kilometre, so within-hexagon
spread is far below between-hexagon signal.

The sample point is not the hexagon centre. 979 of the 11,593 hexagon centres
fall in the sea, 615 of them Scottish, and the API returns an error for a marine
point; sampling at centres would have silently dropped every island and much of
the west coast. Each hexagon is sampled at a representative point of its
intersection with land. All 11,593 points returned a value.

**Interpretation caveat.** Irradiance over Great Britain is a smooth north–south
gradient, from roughly 1,050 kWh/kWp on the south coast to about 650 in Shetland.
Read alone, this indicator approximates a map of latitude, and it will correlate
with anything else that varies north to south.

| Pair | Pearson |
|---|---|
| `solar_yield` × northing | −0.837 |
| `solar_yield` × `food_score` | **+0.765** |

The best land for solar is the best land for food: both are southern and eastern,
so they compete for the same ground. That is a policy-relevant observation, but
it is not evidence of a mechanism — much of the correlation is the two variables
sharing a latitude gradient rather than anything about soil.

**Validation.** No published per-hexagon figure exists to check against, so the
surface is checked for the gradient it must exhibit:

| Band | mean kWh/kWp/yr | | Place | mean kWh/kWp/yr |
|---|---|---|---|---|
| South coast | 1,068 | | Kent | 1,115 |
| Southern England | 1,016 | | Cornwall | 1,066 |
| Midlands | 973 | | Norfolk | 1,047 |
| Northern England | 912 | | Yorkshire | 972 |
| Southern Scotland | 859 | | Edinburgh | 888 |
| Highlands | 794 | | Inverness | 824 |
| Far north | 789 | | Shetland | 750 |

Kent above Cornwall is correct: the south-east is the sunniest part of Britain,
not the south-west. The 471–1,202 range across 11,593 hexagons is consistent with
published UK figures for optimally tilted systems.

---

## 10. Banding

Each indicator is divided into three bands — **high, medium and low** — on **two
fixed cut points stated in that indicator's own units**. A hexagon is high at or
above the upper cut, low below the lower cut, and medium between them. The cut
points are carried in the data payload, so the map and each indicator's
explanatory panel cannot quote different lines.

**Three properties follow from using fixed cuts rather than ranked groups.**

First, band sizes are unconstrained. A band may hold most of the country or very
little of it, and does: 88% of Scotland falls in the low band for farm
productivity, and 90% of England falls in the low band for rough grazing. A
classification that assigned a fixed share of hexagons to each band — terciles,
quartiles — could not express either statement, because it describes a hexagon's
position within a distribution rather than its absolute quality.

Second, equal values necessarily share a band, without needing a tie rule.

Third, a hexagon's band depends on nothing but its own value. Extending the grid
or revising one county's survey cannot move a band elsewhere in Britain. The
data therefore carries one banding, not one per geographic scope.

**The cost.** Fixed cuts spend the three colours unevenly. A band holding 3% of
hexagons is close to invisible at national zoom, and a band holding 70% renders
most of Britain in one shade, discarding information the three-colour scheme
exists to carry. This is a deliberate trade of legibility for accuracy.

### 10.1 The cut points

| Indicator | Low → medium | Medium → high |
|---|---|---|
| Farm productivity | 0.40 | 0.60 |
| Prime farmland | presence | 50% |
| Farmland extent | 30% | 69% |
| Arable and horticulture | 21% | 50% |
| Improved grassland | 32% | 50% |
| Rough grazing | 24% | 50% |
| Woodland creation potential | 16.5% | 50% |
| Habitat creation potential | 7% | 30% |
| Core habitat restoration | 7% | 30% |
| Peatland extent | 8.5% | 50% |
| Solar yield potential | 934 | 1,000 |

### 10.2 The authority of each cut point

**This is the most important table in this document.** The twenty-two cut points
do not carry equal weight, and the difference should be read before any band is
relied on.

| Grade | Meaning | Count |
|---|---|---|
| Policy | A line drawn by government in legislation, a statutory target or adopted national planning policy | 2 |
| Policy (approximated) | The boundary is genuine policy; the number standing for it is this project's translation of it | 1 |
| Classification | A boundary defined within the source survey, carrying no policy consequence | 1 |
| Statistic | A published national measurement used as a line, or where none matched, the indicator's own British mean | 7 |
| My assumption | No line existed; one is adopted here and labelled as mine | 11 |

| Indicator | Low → medium | Medium → high |
|---|---|---|
| Farm productivity | classification | policy (approximated) |
| Prime farmland | **policy** | my assumption |
| Farmland extent | my assumption | statistic |
| Arable and horticulture | statistic | my assumption |
| Improved grassland | statistic | my assumption |
| Rough grazing | statistic | my assumption |
| Woodland creation potential | **policy** | my assumption |
| Habitat creation potential | my assumption | my assumption |
| Core habitat restoration | my assumption | my assumption |
| Peatland extent | statistic | my assumption |
| Solar yield potential | statistic | statistic |

Note that on woodland creation the policy line is the **lower** cut and the upper
one is a bare convention — the reverse of most indicators. This is the reason the
two cut points are justified separately below rather than together.

### 10.3 Justification, cut by cut

**Farm productivity — upper cut, 0.60, policy (approximated).** Set at the
statutory floor of protected agricultural quality. The designations are genuine
policy: best and most versatile land comprises ALC grades 1, 2 and 3a, and prime
agricultural land comprises LCA classes 1, 2 and 3.1, and both attract a
presumption against development. The *number* is not policy. The 0.60 equivalence
is this project's own crosswalk (§3), and no official conversion between the two
classifications exists. What is being asserted is that both jurisdictions draw a
protection line at the top of their own third grade; expressing that as a single
value on a shared scale is a modelling convenience.

**Farm productivity — lower cut, 0.40, classification.** Set at the internal
division of grade 3, the last boundary either survey defines in terms of cropping
capability rather than grazing. It corresponds to ALC grade 3b and LCA class 3.2.
Below it sit ALC grades 4 and 5 and LCA classes 4.1 and below, which both surveys
describe in terms of grass and rough grazing. No policy consequence attaches to
this boundary in either jurisdiction; it is used because it is the point at which
both systems independently stop describing land as arable.

**Prime farmland — lower cut, presence, policy.** The low band is absence rather
than a value. Both designations are binary in law and the planning tests engage
on presence rather than proportion, so a hexagon holding none is making a
different statement from one holding a little, and that boundary is drawn by
policy rather than by me. The one quantitative threshold in this area is the 20
hectare statutory consultation trigger described in §4. Tested against it,
presence is a close proxy for materiality at this resolution: estimating
designated area as hexagon land area × farmed share × designated share of
farmland, the median hexagon holding any designated land holds approximately 786
hectares of it, and only 349 of the 7,100 such hexagons (4.9%) fall below the
trigger. The estimate is approximate, because the capability surveys and the land
cover map do not draw the farmland boundary identically, but the margin is wide
enough that the conclusion does not depend on it. The trigger itself is English;
Scotland's protection runs through NPF4 Policy 5 with no hectare threshold, so the
materiality test corroborates the boundary in England only. The boundary itself
holds in all three countries, because both designations are binary in both
systems.

**Prime farmland — upper cut, 50%, my assumption.** No published threshold states
how much of a landscape should be top grade. A majority test is the plainest
reading available. This is the weakest pair of cut points in the tool alongside
the habitat indicators.

**Farmland extent — upper cut, 69%, statistic.** The share of the United Kingdom
land area recorded as utilised agricultural area: 16.76 million hectares of
approximately 24.2 million in 2024. A hexagon at or above it is under agricultural
cover at least as extensively as the national land budget. Two qualifications.
The anchor moves — utilised agricultural area was 17.2–17.5 million hectares and
70–72% of UK land between 2016 and 2022 — so a threshold pinned to it is fixed
only between publications. And the instruments differ: the national figure is
compiled from holding-level returns, while this indicator is a satellite land
cover classification weighted for rough grazing. It is a reference point rather
than a like-for-like comparison.

**Farmland extent — lower cut, 30%, my assumption.** No published threshold
separates a predominantly agricultural landscape from a mixed one. Below roughly
three tenths, agricultural cover is not the leading land use. Set at a round
number rather than a derived one so that its arbitrariness is visible.

**Arable, improved grassland and rough grazing — upper cut, 50%, my assumption.**
A dominance test, asserting only that more of the hexagon's land area is under
this cover than under everything else combined. No published standard defines
land cover dominance. The same test is used on all three so that a high reading
means the same structural thing on each.

**Arable, improved grassland and rough grazing — lower cuts, the indicator's own
British mean (21%, 32%, 24%), statistic.** Defra's crop and grassland areas are
compiled from holding-level returns and do not correspond to the satellite land
cover classes used here; the common rough grazing series covers only land held in
common and is substantially narrower. Applying an external ratio would import a
definition the data does not use. These cuts are objective but circular: they
describe this dataset rather than an independent standard, and they move when the
land cover map is reprocessed. They are computed at build time so that the
published figure and the figure the map uses cannot diverge.

**Woodland creation potential — lower cut, 16.5%, policy.** England's statutory
target for woodland and tree canopy cover: at least 16.5% of all land in England
by the end of 31 December 2050, under regulation 3(2) of the Environmental
Targets (Woodland and Trees Outside Woodland) (England) Regulations 2023 (SI
2023/90), a long-term target made under section 1(1) of the Environment Act 2021,
with progress calculated by the Forestry Commission. Current cover on the same
measure is 14.5%. A hexagon at or above the threshold contains enough plantable
land to accommodate the national target within its own boundary.

**This threshold is valid in England only, and the limitation is material.**
Scotland and Wales set planting-*rate* targets — Scotland 18,000 hectares of new
woodland a year, Wales approximately 2,000 against an aspiration of 100,000
hectares by 2030 — which are flow measures and cannot substitute for a stock
measure expressed as a percentage of cover. On Forest Research's woodland area
basis, Scotland already stands at 19% cover and Wales at 15%, both above the
English figure, so applying it there tests nothing. England reads 10% on that
basis against 14.5% on the statutory one, because the statutory measure counts
trees outside woodland and the stock measure does not; the two should not be
compared directly.

**Woodland creation potential — upper cut, 50%, my assumption.** The dominance
test.

**Habitat creation potential and core habitat restoration — 30% and 7%, my
assumption on all four cuts.** The figures are real and correctly quoted: 30% is
the area commitment in Target 3 of the Kunming-Montreal Global Biodiversity
Framework, as adopted for England in Defra's 30by30 land delivery plan, and 7% is
Defra's assessment of the share of England currently meeting those criteria —
SSSIs in favourable or unfavourable-recovering condition at approximately 6.4%
plus national nature reserves and public woodland managed for wildlife at
approximately 0.7%, against roughly 32% judged to meet or have the potential to
meet them. The delivery plan restates this within a tiered framework:
approximately 7% gold tier, 9% silver, and a further 3.5% anticipated.

**Their use here is not licensed by those figures.** Target 3 is a *protection*
commitment covering protected areas and other effective area-based conservation
measures over existing important nature, expressed as a *national aggregate*.
These indicators measure per-hexagon *creation opportunity*. The quantity differs,
the spatial unit differs, and nothing in the target or the delivery plan sets a
per-cell figure. The 7% figure is likewise a national coverage statistic for land
under qualifying management, not a per-hexagon opportunity value. The numbers are
adopted here as assumptions, and the only argument for them is that 30% of a
hexagon is a substantial share and the target makes it a familiar one.

Restoration relates more closely to Target 2 of the same framework — 30% of
degraded ecosystems under effective restoration by 2030 — than to Target 3, and
core habitat restoration is cited against it. Substituting the citation does not
promote the anchor: Target 2's 30% is 30% of *degraded ecosystems*, not of all
land, and remains a national aggregate.

The two indicators share cut points deliberately. Two different quantities cannot
both have the same correct line except by coincidence, so equality here is not a
claim that each is independently right — both are assumptions. Holding the lines
still while the measurement narrows from the whole opportunity network to its
strict core is what makes the two readable against each other, and the fall in
how many hexagons clear the line is the finding.

**Peatland extent — lower cut, 8.5%, statistic.** The share of England's surface
covered by peaty soil according to the England Peat Map published in May 2025 —
the same map this indicator's English component is built from. This indicator's
own mean across English hexagons is 8.2%, which agrees with the published figure
within modelling noise. The published figure is itself a modelled estimate on a
spatially rare feature and carries a non-trivial error band.

**Read the low band carefully on this indicator.** Low means below average, not
absent. A hexagon at 8% still carries approximately 173 hectares of peaty soil,
and the emissions consequence of draining, planting or building on it is not
proportionately small. On every other indicator a low reading indicates the
absence of an opportunity; here it indicates a smaller quantity of something that
matters at any quantity, and the band name understates it.

**Peatland extent — upper cut, 50%, my assumption.** The dominance test. A
stronger anchor exists in the deep peat regulatory definition but cannot be used,
because this indicator carries extent without depth (§8).

**Solar yield potential — upper cut, 1,000 kWh/kWp/yr, statistic.** The level
attained along the English south coast, where Cornwall, Devon and Dorset run at
approximately 1,000–1,050. It marks the top of the national resource. It is not a
viability threshold: no published threshold defines the specific yield at which a
scheme becomes viable, because viability turns on grid connection, land cost and
offtake price alongside irradiance, and irradiance varies least of the four across
Britain.

**Solar yield potential — lower cut, 934 kWh/kWp/yr, statistic.** The indicator's
own British mean. Published UK figures for national average yield measure a
different quantity: they report the realised output of the *installed fleet*,
which is capacity-weighted towards southern England. Using a fleet-weighted figure
as the midpoint of a resource surface would place the line above the centre of the
distribution and systematically over-report the low band. This cut carries the
same circularity as the land cover means, and moves when the model is revised.

### 10.4 Absence as a band

Prime farmland is banded on presence at its lower cut rather than on a value,
because the underlying quantity is a designation a parcel either carries or does
not (§4).

**The simplification is lossy and should be read as such.** The designation is
binary but the policy built on it is not: national planning policy directs that
poorer-quality land should be preferred for development, and higher grades attract
greater weight in the planning balance. A presence test discards that gradient — a
hexagon whose farmland is wholly grade 1 and one with a single qualifying field
both read high. Readers needing the gradient should use the farm productivity
indicator alongside it.

### 10.5 What fixed thresholds do not achieve

They do not make indicators comparable with one another. High on solar yield is a
measured physical quantity; high on prime farmland is the presence of an
interpretive designation; high on habitat creation potential is headroom against
a target rather than delivery against it. The map pairs two indicators at a time
and is still pairing quantities of different kinds. Fixed thresholds fix the
meaning of each band within its own indicator, and nothing more.

---

## 11. Scoring and weighting

Eight of the eleven indicators require nothing between the source survey and the
band. They are a measured share of the hexagon, or a modelled physical value, and
a cut point applies to them directly.

Three carry an intermediate step, and only one of the three is a score.

**Farm productivity is scored, because it must be.** Farmland carries a grade
rather than a number, and a hexagon holds a mixture of grades, so something has
to reduce that mixture to one value before any threshold can act on it. The
grade-to-value scale and the area-weighted mean are set out in §3.

The known objection is that the mean of a non-linear score is not itself a grade:
a hexagon could average 0.60 while holding no grade 3a land at all. Tested against
the designated share, the objection is small in practice — of the 2,283 hexagons
in the high band, 23 hold designated land as a minority of their farmland and none
below 25%, with a median of 92%. The band a hexagon falls in describes the average
of its farmland, not every field within it.

**Farmland extent and woodland creation potential are weighted, not scored.**
Rough grazing counts at half; England's medium woodland sensitivity class and
Scotland's F5 count at half. Nothing is rescaled and no new scale is constructed —
some land is counted once and some counted half, and the result remains a
percentage of the hexagon's land area.

**Farmland extent: why half.** Britain farms two ways and the gap between them is
wide. Arable and improved grassland are sown, fertilised and managed; acid
grassland, heather, moor and bog carry livestock at a fraction of the intensity.
Neither extreme works as a single indicator:

| | England | Scotland | Wales | |
|---|---|---|---|---|
| Intensive cover only | 68.0% | 28.8% | 55.7% | reads Rannoch Moor and the Cairngorms at 0.0% |
| All cover, unweighted | 75.9% | 77.1% | 76.2% | ceases to discriminate between countries |
| **Rough grazing at 0.5** | **72.0%** | **53.0%** | **66.0%** | as published |

Excluding rough grazing presents the exact ground that woodland creation and
peatland restoration compete for as empty. Including it at face value collapses
the three countries into each other. Half is the compromise, and it is stated in
the tool because a reader cannot infer it from the number.

| Place | intensive only | as published |
|---|---|---|
| Fens | 93.7% | 94.0% |
| Lincolnshire | 92.0% | 92.0% |
| Lake District | 1.4% | 46.1% |
| Rannoch Moor | 0.0% | 44.7% |
| Cairngorms | 0.0% | 19.6% |
| Snowdonia | 2.8% | 49.5% |

**The consequence for both weighted indicators.** Farmland extent measures
farming intensity rather than farmed extent, and is not directly comparable with
published utilised agricultural area figures, which count rough grazing in full.
Woodland creation potential is compared against a target for a real area of
canopy, so a weighted share should be read as an indication rather than as an
area that could be planted. Both statements appear in the tool.

---

## 12. Land area and hectare reporting

The tool reports the land area falling in each combination of bands, in hectares.

**Land area is summed per hexagon, from each hexagon's own measured land.** A
hexagon's land area is its nominal 21.65 km² multiplied by `land_frac`, the share
of it that is land, measured when the grid was built by intersecting the hexagon
with the full-resolution coastline (§1). The areas of the hexagons in a band are
added together. There is no whole-hexagon step.

**This matters because 2,385 of the 11,593 hexagons cross the coastline.**
Treating every hexagon as entirely land would count sea with land:

| | hectares |
|---|---|
| Nominal whole hexagons | 25,098,845 |
| Measured land area | **22,981,333** |
| Great Britain, published | ~22,895,200 |

| | measured | nominal | published | error, measured |
|---|---|---|---|---|
| England | 13,032,173 | 13,624,345 | 13,027,800 | +0.03% |
| Scotland | 7,871,894 | 9,199,085 | 7,789,500 | +1.1% |
| Wales | 2,077,266 | 2,275,415 | 2,077,900 | −0.03% |

The nominal method overstates Great Britain by 8.4% and Scotland by 18%, because
Scotland has far more coastline and island per unit of land. The error is
therefore not a uniform inflation but a distortion of comparisons between
countries, and between band combinations: coastal hexagons cluster in
low-productivity ground, so the nominal method inflates the cells describing poor
land far more than those describing good land.

The residual +1.1% on Scotland reflects the coastline used here generalising the
coast slightly differently from national statistics; it is most visible where the
coast is most intricate.

**What a hectare figure means.** A hexagon takes one band from its aggregate
value, so "hectares in the high band" means "land area of the hexagons that are
high". It is not the area of the feature being measured: the high band for
peatland is land where peat covers more than half the hexagon, not the area of
peat itself. Assigning a band to an individual hectare would require re-running
the banding at raster resolution, which is a different measurement.

The land area in every band of every indicator, with the method and the evidence
behind each figure, is published as a spreadsheet linked from beneath the table in
the tool.

---

## 13. Cross-border comparability

Each indicator carries a comparability rating, and the tool states it.

| Rating | Meaning | Indicators |
|---|---|---|
| Good | One product covers all three countries on one method | Farmland extent and its three components; solar yield |
| Partial | The same measurement joined across countries by a documented crosswalk | Farm productivity; prime farmland; peatland extent |
| Poor | Different products answering different questions per country | Woodland creation potential; habitat creation potential |
| England only | No equivalent published elsewhere | Core habitat restoration |

**For indicators rated poor, absolute values must not be compared across the
border.** A difference between England and Scotland on woodland creation
potential is, in the first instance, a difference between a constraint survey and
a capability survey. The same applies to habitat creation potential, where
England's graded zones and Wales' protected-site networks differ mainly in
construction.

Fixed banding does not resolve this. It applies the same cut point everywhere,
which makes the *rule* consistent, but the quantity being cut still differs by
country. This is a limitation of the source data and cannot be corrected
downstream.

---

## 14. Unmeasured ground

**Hexagons with no value for an indicator are drawn in a distinct no-data grey
and are not banded.** They are not treated as zero, and the distinction is
enforced throughout: an absence of survey is not an absence of the thing surveyed.

The summary table reports unmeasured land as its own row rather than omitting it,
so every indicator's rows sum to Great Britain's land area. Unmeasured land is
substantial on two indicators:

| Indicator | Unmeasured | Share of GB land |
|---|---|---|
| Farm productivity, prime farmland | 90,111 ha | 0.4% |
| Farmland extent and components | ~800 ha | <0.1% |
| Habitat creation potential | 7,871,894 ha (Scotland) | 34% |
| Core habitat restoration | 9,949,143 ha (Scotland, Wales) | 43% |

---

## 15. Local planning authority

**Not an indicator.** Nothing is drawn from it, it does not appear in the
indicator selectors, and it carries no band. It is a label on the hexagon readout
— "England – Fenland" — so that a grid reference means something to a reader who
does not work in eastings and northings.

**Source.** ONS Open Geography Portal, Local Planning Authorities (April 2023)
Boundaries UK BFC V2, OGL v3. 379 authorities UK-wide, 368 in Great Britain
(England 309, Scotland 34, Wales 25). The published " LPA" name suffix is
stripped.

**Assignment.** The authority covering the largest area of the hexagon, the same
rule used for country attribution. Because boundaries do not respect a 21.65 km²
hexagon, the assignment was measured before being relied on:

| | hexagons | |
|---|---|---|
| Wholly within one authority | 7,721 | 66.6% |
| Split across two | 3,046 | 26.3% |
| Split across three | 730 | 6.3% |
| Split across four or more | 96 | 0.8% |

The assigned authority holds a mean 92.9% of its hexagon and a median of 100%.
86.2% of hexagons are above three quarters; 1.1% fall at or below half, which is
the only band where naming a single authority is arguable. All 11,593 hexagons
are labelled, and 362 of the 368 authorities appear.

Boundaries are simplified to 50 m before overlay, which against a hexagon 5 km
across is a rounding error. The names ship as a lookup table with one integer per
hexagon rather than a string per row.

---

## 16. The coastline outline

The country boundaries drawn beneath the grid are ONS boundaries simplified to
500 m, with parts under 8 km² dropped, reducing 9.4 MB to 133 kB. They exist so
that the grid has something to sit against.

**This outline is a locator sketch and is used for nothing else.** At 500 m
tolerance the line is inaccurate by up to half a kilometre and several hundred
islands are absent. Country attribution and `land_frac` come from the
full-resolution geometry (§1), never from this file.

---

## 17. Internal consistency checks

The following are asserted in the pipeline or verified against published figures.

| Check | Result |
|---|---|
| Land area by country against published totals | England +0.03%, Wales −0.03%, Scotland +1.1% |
| Peatland extent by country against published totals | Within ~2.7 points on all three |
| Farm productivity, Wales | 13.1% BMV against a stated 10–15% |
| Farm productivity, Scotland | 6.3% prime against an approximate 7% |
| Farmland extent, weighted | 14.75 Mha against a GB utilised agricultural area of roughly 15.7 Mha |
| Farmland extent, unweighted components | 12.07 Mha arable and improved grass against a published ~12 Mha; 5.37 Mha rough grazing |
| Total mapped land excluding coastal classes | 228,774 km² against Great Britain's 229,848 km² |
| Solar yield range | 471–1,202 kWh/kWp/yr, consistent with published UK figures |
| Woodland sensitivity classes | Low + medium never exceeds 100% of any hexagon |
| Band rows in the summary table | Sum to Great Britain's land area on every indicator |
| Every hexagon banded once per indicator pair | 11,543 across the nine cells plus 50 unmeasured = 11,593 |

**Unverified.** England's 50.9% best and most versatile share (§3) has no
published counterpart for the revised map and should not be cited.

---

## 18. Licensing and attribution

Required attribution on the published site:

- Contains OS data © Crown copyright and database right
- © Natural England copyright, `www.gov.uk/natural-england`
- Contains Forestry Commission information licensed under OGL v3
- Contains Welsh Government / Natural Resources Wales information, OGL v3
- Contains James Hutton Institute materials, OGL v3
- Contains NatureScot information, OGL v3
- Office for National Statistics boundaries, OGL v3

All inputs are Open Government Licence v3 or equivalent **except two**:

- **Unified Peat Map of Wales** — Public Sector Geospatial Agreement / Ordnance
  Survey derived. Viewable under OGL, downloadable for non-commercial use only,
  and its published terms are silent on redistributing derived data, which is
  what this tool does when it publishes per-hexagon percentages and offers them
  as CSV. It is included on the basis that this site is non-commercial.
  Aggregating 7,010 polygons to a percentage of a 21.65 km² hexagon is a strong
  derivation and very likely outside what the restriction is aimed at, but that
  is a judgement rather than a licence term. `data@gov.wales` is the contact.
  Dropping Wales from the peatland indicator would cost 1,051 hexagons and
  nothing else.
- **UKCEH Land Cover Map 2024** — free for non-commercial use, not OGL, and it
  requires a registered download.

**This site is non-commercial. Any commercial reuse would have to drop both
inputs or obtain separate permission.**