Land cover heterogeneity in Great Britain as identified in Land Cover
Map 2000
R. A. HILL* and G. M. SMITH
Centre for Ecology and Hydrology, Monks Wood, Abbots Ripton, Huntingdon,
Cambridgeshire PE28 2LS, UK
Land cover is an important component of landscape character. The spatial
configuration and heterogeneity of land cover can influence species distributions
and patterns of biodiversity. Sustainable countryside planning and land
management policy in Great Britain require the mapping of land cover
heterogeneity at a national scale. Here, we map four measures of land cover
heterogeneity across Great Britain using Land Cover Map 2000. We calculate
land cover richness, diversity, evenness and similarity within and between
1 6 1 km grid cells of the British National Grid. From this we are able to identify
assemblages of land cover types that are associated with high or low landscape
heterogeneity, and where they occur geographically.
1. Background and heterogeneity measures
Defining landscape character and how this contributes to more intangible measures
of countryside quality is now seen as essential to sustainable countryside planning
(Haines-Young et al. 2004). In Britain, the Countryside Character Network defines
landscape character as `a distinct, recognisable and consistent pattern of elements in
the landscape that makes one landscape different from another' (see http://
www.ccnetwork.org.uk). Landscape character is about uniqueness. Its assessment is
based on several components and these can be grouped as natural, cultural/social,
perceptual and aesthetic (Swanwick 2002). Land cover, and particularly its spatial
characteristics (e.g. pattern, diversity, scale, configuration), are key elements that
influence landscape character. In addition, there are well known relationships
between landscape-scale patterns and ecological processes which influence
biodiversity at the beta-scale (Tracy and Brusard 1994).
Land Cover Map 2000 (LCM2000) mapped the land cover of Great Britain and
Northern Ireland using, as far as possible, composites of summer and winter
Landsat Thematic Mapper imagery (Fuller et al. 2002). The imagery was classified
using an object-based approach that identified features in the landscape such as
fields, woods and lakes. The minimum mappable unit of LCM2000 is 0.5 ha, which
is approximately equivalent to eight Thematic Mapper pixels. To support reporting
under the UK Biodiversity Action Plan (UK Biodiversity Steering Group 1995),
LCM2000 identified widespread Broad Habitats (Jackson 2000). The 20 Broad
Habitats mapped by LCM2000 are shown in figure 1.
From the full spatial resolution LCM2000 data, the percentage cover of each
Broad Habitat was calculated per 1 km61 km square of the British National Grid
*Corresponding author. Email: rhill@ceh.ac.uk
International Journal of Remote Sensing
Vol. 26, No. 24, 20 December 2005, 5467­5473
International Journal of Remote Sensing
ISSN 0143-1161 print/ISSN 1366-5901 online # 2005 Taylor & Francis
http://www.tandf.co.uk/journals
DOI: 10.1080/01431160500259931
Figure 1. The 20 Broad Habitats mapped in Land Cover Map 2000.
5468 R. A. Hill and G. M. Smith
(BNG). The BNG provided a regular spatial framework in which the land cover
information of over 6.6 million parcels of LCM2000 could be summarized for
calculating heterogeneity measures. By adopting a regular grid it remains possible,
albeit at a 1 km spatial resolution, to extract land cover information and
heterogeneity measures for any area of interest, such as administrative areas, river
catchments or environmental zones. Land cover heterogeneity was examined both
within and between neighbouring 1 km BNG squares, based on the percentage cover
of Broad Habitats. Four measures of heterogeneity were generated.
1. Land cover richness (R): the total number of Broad Habitats per BNG square.
2. Land cover diversity (D): calculated per BNG square using Simpson's Index
(Simpson 1949):
D~1
.X
p2
i 1
where D is Simpson's index, pi is the percentage cover of Broad Habitat i.
3. Land cover evenness (E): calculated per BNG square as:
E~D=R 2
where E is evenness, D is Simpson's index and R is land cover richness.
4. Land cover similarity (S): calculated between neighbouring BNG squares
using the Bray­Curtis similarity index (Bray and Curtis 1957):
S~1{
X
jxi{yij
.X
xizyi  3
where S is the Bray­Curtis measure of similarity, xi is the percentage cover of
Broad Habitat i in one square, and yi is the percentage cover of the same
Broad Habitat in an adjoining square. The similarity score for each 1 km
BNG square was calculated as the average when comparing with the
surrounding eight squares.
Based on 20 Broad Habitats the possible range of scores for each BNG square is 1
to 20 for both land cover richness (R) and diversity (D). Land cover evenness (E )
has a possible range of 1/R to 1 and land cover similarity (S ) of 0 to 1. Compared
with land cover richness (R), which indicates the total number of Broad Habitats per
BNG square, land cover diversity (D) gives more weight to common land cover
types. Thus, D5R only if all land cover types are equally represented. Similarly,
land cover evenness (E ) will have a score of 1 when all land cover types are equally
represented. For land cover similarity (S ), the Bray­Curtis index will equal 0 when
no land cover types are shared and 1 when all land cover types are shared between
neighbouring BNG squares.
2. Results and discussion
The maps of land cover richness, diversity, evenness and similarity at a 1 km scale
for Great Britain are shown in figures 2(a)­(d ).
Perhaps unsurprisingly no BNG square contains all 20 Broad Habitats
(figure 2(a)); the maximum present in one square is 14. Figure 3(a) shows the
histogram of land cover richness. Over 90% of BNG squares have between 3 and 9
Broad Habitats, 72% have between 4 and 7. The average number of Broad Habitats
per 1 km BNG square is 5.4.
GB land cover heterogeneity 5469
The range of land cover diversity scores is 1 to 8.89 (figure 2(b) and cover). The
average diversity score is 2.41; over 90% of BNG squares have a diversity score
between 1 and 4 (figure 3(b)). Compared with the land cover richness scores, this
demonstrates that in most BNG squares one or more Broad Habitats has a limited
coverage. This can be quantified by examining land cover evenness (figure 2(c)). The
range of scores is 0.11 to 1 (figure 3(c)). An evenness score of 1 means that all Broad
Habitats present within a BNG square have the same percentage cover. Less than 1.5%
of BNG squares have an evenness score of 1.0 and all of these have only one Broad
Habitat present. Only 4% of BNG squares have an evenness score .0.8; these have
between 1 and 10 Broad Habitats present. Over 80% of BNG squares have an evenness
score ,0.6. The average evenness score is 0.47. Land cover similarity (figure 2(d)) has a
spatial dimension, being an average score for each BNG square with its eight
neighbours. The range in land cover similarity scores is 0 to 1 (figure 3(d)), with a mean
of 0.66. Thus, as would be expected, most BNG squares are reasonably similar to their
neighbours, with less than 13% having an average similarity score of ,0.5.
(a) (b)
Figure 2. Land cover heterogeneity measures calculated per 1 km BNG square from
LCM2000: (a) richness; (b) diversity; (c) evenness; and (d ) similarity. (Continued opposite.)
5470 R. A. Hill and G. M. Smith
What does this tell us about the British landscape? Certain assemblages of land
cover types occur in landscapes of low heterogeneity. Table 1 gives diversity scores
for BNG squares grouped into eight land cover assemblages. Each BNG square is
assigned to a land cover assemblage based on whichever has the highest total
coverage of the component land cover types. Compared with the national average
diversity score of 2.41, low diversity occurs in BNG squares with a dominant land
cover of `heath, bog, montane' (1.89), `arable and horticulture' (2.23) or `coastal
habitats' (2.29), whilst high diversity occurs where the dominant land cover is
`woodland' (2.94) or `semi-natural grass, bracken, fen, marsh' (2.90). The spatial
configurations these results represent can be seen in figure 2(b) (and cover). Low
diversity occurs in patches along the eastern fringes of Great Britain, most notably
East Anglia and Lincolnshire, associated with intensive agricultural land use. Low
diversity also occurs in upland areas dominated by heaths and bogs; most notably in
the Scottish Highlands and Hebridean Islands and in the Cambrian Mountains of
central Wales. The Flow Country of peatland bog in northern Scotland also has low
land cover diversity. Finally, the major conurbations, in particular London and
Birmingham, also stand out as islands of low land cover diversity compared with the
(c) (d)
Figure 2. Continued.
GB land cover heterogeneity 5471
national average. Areas of high land cover diversity are more diffusely distributed,
but can be found: in the North and South Downs and Weald of south-east England;
around the fringes of the Brecon Beacons in south Wales and of the Peak District
and Pennines in north-central England; in northern County Durham; and the
Carron Valley area north of Glasgow in Scotland. These are areas of semi-natural
grasslands (either calcareous or acidic) and, in particular, a mixture of or transition
between unenclosed semi-natural grasslands and predominantly small-scale enclosed
pasture, often with scattered woodlands. These are the locally heterogeneous
Figure 3. Histograms of land cover heterogeneity measures for Great Britain: (a) richness;
(b) diversity; (c) evenness; and (d ) similarity.
Table 1. Diversity scores for eight terrestrial land cover assemblages and for Great Britain as
a whole.
Land cover assemblage Minimum Maximum Mean Standard deviation
Coniferous, broadleaved, mixed
woodland
1 8.89 2.94 1.19
Arable and horticulture 1 7.25 2.23 0.86
Improved grassland 1 7.11 2.58 0.68
Semi-natural grass, bracken, fens,
marshes
1 8.85 2.90 1.19
Heath, bog, montane habitats 1 7.87 1.89 0.85
Inland water 1 7.40 2.52 1.12
Built up and gardens 1 7.83 2.47 1.03
Coastal habitats 1 7.68 2.29 1.32
GB total 1 8.89 2.41 1.02
5472 R. A. Hill and G. M. Smith
cultural landscapes that score positively in the natural, cultural/social, perceptual
and aesthetic categories of landscape character assessment.
3. Conclusions
The heterogeneity of the British landscape reflects the nature of the environment
and how it has been managed. Understanding and representing this using LCM2000
data could be key to explaining species distributions and patterns of biodiversity in
Britain (Fuller et al. 2005), thereby leading to informed land management policy.
The datasets presented in figure 2 are produced consistently and objectively and
therefore can be a valuable input to national-scale landscape character assessment
and the determination of countryside quality. The approach described in this paper
could be applied to other regionalizations, such as the National Land Classes,
Environmental Zones or Countryside Character Areas. This work shows that
environmental information derived from Earth Observation data at the national
level can support complex multi-disciplinary research and government policy.
Acknowledgement
LCM2000 was produced at CEH Monks Wood by the authors, together with Robin
Fuller, Jane Sanderson, Andy Thomson and Nigel Brown. This work was carried
out under the Defra-funded project: CS2000 Module 9, Data Integration for
Localised Results and Support for Indicators of Countryside Character and Quality.
References
BRAY, J.R. and CURTIS, J.T., 1957, An ordination of the upland forest communities of S.
Wisconsin. Ecological Monographs, 27, pp. 325­349.
FULLER, R.M., DEVEREUX, B.J., GILLINGS, S., AMABLE, G.S. and HILL, R.A., 2005, Indices
of bird-habitat preference from bird field surveys and remote sensing of land cover: a
study of south-eastern England with wider implications for conservation and
biodiversity assessment. Global Ecology and Biogeography, 14, pp. 223­239.
FULLER, R.M., SMITH, G.M., SANDERSON, J.M., HILL, R.A. and THOMSON, A.G., 2002, The
UK Land Cover Map 2000: construction of a parcel-based vector map from satellite
images. Cartographic Journal, 39, pp. 15­25.
HAINES-YOUNG, R., MARTIN, J., TANTRAM, D. and SWANWICK, C., 2004, Countryside Quality
Counts, Tracking Change in the English Countryside: Constructing an Indicator of
Change in Countryside Quality. Final Report (Nottingham: Nottingham University
Consultants Ltd).
JACKSON, D.L., 2000, Guidance on the interpretation of the Biodiversity Broad Habitat
Classification (terrestrial and freshwater types): definitions and the relationships with
other habitat classifications. JNCC Report No. 307 (Peterborough: Joint Nature
Conservation Committee).
SIMPSON, E.H., 1949, Measurement of diversity. Nature, 163, p. 688.
SWANWICK, C., 2002, Landscape Character Assessment: Guidance for England and Scotland
(London: The Countryside Agency and Scottish Natural Heritage).
TRACY, C.R. and BRUSARD, P.F., 1994, Preserving biodiversity: species in landscapes.
Ecological Applications, 4, pp. 205­207.
UK BIODIVERSITY STEERING GROUP, 1995, Biodiversity: The UK Steering Group Report.
Volumes 1 and 2 (London: HMSO).
GB land cover heterogeneity 5473