Information and Advsiory Note Number 67, January 1997
1.1 There is a range of methods currently available for determining the extent and composition of Iandcover, habitat and plant community types. Three examples varying in scale are:
1.2 This Note describes on-going work to provide a comprehensive inventory of the blanket mire resource of Scotland, using remote sensing (satellite imagery), as part of the Scottish Blanket Bog Inventory (SBBI).
1.3 Blanket mire is normally defined as a wetland which supports a surface vegetation which is peat-forming.
1.4 Blanket bog is listed as a ‘priority’ (active bog only) habitat under the EC Habitats Directive and the UK Biodiversity Action Plan. It is the most extensive semi-natural habitat found on land in the UK, and has been sparsely surveyed.
1.5 Blanket mires in Scotland cover just over 1.0 million hectares (1.4 mha in Great Britain). Survey and inventory is currently made difficult because of the remoteness, apparent vegetation homogeneity and featureless nature of the habitat.
2.1 Before outlining the use of satellite imagery for habitat characterisation, we mention other techniques available for surveying and assessing blanket bogs.
2.2 The NVC is being used extensively for mapping particular blanket mire vegetation types throughout specified sites in Scotland. This provides a standard method for classifying vegetation and is used by ecologists throughout Britain. The NVC, however, was not intended as a means for mapping large areas of semi-natural habitats, particularly where mosaics of vegetation types occur. It was, rather, designed to classify vegetation on the basis of a limited sample of plant r’eleves (quadrats - 2 m × 2 m square). It cannot always classify or evaluate previously unsampled vegetation types throughout Scotland (i.e due to the low sample numbers in the original NVC mires survey data for much of Scotland), habitat condition, site hydrology or micro-topography, all of which are important for assessing the scientific and conservation interest of peatlands. For these reasons peatland vegetation is inadequately represented in published NVC maps (Rodwell, 1991; Horsfield et al., 1996). Furthermore, NVC mapping requires a substantial capital investment in the form of personnel and time needed to cover large areas of peatland (one surveyor can map 100-lSOha day−1 of blanket bog). For full details on NVC and peatland mapping techniques make reference to the further reading section (Renwick and Reid, 1997).
2.3 Intermediate scale - landcover survey based on air photo interpretation (API).
There are several examples of this type of Iandcover survey which SNH staff will be familiar with. For further details of these make reference to the further reading at the end of this Note, notably on:
2.4 API, allied with Geographical Information Systems (GIS), can provide a means of assessing the distribution and composition of Iandcover types deemed to be of conservation importance (for example heather moorland) (Thompson et al., 1995). API is also an extremely useful tool for estimating change over time in the appearance of the countryside. The categories of Iandcover definition, however, are quite coarse for peatlands (e.g. heather moorland, unimproved grassland, blanket mire etc.). They do not give much information on the ecological variability within habitats and provide inadequate information on habitat ‘condition’. In the case of blanket bog, the variability across plant communities dominated by different sedges, grasses, mosses and shrub species is critical to the determination of condition and management requirements.
3.1 In 1995 SNH identified the scale of remote sensing systems (e.g. satellite, aerial) which could be used to map different types of semi-natural habitat. Blanket mire was identified as being well suited to large scale mapping by Landsat and SPOT satellites (Pooley and Jones, 1995). This was supported by earlier research by Reid et al., 1994.
3.2 The SBBI uses a methodology developed specifically for mapping the extent, variability, condition and hence conservation interest of the blanket mire habitat throughout Scotland. The method uses data recorded by the Thematic Mapper (TM) Sensor from the Landsat 5 satellite. The significant difference between the approach adopted by the SBBI and other remote sensing methods is that it is designed to map blanket mire condition and extent only, and thus makes no attempt to map other landcover types. Techniques have been developed to interpret the spectral characteristics of the TM imagery allied with different features of blanket mire. (For details on GIS and remote sensing terms in this Note see Gardner, 1996 and Quarmby and Reid, 1997).
3.3 The SBBI favoured the Landsat 5 satellite data and technique developed, because:
3.4 The specific aims of the SBBI analysis and interpretation of each TM image are three fold:
The SBBI plan for completing the above aims are focused on the characterisation of blanket mire across each of the images shown in Figure 2.
4.1 Ground survey and ancillary information
The satellite data used were six Thematic Mapper image bands acquired by the Landsat 5 satellite on 6 June 1992. Survey sites are targeted on the Landsat TM image using local knowledge and an automated classification approach (this is termed the 1st stage classification; Reid et al., 1996). Within these targeted sites, boundaries (lines) are produced around all of the colour classes on the image. These lines (polygons) are transferred onto ‘raw’ satellite imagery where homogeneous areas (blocks) are then marked for ground reference survey collection (Figure 3a and 3b, page 6).
4.1.1 For this scene a total of 139 ground reference areas (blocks) were visited (Figure 4, page 6). A survey data sheet was completed for each area containing information on, for example, NVC type, condition, micro-topography, erosion features and vegetation structure, etc. In total, 30 attributes are collected for each area surveyed.
4.1.2 On completion of survey all information is converted into digital formats.
4.2 Satellite data processing - classification and interpretation
4.2.1 To aid the 2nd stage of the classification process, a number of masks (areas of the imagery which are deliberately excluded from any analysis) of the imagery are produced to restrict the analysis to peatlands. For example, British Geological Soil data are used to highlight the presence or absence of blanket bog (Figure 5, page 6).
4.2.2 In addition, a number of statistical analyses (e.g. Principal Component Analysis) are performed on the imagery to reduce the size of the dataset and in order to remove any correlated or confusing spectral features in the imagery.
4.2.3 The final classification of the image produced output, for Lewis, in three stages:
4.3 Quantitative analysis - accuracy of the classification and ground information
4.3.1 Quantitative analysis of the peatland classification involved:
For details on all of the statistical applications make reference to Quarmby and Reid, 1997. In summary, the following statistical techniques were applied:
4.3.2 Overall class accuracy was 77% (class accuracy ranged between 64-90% correctly classified pixels).
Field measurements of shrub and herb cover, erosion and bare peat from the ground reference survey were calculated to be clearly linked (in the 95% confidence limits) to each of the classes resulting from the classification procedure (Quarmby et al., 1997).
4.3.3 Due to the continuous nature of this semi-natural landscape a classification accuracy would not be expected to be greater than 90%. An accuracy of 77% is therefore highly acceptable and is reflective of the real spatial variability in vegetation cover.
4.4 Descriptions of each class according to the NVC type
4.4.1 The proportions of NVC types in each spectral class are calculated on the basis of the presence or absence of each. For each class, the proportion of each NVC type (Y-axis) has been plotted on a bar chart (Figure 6).
4.5 Final classified map and class descriptions
4.5.1 Using the results from the above analysis, final habitat descriptions are developed which are related to SSSI guidelines and the EC Habitats Directive (HD). A short summary report is produced for each class (colour) across the peatlands in the satellite scene. This provides:
For definitions of peatland technical terms make reference to Lindsay et al. 1988; Rodwell, 1991; Quarmby et al., 1996
M17 = Scirpus cespitosus-Eriophorum vaginatum blanket mire
M19 = Calluna vulgaris-Eriophorum vaginatum blanket mire
Sub-communities:
M17a = Drosera rotundifolia-Sphagnum spp
M17b = Cladonia spp
4.5.2 As a result of the image processing of the island of Lewis and Harris, 7 peatland classes have been determined (Table 1 and Figure 7, page 7).
5.1 The majority of the peatland of Lewis and Harris is composed of active bog (Table 2 and Figure 8, page 7). Spectral classes 1-5 and 7 account for 67% of the total land area covered by this satellite image.
6.1 The SBBI technique of peatland assessment based on satellite imagery has become a valuable and cost effective tool for mapping and quantifying, at a coarse scale, large expanses of blanket bog in Scotland. It could also offer considerable potential for mapping blanket bog habitats in other parts of the E.U. and provides a valuable means of monitoring change between different vegetation types through time.
6.2 By applying this kind of mapping, SNH can identify and then target, for further detailed ground survey, areas of high natural heritage importance. Massive parts of Scotland’s blanket bogs have not been studied in any detail - some have only been walked over in more recent times. This technique therefore provides the first opportunity to determine the full extent and composition of blanket bogs in Scotland,
6.3 The SBBI is currently concentrating on those parts of Scotland which are likely to have substantial areas of active blanket bog but which have been unmapped or poorly surveyed.
European Commission (1994). Manual for the Interpretation of Annex 1 Priority Habitat Types of the Directive 92/43/EEC. European Commission, Directorate-General Environment, Nuclear Safety and Civil Protection. Nature Protection and Soil Conservation (Habitats 94/3 FINAL), Brussels.
Gardner, S. (1996). An Introduction to Geographic Information Systems. Information and Advisory Note No. 34. SNH, Battleby.
Horsfield, D., Thompson, D.B.A., and Tidswell, R. (1996). Revised atlas of upland NVC plant communities in Scotland. Information and Advisory Note No. 56. SNH, Battleby.
Joint Nature Conservation Committee (1994). Supplement to The guidelines for the selection of biological Sites of Special Scientific Interest: bogs, Nature Conservancy Council (1989). Joint Nature Conservation Committee, Peterborough.
Lindsay, R.A., Charman, D.J., Everingham, F., O’Reilly, R.M., Palmer, M.A., Rowell, T.A. and Stroud, D.A. (1988). The Flow Country: the Peatlands of Caithness and Sutherland. Nature Conservancy Council, Peterborough.
Macaulay Land Use Research Institute. (1993). The Landcover of Scotland 1988 Final Report. The Macaulay Land Use Research Institute, Aberdeen.
Pooley, M.R. and Jones, M.M. (1995). Application of Remote Sensing to Habitat Mapping and Monitoring. SNH Review No. 57. SNH, Battleby.
Quarmby, N.A. and Reid, E. (1997). Environmental remote sensing. Information and Advisory Note No. 63. SNH, Battleby.
Quarmby, N.A. and Reid, E. (1997). Determining the composition of the blanket bogs of Scotland using Landsat TM. In From Patch to Planet. (Ed. by A. Millington), Biogeography Research Group Conference, Chester 1995, Wiley.
Quarmby, N.A., Everingham, F, and Reid, E. (1997). The composition of the blanket bogs of Lewis and Harris determined using Landsat TM. SNH Research, Survey and Monitoring Report - in press. SNH, Battleby.
Reid, E., Mortimer, G.N., Thompson, D.B.A. and Lindsay, R.A. (1994). Blanket Bogs in Great Britain: An Assessment of large-scale pattern and distribution using Remote Sensing and GIS. In Large Scale Ecology and Conservation Biology (Ed. by P.G. Edwards, R.M. May and N.R. Webb), pp. 229-246. 35th Symposium of the British Ecological Society, Blackwell Scientific Publications, Oxford.
Reid, E., Ross, S.Y., Thompson, D.B.A., and Lindsay, R.A. (1996). From Sphagnum to a satellite: towards a comprehensive inventory of the blanket mires of Scotland. In: Conserving Peatlands (Ed. L. Parkyn, R.E. Stoneman and H.A.P. Ingram), pp. 204-216. Centre for Agriculture and Bioscience Publications, Wallingford.
Renwick, L.J. and E. (1997). Ecological mapping through a Geographic Information System. SNH Research Survey and Monitoring Report - in preparation. SNH, Battleby.
Rodwell, J.S. (ed) (1991). British Plant Communities. Vol. 2: Mires and Heaths. Cambridge University Press. Cambridge.
Thompson, D.B.A., Hester, A.J. and Usher M.B. (eds) (1995). Heaths and Moorland Cultural Landscapes. HMSO, Edinburgh.
Tudor, G.J., Mackey, E.C. and Underwood, F.M. (1994). The National Countryside Monitoring Scheme: the changing face of Scotland, 1940s to 1970s. Main Report. SNH, Battleby.
Many individuals have been involved in the development of the Scottish Blanket Bog Inventory (formerly the National Peatland Resource Inventory, NPRI). We would like to express special thanks to: Richard Lindsay for initiating the research and launching the project within SNH; Fiona Everingham for analysing, summarising and reporting on the peatland data for the assessment of the peatlands on Lewis; Joanna Drewitt who aided in the production of diagrams for this I&AN; and the past and present members of the NPRI and SBBI teams, who have advised and contributed enormously to establishing the method, field analysis, map production and project outputs. Finally, we thank those individuals who have offered comments and guidance on the development of this project.
Further information and copies of the reports can be obtained from Eliane Reid.