SUITABILITY OF QUICKBIRD SATELLITE IMAGERY
FOR THECLASSIFICATION OF THE SEMI-NATURAL VEGETATION
IN THE FORVIE NATIONAL NATURE RESERVE (SCOTLAND)


Isle of Skye DANIELA GURLIN


MSc in GIS 2005/2006

Institute of Geography
The University of Edinburgh
Drummond Street
EH8 9XP


Introduction

Natural and semi-natural coastal environments perceive an increased pressure by the intensification of agricultural activities, fisheries, development activities and recreational activities. The political framework for the conservation of these environments is given by the EU NATURA 2000. The NATURA 2000 includes the 'Council Directive 92/43/EEC on the Conservation of Natural Habitats and Wild Flora and Fauna' (Habitats Directive) for the management of Special Areas of Conservation (SAC)and the 'Council Directive 79/409/EEC on the Conservation of Wild Birds' (Birds Directive) for the management of Special Protection Areas (SPA). The prerequisite for the development of efficient conservation strategies is the comprehension of the natural dynamics of these environments.

Objectives

The study will explore the potential of QuickBird satellite imagery for the creation of vegetation maps and the analysis of changes in vegetation cover in an object-oriented classification approach in eCognitionTM for the Forvie National Nature Reserve in Aberdeenshire, Scotland. The analysis of changes in vegetation cover will include the comparison of aerial photography from July 1994 and a QuickBird satellite image from the 27th of September 2005.


QuickBird satellite image of the Forvie National Nature Reserve
The classification of the semi-natural vegetation
in the Forvie National Nature Reserve
was performed for 2 study areas of the Sands of Forvie

Study Area

The Forvie National Nature Reserve includes the estuary of the river Ythan (242.24 ha), a wetland of international importance (RAMSAR convention) and SPA (NATURA 2000) and the Sands of Forvie (734.05 ha), a complex coastal dune system and SAC and SPA (NATURA 2000).

Habitats of international importance in the Forvie National Nature Reserve include
• Shifting Dunes
• Shifting Dunes with Ammophila arenaria (marram grass)
• Decalcified Dunes with Empetrum nigrum (crowberry) and
• Humid Dune Slacks.

The classification of the semi-natural vegetation of the Sands of Forvie followed the national vegetation classification (NVC) standard. The main NVC vegetation types of the Sands of Forvie include mesotrophic grasslands (MG), swamps (S), mires (M), heaths (H), shingle, strandline and sand-dune communities (SD) and salt-marsh communities (SM).



Shifting dunes with Ammophila arenaria
Sunset in the Forvie National Nature Reserve

Typical plant communities of the Sands of Forvie include Erica tetralix - Sphagnum compactum wet heath (M16), Calluna vulgaris - Carex arenaria heath (H11), Carex arenaria - Festuca ovina - Agrostis capellaris dune grassland (SD12), Carex arenaria - Cornicularia aculeata dune communities (SD11), Salix repens - Holcus lanatus dune slack communities (SD16), Potentilla anserine - Carex nigra dune slack communities (SD17) and Ammophila arenaria mobile dune communities (SD6).


Calluna vulgaris - Carex arenaria heath (H11)
Heathlands are typical for the northern part of the Sands of Forvie
Ammophila arenaria mobile dune community (SD6)
Shifting Dunes are typical for the southern part of the Sands of Forvie

Material and Methods

(1)  QuickBird satellite image of the Ythan estuary and the Sands of Forvie from September 2005
(2)  Colour aerial photographs of the Ythan estuary and the Sands of Forvie from July 1994 in a scale of 1:5000
(3)  Results of the Sand Dune Vegetation Survey of Scotland (SDVSS) for the Sands of Forvie from 2001


Methodology
Methodolgy


Results and Discussion

The result of the classification of the semi-natural vegetation of the Sands of Forvie for study area 2 indicates a typical mosaic of shifting dunes with Ammophila arenaria, Carex arenaria - Festuca ovina - Agrostis capellaris dune grassland and Calluna vulgaris - Carex arenaria heath. The analysis of changes in vegetation cover revealed a vegetation increase of 3.77 ha (increase of 4.00 ha and decrease of 0.23 ha). The accuracy of the classification is 91.40 % for the aerial photography and 83.40 % for the QuickBird satellite image. The visual comparison of the classification results indicates changes in several plant communities.

Several problems complicated the classification of the semi-natural vegetation of the Sands of Forvie. These problems are a consequence of spectral similarities of several SDVSS vegetation types and spectral variations of the vegetation within individual vegetation types. Typical examples include Calluna vulgaris - Carex arenaria heath (with the only sub-community Empetrum nigrum ssp. nigrum) and Carex arenaria - Cornicularia aculeata dune communities. The spectral variation within a vegetation type is mainly related to different succession states of the vegetation, different species abundances and different percentages of sand and vegetation cover.



Study area 2
Result of the analysis of changes in vegetation cover


Study area 2
Result of the classification of the aerial photography (1994)
Study area 2
Result of the classification of the QuickBird satellite image (2005)

Calluna vulgaris - Carex arenaria heath (H11)
Calluna vulgaris - Carex arenaria heath in the Sands of Forvie
Calluna vulgaris - Carex arenaria heath (H11)
Calluna vulgaris and Empetrum nigrum are typical species
of the Calluna vulgaris - Carex arenaria heath

The result of the classification of the semi-natural vegetation of the Sands of Forvie for study area 1 indicates a typical succession from shifting dunes with Ammophila arenaria to Carex arenaria - Festuca ovina - Agrostis capellaris dune grassland. The analysis of changes in vegetation cover revealed a vegetation increase of 10.07 ha (increase of 13.31 ha and decrease of 3.24 ha). The changes appeared mainly in areas of Ammophila arenaria mobile dune communities and indicate a very dynamic dune system in the south of the Sands of Forvie. The accuracy of the classification is 93.25 % for the aerial photography and 81.95 % for the QuickBird satellite image. The visual comparison of the classification results indicates changes in several plant communities.

The result of the analysis of changes in vegetation cover indicates a reasonable suitability of QuickBird satellite imagery for the analysis of changes in the relative proportion of sand and vegetation. The application of a resolution merge of the multi-spectral and panchromatic wavebands allows the identification of relatively small patches of Ammophila arenaria on the shifting dunes in the south of the Sands of Forvie. Problems are only caused by the separation of sand and shadows. The suitability of QuickBird satellite imagery for the analysis of changes within the vegetation is dependent on the spectral separability of the vegetation types.

Study area 1
Result of the analysis of changes in vegetation cover


Study area 1
Result of the classification of the aerial photography (1994)
Study area 1
Result of the classification of the QuickBird satellite image (2005)

Ammophila arenaria mobile dune community (SD6)
Ammophila arenaria mobile dune community
with Cakile maritima and Ammophila arenaria
in the Forvean Links
Cakile maritima
Cakile maritima is a typical species of several
Ammophila arenaria mobile dune communities

Conclusions

The result of the classification of the semi-natural vegetation of the Sands of Forvie indicates a reasonable suitability of QuickBird satellite imagery for the analysis of changes in vegetation cover. The suitability of QuickBird satellite imagery for the analysis of changes within the vegetation is dependent on the spectral separability of the vegetation types.

Acknowledgements

I would like to thank T. Malthus and C. Place, University of Edinburgh, and G. Grenzdörffer, University of Rostock, for their technical support, C. Stuart and S. Angus, SNH, for the provision of the aerial photographs and the results of the SDVSS and A. Drysdale, SNH, for the research permit for the Forvie National Nature Reserve.