Mapping large, shallow inlets and bays: modelling a Natura 2000 habitat with digital terrain and wave-exposure models
1 Norwegian Institute for Water Research, Gaustadalléen 21, N-0349 Oslo, Norway
2 AquaBiota Water Research, Svante Arrhenius väg 21 A, SE-10405 Stockholm, Sweden
Correspondence to T. Bekkby: tel: +47 22185100; fax: +47 22185200; e-mail: trine.bekkby{at}niva.no
Bekkby, T., and Isæus, M. 2008. Mapping large, shallow inlets and bays: modelling a Natura 2000 habitat with digital terrain and wave-exposure models. – ICES Journal of Marine Science, 65: 238–241.EU member countries are obliged to protect a certain share of Natura 2000 habitats. Hence, these habitats must be mapped. This paper is an attempt to provide a tool for modelling one of the Natura 2000 habitat, the "large shallow inlets and bays" (Natura 2000 habitat 1160), using a Norwegian archipelagic area as a case study. The Natura 2000 definition of the habitat is interpreted into criteria used for modelling, and a spatial prediction is presented on a map. The effect of scale, regarding both spatial resolution of data and methodology, is also tested. This is the first publicly accessible attempt to model the Natura 2000 habitat. It shows that the result of the modelling depends on the spatial resolution of the data and the methods used in the modelling process. Using data at a 10-m and a 25-m resolution provides good results, and even the model based on the 50-m data provided an acceptable overall picture.
Keywords: bays, GIS, habitat directive, habitats, inlets, modelling, Natura 2000
Received 2 April 2007; accepted 10 December 2007.
 |
Background and introduction |
|---|
 Top Background and introduction Our interpretation of the...Our approach: GIS modellingResults and discussionReferences |
|---|
EU member countries are obliged to protect a certain share of Natura 2000 habitats. Hence, these habitats must be mapped. Even though Norway is not a member of the EU and does not have the same obligations, we believe that working within the same framework and using the same definitions as our neighbouring countries is critical to the management of common resources and interests.
Mapping a habitat can be time-consuming and costly. Also, in terms of inlets and bays, it is difficult, because choices, definitions, and interpretations are subjective. Hence, we have attempted to provide a tool for GIS modelling of habitat through the use of identifiable and re-examinable geophysical criteria. We also evaluate the effect of using data at different scales in respect of both spatial resolution of data and methodology. The work is based on programmes carried out in Norway and Sweden (Bekkby et al., 2002, 2004; Axelsson, 2004; Bekkby, 2006; Bekkby and Rosenberg, 2006; Rinde et al., 2006).
According to the Natura 2000 Interpretation Manual (EUR25, http://ec.europa.eu/environment/nature/), "large, shallow inlets and bays" are defined as large and shallow indentations generally sheltered from wave action. The term "shallow" may vary with geographical location, and it has been considered inappropriate to fix a maximum water depth. However, the depth limit is often defined by the distribution of the Zostera and Potametea associations. In contrast to estuaries, the influence of fresh water on inlets and bays is generally limited.
|
Our interpretation of the Natura 2000 definition |
|---|
|
TopBackground and introductionOur interpretation of the...Our approach: GIS modellingResults and discussionReferences |
|---|
Using the geophysical criteria from the Natura 2000 definition (above), we used indentation, wave exposure, and depth as criteria for modelling the habitats:
- Indentations: defined as semi-enclosed areas.
- Wave exposure: the habitat is generally sheltered from wave action. We used a wave-exposure model (Isæus, 2004) and the distribution of Zostera marina (provided in a sister paper in preparation) to define areas sheltered from wave action.
- Depth: shallow areas were defined by the distribution of Z. marina
|
Our approach: GIS modelling |
|---|
|
TopBackground and introductionOur interpretation of the...Our approach: GIS modellingResults and discussionReferences |
|---|
Preparing the data
The basic dataset for this modelling exercise may be a grid or a point dataset. In our case, it was a digital-terrain model DTM at a 10-m spatial resolution. The models are developed in ArcView 3.3. We transformed the grid into data points (a xyz file) using the ArcView 3.3 extension "GridPig-Tools" (USGS, 2003). To provide each point with an "id" number needed later, an avenue script, "addrecno", was developed by V. Bakkestuen, Norwegian Institute for Nature Research. Some of these processes may take time, in particular if the spatial resolution is high and the datasets are large. Therefore, removing data >400 m from land, where inlets and bays will not occur by definition, so reducing the study and modelling area, and removing land areas may save time and effort.
Finding the indentations
In modelling the indentations, we used the definitions presented by Axelsson (2004), i.e. finding the data points from the grid that were <400 m from land in at least five of eight compass directions. The extension "radiation lines" (Jenness, 2006) was used, giving each data point 400-m long radiation lines in the eight compass directions. A land polygon was used to erase all radiation-line fragments crossing the coastline on to land. All lines not cut by the land polygon (i.e. those >400 m from land) were deleted (Figure 1). The number of compass directions <400 m from land was estimated for each data point using the "Summarize" function in ArcView. The point theme was converted to a grid at the same resolution as the point dataset. An indentation was defined as a grid cell with a value of 5 or more, i.e. land was found within 400 m in at least five of the eight compass directions.
|
Finding the areas sheltered from wave action
The target habitat is generally sheltered from wave action. Wave exposure (average over 5 years, at a spatial resolution of 10 m) was modelled using the simplified wave model (SWM) developed by Isæus (2004). The model has been refined to fit the EUNIS habitat classification system (developed by EEA; see Davies and Moss, 1999). As Z. marina is regarded as a one of the key species in the habitat, according to the Natura 2000 Interpretation Manual, we used the exposure limit of Z. marina (paper in preparation) as our definition of sheltered areas. The data show that 90% of the Z. marina sites are in areas with a SWM value of <59 521, i.e. ranging from the EUNIS class "ultra-sheltered" to the middle of class "moderately exposed". As Rinde et al. (2006) used a SWM value of 100 000 as a limit, i.e. including the whole EUNIS class of "moderately exposed", we also decided to include the whole "moderately exposed" class and to use this limit.
Finding the shallow areas
We believe that it is inappropriate to fix a maximum water-depth limit. Norway has a steep and rocky shoreline, different from many other European countries, and a fixed depth limit would not easily fit such a coast. Hence, as suggested by the Interpretation Manual, we defined the depth limit based on the distribution of Z. marina. Our own unpublished data show that 90% of the Z. marina sites are <7.4 m deep, so we used 7 m as a depth limit, as also used by Rinde et al. (2006).
The effect of scale
As the result of the modelling may depend on the spatial scale of the data, we modelled large, shallow inlets and bays using models with a spatial resolution of 10, 25, 50, and 100 m. We also tested the result of using 8 and 16 compass directions when finding the indentations. A grid cell was part of an indentation if the cell value was 5 when doing the calculations in eight compass directions and 10 when calculating for 16 compass directions.
|
Results and discussion |
|---|
|
TopBackground and introductionOur interpretation of the...Our approach: GIS modellingResults and discussionReferences |
|---|
Figure 2a shows the result of modelling using input data with a spatial resolution of 10 m and eight compass directions. Figures 2b–d show the results using eight compass directions and input data with a spatial resolution of 25, 50, and 100 m, respectively. Increasing the resolution from 10 to 25 to 50 m (Figures 2a–c), the coverage of the model does not change much. This is also the case when measuring the area covered by the model (Figure 3). However, increasing the resolution from 50 to 100 m resolution (Figures 2b and 2c), the small inlets and bays are not covered, and the area decreases (Figure 3). Using 16 compass directions (Figures 2e–h), the trend is the same. However, the decrease in area appears at a spatial resolution of 50 m (Figure 3), not at 100 m as for the model with eight compass directions.
|
|
The reason that the small inlets and bays are not covered, resulting in a decrease in measured area as spatial resolution increases, is that the small islands are generally completely covered by a single grid cell. In such cases, the distance from the data pixel to the islands will not be measured, and no indentations will be found.
As far as we know, this paper provides the first published attempt to model a Habitat Directive Annex 1 habitat (i.e. the Natura 2000 habitat 1160) and is a good example of how Natura 2000 habitats may be modelled and mapped. It shows that the result of the modelling depends on the spatial resolution of the data and the methods used in the modelling process. Using data at a resolution of 10 or 25 m provides good results, and even a model based on 50 m data provides an adequate picture.
Greenlaw et al. (2007) developed a classification system for coastal, marine bays based on GIS analyses of geophysical factors. The factors were also used to predict fine substratum and biological communities within bays. Such a classification system may overlap to some degree with the Natura 2000 system, and hence our approach. Both approaches include identifiable and re-examinable geophysical criteria, though in-depth comparison of methods is needed before any conclusion on the extent of overlap and possible coordination of methodology is made.
|
Acknowledgements |
|---|
Funding for this project was provided by the Research Council of Norway and the Norwegian Institute for Water Research. We thank S. Axelsson (Metria, Sweden) for providing information on the Swedish approach, V. Bakkestuen (Norwegian Institute for Nature Research, and the University of Oslo) for developing the GIS scripting of the algorithms, and J. Ekebom (Metsähallitus Natural Heritage Services, Finland) for valuable comments.
|
References |
|---|
|
TopBackground and introductionOur interpretation of the...Our approach: GIS modellingResults and discussionReferences |
|---|
-
Axelsson S. Kartering av vissa kustbiotoper som utpekas i EU:s Habitatdirektiv. (2004) Metria Report M2003/03473.9.
Bekkby T. GIS-modellering av viker. Vann (2006) 2:127–131.
Bekkby T., Erikstad L., Bakkestuen V., Bjørge A. A landscape ecological approach to coastal zone applications. Sarsia (2002) 87:396–408.[CrossRef]
Bekkby T., Rinde E., Rosenberg R., Bakkestuen V., Erikstad L. The effect of terrain structures and environmental factors on the distribution marine habitats. ICES Document CM 2004/P: 13 (2004).
Bekkby T., Rosenberg R. Marine habitaters utbredelse—terrengmodellering i Gullmarsfjorden. Länsstyrelsen i Västra Götaland län, Vattenvårdsenheten. Report 2006:07 (2006) 33. (in Norwegian and Swedish, with English abstract).
Davies C. E., Moss D. EUNIS Habitat Classification. Final Report to the European Topic Centre on Nature Conservation, European Environment Agency. (1999) 256.
Greenlaw M., O’Connor S., Roff J. Conservation of Nova Scotia’s bays: are we just coasting? Defining coastal zone representative bay types. In: DFO/FSRS Inshore Ecosystem Project Data Synthesis Workshop, 19–20 March 2007 (2007) 5–6. DFO Canadian Science Advisory Secretariat Proceedings Series 2007/028. http://www.dfo-mpo.gc.ca/csas/Csas/Proceedings/2007/PRO2007_028_E.pdf.
Isæus M. Factors strukturing Fucus communities at open and complex coastlines in the Baltic Sea. Doctoral thesis at the naturvetenskapeliga fakulteten, Botaniske institutionen. (2004) Sweden: University of Stockholm. 165.
Jenness J. Radiating lines and points. (2006) Version 1.1, downloaded from http://www.jennessent.com/arcview/arcview_extensions.htm (freely available), Jenness Enterprise.
Rinde E., Rygg B., Bekkby T., Isæus M., Erikstad L., Sloreid S-E., Longva O. Documentation of marine nature type models included in Directorate of Nature Management’s database Naturbase. First generation models for the municipalities mapping of marine biodiversity 2007. (2006) NIVA Report LNR 5321-2006 (in Norwegian with English abstract).
USGS. "Grid Pig". (2003) Version 2.6. ArcView extension, downloaded from http://arcscripts.esri.com.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  T. Bekkby, E. Rinde, L. Erikstad, V. Bakkestuen, O. Longva, O. Christensen, M. Isaeus, and P. E. Isachsen Spatial probability modelling of eelgrass (Zostera marina) distribution on the west coast of Norway ICES J. Mar. Sci., October 1, 2008; 65(7): 1093 - 1101. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||