Download PDF - COINAtlantic

Methods11 th International Symposium for GIS and Computer Cartography for Coastal Zones ManagementWe develop barriers for spatial neighbourhoods to account for indentations and protrusions in coastlines commonin marine environments. By means of a Graphic User Interface (GUI), the user will be able to select observationpoints as well as constraints files, then select spatial method definition. The new approach works as follows: first,each marine mammal observation is connected to every other observation regardless of distance or barrier. Second,lines are intersected with barriers in order to locate points that due to their geographic location cannot be neighbours;lines that do not cross any barrier are given a value, for this case in a binary format (one), subsequently, linesmaintain the unique identifier of each observation point. Fourth, an additional binary format value (one) is given tothe line that is within a threshold distance given by the user. Fifth, a spatial weighting matrix using k-neighbours,distance or inverse distance is generated. In general, the points that are connected by lines are neighbours whilepoints that are not connected by lines cannot be considered as neighbours because they are separated by land.ResultsIn this study we have given the name of links to the lines that connect observation points. The total number oflinks between a pair of observation points for a non-constrained approach is 26 whereas the total number of links fora constrained approach is 17. Comparing the approaches for delineating spatial neighbourhoods by distance, a meanof 3.71 links per observation point was found for the non-constrained neighbours, compared to constrained neighbourswith a mean of 2.43 links. A standard deviation of 1.11 corresponds to links with non-constrained neighboursand 0.53 for constrained, making the low value of the constrained approach closer to the mean value. Figure 1cshows the results after applying the methodology for identification of neighbourhoods in a marine environmentwhere land barriers are identified as constraints.(a) (b) (c)Figure 1. Spatial neighbourhood definition for: (a) non-constraint fix distance; (b) distance fix with constraint; (c) observationswith constraint neighbors. White background represents water and colour represents land.DiscussionImplementation of constraint approaches using Voronoi or Delaunay triangulation in spatial neighbourhoods definitionsby adjacency of unconnected points have been previously implemented (Gold, 1992; Nordvik and Harding,2008). With these approaches the spatial relationship among neighbours can be altered by adding, changing or removinga neighbour (i.e, Gold and Condal, 1995; Gold, 1994a, 1994b) based on ancillary data on the marine environment,in this case the coastline.Using a barrier approach to delineate spatial neighbourhoods in marine environments improves different spatialanalysis applications as it takes into consideration true geographical limitations Figure 1c presents the correct definitionof neighbourhood for observations of marine species separated by land. On the contrary, traditional approacheswould neglect land and calculate the neighbourhood as presented in Figure 1b.Neighbourhood definitions for spatial studies of marine environments can benefit from methods that integrate additionalphysical data. In this example we show how coastline covariate data may be used to improve delineation ofneighbourhoods. Islands, atolls, and keys among others, can be considered as barriers for species and also, dependingon their morphology, they can influence the direction of nutrients dispersion or currents. For instance, an islandmay affect current direction, leading to phytoplankton upwelling occurring in a different location due to topographyor geostrophy (Yang, 2007).219