Methods
Islands & DEM
We used a point dataset[1] generously provided by Hedvig Skirgård as the basic dataset of the islands with all associated information such as language, coordinates, area and regional affiliation. To find out the elevation of the individual islands, we downloaded the digital elevation model[2] (DEM) from General Bathymetric Chart of the Oceans (Gebco), which has a high resolution of 15 arc seconds). We intersected the point dataset of the islands with the raster dataset of the DEM and assigned the pixel values of the raster to the points lying within it.
Climate Scenarios
To ensure that the influence of sea level is not only visible in the map as a direct response to sea level rise, we added a temporal component using climate scenarios. To do this, we used the two representative concentration pathways (RCP) 2.6 and 8.5 from the Intergovernmental Panel on Climate Change (IPCC)[3]. The RCP2.6 assumes drastic climate-impacting interventions, while the RCP8.5 assumes no interventions at all, i.e. business as usual. We have limited ourselves to these two extreme scenarios, as we did not want to overwhelm the user with too many choices. Combined with a manual sea level setting, there are three different choices, which is a manageable selection. To calculate the exact rise of sea level, we used several scientific publications[3, 4, 5, 6]. From these we have calculated the sea level rise in the mm range in annual steps up to the year 2300, so that the year can be used as input for the interactive map and the sea level rise is derived directly from it.
Adjustment of Elevation
Since the resolution of the elevation data in the DEM is 1 metre, no changes would be visible on the map for scenario RCP2.6, since sea level does not rise to one metre until the year 2300. For scenario RCP8.5, the first effects of sea level rise on the islands in Remote Oceania would only be visible in about 100 years. Since this is very unrealistic on the one hand and on the other hand the resolution of the DEM of 15 arc seconds (about 500m) does not allow a realistic height estimation for many very small islands, we adjusted the heights. We proceeded in such a way that we normalised the heights by the areas of the islands. For example, we changed all islands with a height of 1m so that the island with the largest area has a height of 1m and an area of 0 corresponds to height 0. This is how we proceeded for all altitude levels. Thus we have obtained an even distribution of islands over all heights in the cm range. This finally results in an enhanced interactive response of the map to the user's input.
Needless to say that this does not correspond to reality. But under the given circumstances it is the most accurate solution we could generate without acquiring highest-resolution DEMs. Statements based on the use of the map must therefore be taken with caution and always considered in the context of the restrictions.
Implementation in Maps
To visualise the impact of sea level rise, both physically and culturally, on the islands in Remote Oceania, we decided to generate two maps.
In the first map, the islands are shown with their associated elevation. If the island is higher than the current sea level, it is shown as white (secure). If the sea level approaches the height of an island, it is shown in red (endangered). If the sea level has reached the height of an island, it is shown in black (submerged). For the sake of clarity, the current status is shown in a legend in the lower right-hand area of the map in such a way that the number of islands per category is displayed.
The second map shows the influence of the sea level on linguistic diversity in Remote Oceania. For a better overview, we have aggregated the languages on different islands into polygons, which are visible as language areas. The area of a language disappears from the map when the sea level has reached the height of the highest island within the respective area. Again, a legend in the lower right area of the map shows the number of languages present at the selected sea level. The view can be changed from language areas to individual islands via the layer options if required. In this case, the number of islands displayed is no longer 5525 as in the first map, but 14501, since certain islands are home to several languages, which had to be split up for the display of the languages. For a more realistic picture, a model could have been used here that not only binarily decides whether an area remains present or not, but adjusts the area of the languages to the number of islands. However, this would have gone beyond the scope of this work. In order to experience this interactively by yourself, we have decided to make the map accessible on this website nevertheless.