Wintersteller, Paul; Strack, Anne (2019): Derivates of a GIS based analysis of the recorded bathymetry and backscatter around Tristan da Cunha [dataset]. PANGAEA, https://doi.org/10.1594/PANGAEA.906110,

---

In: Geissler, Wolfram H; Wintersteller, Paul; Strack, Anne; Kammann, Janina; Jegen, Marion; Maia, Marcia; Schloemer, Antje; Jokat, Wilfried (2019): GIS based analysis, processed sub bottom profiler, and composite grid and backscatter mosaics around Tristan da Cunha [dataset publication series]. PANGAEA, https://doi.org/10.1594/PANGAEA.906154

Content

The attached raster-files are saved in the formats PDF and georeferenced PNG. The shapefile "VolcanicCones_Tristan", including its auxiliary files, as well as all other georeferenced files are shown in the projected coordinate system UTM28S with geodetic datum WGS84 (EPSG # 32728).

The two Excel files contain a, the volume calculation of the volcanic cones and b, the decision table for the BPI classification.

1. Rugosity/Ruggedness raster & Benthic Terrain Modeler:

To investigate the structural complexity of the study area, a rugosity (surface roughness) raster was created using the arc-chord ratio (ACR) index after Du Preez (2015). The ACR method is defined as the contoured area of the surface divided by the area of the surface orthogonally projected onto a plane of best fit. In this way, it effectively decouples rugosity from the slope. To reduce artefacts in the rugosity raster caused by the underlying bathymetry data, the ACR rugosity raster was calculated from the three-fold resampled bathymetry data.

The benthic terrain of Tristan da Cunha was analyzed using ESRIs ArcGIS™ with the aid of the ArcGIS application's Benthic Terrain Modeler (BTM) v. 3.0 (Lundblad et al., 2006; Wright et al., 2012) which classifies bathymetry data and analyses seafloor characteristics.

The most important derivative for this analysis is the Bathymetric Position Index (BPI). The BPI is modified from the Topographic Position Index, which is used in terrestrial environments (Weiss, 2001). It compares the elevation of each cell in the bathymetry raster with the mean elevation of a defined neighborhood around that cell. In this analysis, an annulus neighborhood with an inner and outer radius is used. To identify both fine and broad features on the seafloor, two BPI grids were created using an inner radius of 15 km and an outer radius of 22.5 km for the broad-scale BPI raster and 1.5 km and 3 km, respectively, for the fine-scale BPI raster. Since the bathymetric position tends to be auto-correlated (Erdey-Heydorn, 2008), the BPI grids were standardized. The overall structural analysis was based on a classification dictionary (Supplementary Material Table S2) which defines several geomorphological structures by their broad and fine scale BPI and their slope. The transition from flat areas to broad slopes was set to 3° and from broad to steep slopes to 25°.

2. Slope & Aspect

Slope and aspect are first order derivates of the bathymetry. They are both shown in colored categories.

3. Backscatter

The beam time series are shown as a mosaic in stretched greyscale from black & white and in color white to brown. For analysis the dataset was separated in categories and shown in colors from dark green to light brown.

4. Volume calculations of volcanic cones:

We utilized the GIS software GlobalMapper v18.2 for volume calculations of the volcanic cones, though the prework to generate the volcanic cone polygons was conducted with ESRIs ArcGIS™. The result of the BTM classification was used to extract the four classes that represent outcrops or local ridges (see Supplementary Material Table S2: Class 6, 11, 13 and 14). This outcrop raster was then converted into polygon features. Areas smaller than 0.1 km² were excluded from further analyses. Since it is not possible to select polygons by their shape, a manual selection of the polygon features was necessary to remove all polygons that do not represent circular cone-shaped morphologies (e.g., elongated ridges or parts of the main islands). Furthermore, all polygons were buffered as the outcrop classes from the BTM classification result do not extend to the approximate base of the outcrops, which is due to the functionality of the BPI. The best buffer distance was measured exemplarily (here we buffered with 200 m). Finally, the polygons were smoothed by a Bezier Interpolation and exported as a shapefile. The actual volume calculation was conducted with GlobalMapper's analysis & measurement tool "Pile Volume" and exported as a .CSV file (Supplementary Material Table S1).

Mean backscatter values were calculated for each volcanic cone by using the ArcGIS tool "Zonal Statistics". Very low (<-68 dB) and very high (>-8 dB) backscatter values were excluded beforehand using the "Extract by Attribute tool" as they mainly occur near the outer beams and the nadir and, therefore, most likely represent erroneous dB values.