http://doi.io-warnemuende.de/10.12754/prog-2022-0001
doi:10.12754/prog-2022-0001
© Author(s) 2022. This work is distributed
under "MIT Copyright 2022 Susanne Feistel, Michael Naumann, Thomas Ruth, Jakob Zabel, Markus Plangg, Rainer Feistel (susanne.feistel@io-warnemuende.de)"

GisAnox - compiling of hypoxic and euxinic maps for the Baltic Sea with GIS output formats (Version 1.0)

Feistel, Susanne; Naumann, Michael; Ruth, Thomas; Zabel, Jakob; Plangg, Markus; Feistel, Rainer

A short historical summary of the method development: The automated mapping of this kind of maps showing the extend of hypoxic and euxinic zones in the Baltic Sea date back to the year 2002. The initial version of the source code was written by Rainer Feistel in Visual Basic using textual standardized data input as an excerpt of IOW’s database (NAUSCH et al. 2002). After further developments of the central database and transferring the source code to a more advanced programming language, using PHP version 5.6, a modernization of the method was implemented in 2015 (NAUSCH et al. 2015). The output format of this stage of the method was the classic vector graphic SVG-file. The long-term data collection of routinely cruises conducted by the institute in the time-span 1969-2015 were compiled and graphics were published together with a method description as an atlas in 2016 (FEISTEL et al. 2016, method description see pages 12-21). Known issues are listed in chapter 6 (pages 21-22). The goal, already mentioned, was to create output formats which can be processed in geo-information-systems (GIS) to allow the calculation of areas affected by oxygen deficiency and makes them available for broader applicants. A “static” graphical map product was transferred to an interoperable format (GIS-shapefile), which allows multiple options for comparative analysis with other spatial parameters (e.g. habitats of species). This next development step, called “GisAnox”, was done in 2016 by the listed team of authors and is published in the attached ZIP-File. The programming language was transferred to Java and a README textfile (readme.md) of script elements and processing steps is included.

Description of the “GisAnox” method: In advance of the interpolation process an automated database request is performed at the IOW DB (database of the institute). This script excerpts from each cruise of a defined list of stations (locations) the water depth levels where laboratory results get hypoxic and euxinic (dissolved oxygen levels below 2 ml/l and hydrogen sulphide was detected) as well as the bottom values of these parameters. The results are written in a textfile, the so-called “level file”, as basic dataset for the interpolation process. Another basic information is the territorial definition of subbasins (basins2017.wkt). This file contains the subbasin borders as coordinate polygons and is read by the interpolation algorithm as separate basins (see also readme.md for details). During the development of the “GisAnox” different kriging approaches were tested and compared to interpolations of kriging with semi-variogramme analysis and limited search radius done with a professional interpolation software (SURFER - Golden Software). Considering data split into oceanographic units on basin scale led to sufficient results compared to complex kriging functions. Even if you limit the search radius in SURFER, data of different basins is compared to each other. The natural sill sections between the basins and lowered halocline depth control and separate spreading of hypoxic areas. The interpolation can be found in the file: “VolumeInterpolation.java”. Every subbasin is interpolated separately and finally merged for visualization purposes in the GIS-shapefile. The interpolation algorithm chosen is ‘ordinary kriging’ which uses a ‘simple variogram’ that linearly increases the variance with distance. The maximum distance used by the algorithm for data points considered is 2.5 degrees in geographic coordinate system. This distance was chosen to prevent points to be used in interpolation that are too far away to have any relation to the point to be estimated and is roughly oriented at the size of the Baltic Sea’s smaller subbasins. That way, in smaller subbasins such as the Arkona Basin or Bornholm Basin all points are considered but in larger such as the Eastern Gotland Basin for example, points from the north will hardly impact the estimation of points in the south, due to their large distance to each other. The mean is estimated by the average depth of the respective layer in that basin. The assumption needed here for a stationary mean in a basin is likely given by the density stratification of seawater (within a basin). The measured data contains no discernable systematic trend, so ordinary kriging was chosen. Points where no lack of oxygen/H2S was measured were set to -999 m for visualization purposes. Covariance matrices use a simple (linear from min to max over 2.5 degrees) variogram with sill and biased sample variance. These parameters were estimated using the subbasin size (see above) and making no assumptions about the impact of the first derivation on distance for the kriging interpolation. The kriging interpolation is requested for every point in a 1 nautical mile grid and visualized up until the basin borders given by the basins2017.wkt-file. That means the visualization and estimation ends at these basin borders and underwater at the bathymetry given by “iowtopo2_rev03.nc” (SEIFERT et al. 2001) in the resources using a flood fill algorithm. To generate the shapefiles for each surface (Hypoxia and H2S) for the whole territory, all interpolations of subbasins are merged. The final output are four files: Two shapefiles of polygons (hypoxic and euxinic area) and two raster grid shapefiles with 1 nautical mile grid point distance (hypoxic and euxinic area). The validation of these maps with recalculation of maps with a reduced dataset of only 1/3 of the stations show appropriate results. Differences are in mean within 4 %. In general, the limited data leads to overestimation of +3.8 % in mean (+2230 km2) for hypoxic area and +3.7 % (+423 km2) for euxinic area calculations.

Literature: FEISTEL, S.; FEISTEL, R.; NEHRING, D.; MATTHÄUS, W.; NAUSCH, G.; NAUMANN, M. (2016): Hypoxic and anoxic regions in the Baltic Sea, 1969 – 2015. - Marine Science Reports, 100: 76 pages, doi: 10.12754/msr-2016-0100. https://www.io-warnemuende.de/tl_files/forschung/meereswissenschaftliche-berichte/mebe100_2016-hypoxic-and-anoxic-regions.pdf NAUSCH, G., FEISTEL, R., LASS, H. U., NAGEL, K., SIEGEL, H., 2002: Hydrographisch-chemische Zustandseinschätzung der Ostsee 2001. – Meereswiss. Ber., Warnemünde, 49, 2-77, doi: 10.12754/msr-2002-0049. NAUSCH, G., NAUMANN, M., UMLAUF, L., MOHRHOLZ, V., SIEGEL, H., 2015: Hydrographic- hydrochemical assessment of the Baltic Sea 2014. – Meereswiss. Ber., Warnemünde, 96, 1-93, doi: 10.12754/msr-2015-0096. SEIFERT, T., TAUBER, F. & KAYSER, B., http://www.io-warnemuende.de/topografie-der-ostsee.html, (2008) (Date of access: 08/03/2016)

Download

Full Article or Dataset

Related