Global compilation of published gas hydrate-related bottom simulating reflections

The geographic accuracy of the digital bottom simulating reflection (BSR) locations in this dataset depends on (a) the accuracy of the BSR delineations in the original references; (b) the quality of the georeferencing and digitization that could be achieved for the files (all except BOEM2012 and Shedd2012) that required these processing steps; (c) decisions made about how to digitize the BSRs for this dataset. In addition, the dataset as a whole has an inaccuracy in that some known BSRs were not included owing to the difficulty of finding a map with sufficient features for georeferencing and digitizing.
To assess the potential inaccuracy related to (a): The general criteria used to delineate BSRs in marine seismic data acquired with high-energy, impulsive acoustic sources having the frequency characteristics needed to image BSRs include such factors as: (1) negative polarity (reverse impedance, indicating higher over lower velocity sediments) reflections in marine sedimentary section within tens to hundreds of meters below the seafloor; (2) seismic reflections that sometimes roughly parallel the seafloor, consistent with their position at a depth where the hydrostatic pressure and temperature in the sediments correspond to the limit for gas hydrate stability conditions; and (3) seismic reflections that cut across stratigraphic layering, consistent with the reflections' marking a phase boundary, not a stratigraphic boundary, within the sediments. Other bottom simulating features are sometimes recognized in marine seismic data (e.g., the transition of quartz to opal-CT), but such features have different impedance and depth characteristics than BSRs associated with gas hydrates. For the published papers used as the source of maps for georeferencing and digitization of the BSRs, all had been subject to peer review, and gross misidentification of opal-CT features as hydrate-related BSRs is unlikely. There is considerable subjectivity in authors' delineations of BSRs on maps, especially when polygons are drawn encompassing areas in which the raw data indicate fragments of recognized BSRs along individual seismic lines. How much of the area does not contain BSRs within the polygons on original maps is unknown. For Mosher2011_EasternCanada, Bialis2022_BlackSea, and Minshull2020_Norway the BSRs in the original publications were respectively indicated only by circles, ovals, or small, nearly circular shapes on the original maps, meaning that more precise locations and/or spatial extents of BSRs are unknown. Polygons around areas containing BSRs in the original publications often have square edges (possibly correlating to the extent of the original seismic data) or are truncated at the edge of a map, and the original maps associated with BenAvraham2022_SouthAfrica and Neben1998_CelebesSea have question marks on parts of the BSR polygons. Some publications map BSRs not as polygons, but as line segments or points along seismic lines at the places where BSRs are recognized. The publications associated with Jin2003_SouthShetland and Kopp2002_Sumatra show squares over BSRs mapped on seismic lines, and the original maps for Trehu2022_Cascadia, Chi2006_Taiwan, Liu2006_Taiwan, Ohde2018_Nankai, and some others have BSR points along seismic lines. Especially for cases in which the published maps use a polygon to enclose an area where BSRs have been recognized, the delineation of the polygon is likely to represent the greatest inaccuracy in terms of where BSRs actually exist. Note that there are two or more maps for some locations in the dataset. Examples (not exhaustive list) include the Black Sea (Bialis2022_BlackSea, Ludmann2004_BlackSea, and Vasilev2002_BlackSea), the South China Sea (Li2013_SouthChinaSea and Wang2022_SouthChinaSea), Nankai (MH212009_Japan files and Ohde2018_Nankai), and offshore India (Kumar2014_India files and Prakash2010_IndiaEast) have two or more that features within the BSR_polygons layer within the GeoPackage. Users can compare the extent of the BSRs in these maps (in some cases, the BSR polygons will be seen to overlap) and note the differences, but no value judgment was placed by the dataset's authors on the relative accuracy of one of these maps over another.
To assess the potential inaccuracy associated with (b): Georeferencing and digitizing BSRs from maps published over a period of more than 30 years in a wide range of journals and from many geographic areas are not the preferred method for compiling a global dataset, but rather the only method possible when the digital files are unavailable from the original authors. High-quality georeferencing was difficult for BSRs at high latitudes, for maps that were originally published in projections that distorted geometric relationships, and for maps lacking coastlines. The quality of digitization can vary from map to map as well, depending on slight variations in manually picking the points to digitize. It is not possible to quantitatively determine the degree of inaccuracy resulting from georeferencing and digitization, but georeferencing inaccuracies will almost always exceed digitizing inaccuracies.
To assess the potential inaccuracy related to (c): Subjective decisions were made during the digitization process, especially when representing BSRs that appeared in the original published maps as points along seismic lines. For most published maps, polygons in the original were digitized as polygons, and polygons were not combined during digitization. For BSRs identified as points along seismic lines in references such as Chi2006_Taiwan, Liu2006_Taiwan, and Ohde2018_Nankai, polygons were drawn around areas where the BSR points were identified, and the polygon outlines were digitized. Trehu2022_Cascadia and Trehu2022_Hydrate Ridge come from the same published map, which has BSR points identified along seismic lines. For the former file, the points were combined into line segments where BSRs were present on the seismic lines and digitized as polylines. For Hydrate Ridge, which was densely covered by BSR points identified along seismic lines, a polygon was drawn encompassing the dense cluster of points and then digitized. When the dataset authors drew polygons for digitization around points or line segments where BSRs were recognized in original seismic data, this was done judiciously to ensure that large areas without BSR indicators would not be included within the polygons. However, drawing such polygons does introduce an additional inaccuracy.
Note that the two entries associated with BOEM_USAtlantic were provided to the U.S. Geological Survey by the Bureau of Ocean Energy Management (BOEM) as digital coordinates, meaning that there is no inaccuracy associated with (b) or (c) for these BSRs. The entries associated with Shedd2012 were similarly provided by BOEM as shapefiles so that there is also no inaccuracy related to (b) or (c).
BSRs are known to exist in some locations where no published map could be found or where published maps lacked features making it possible to georeference and digitize the maps. For example, BSRs exist offshore parts of Antarctica, but published maps could not be georeferenced for this dataset. BSRs are known to be far more extensive offshore eastern Canada than indicated by the one publication included in this dataset, but no maps have been published, perhaps because the underlying data described in the reference for Mosher2011_EasternCanada are proprietary.