This work was carried as part of Marine Geoscience for Marine Spatial Planning Program

implemented by Natural Resources Canada. We would like to thank Carrie Brayall from

the Canadian Hydrographic Service for her invaluable assistance in helping assemble some

of the required multibeam datasets

Data shall NOT be used for navigation. The authors and Natural Resources Canada make

no warranty, expressed or implied, regarding accuracy, reliability, or completeness of

furnished data, without even the implied warranty of MERCHANTABILITY or FITNESS

FOR a PARTICULAR PURPOSE, nor does the fact of distribution constitute such a

warranty. Users assume responsibility to determine the usability of these data. The user is

responsible for the results of any application of this data for other than its intended purpose. For the complete disclaimer on data use restrictions see:

Patton, E. and Kostylev, V.E. 2023. Bathymetric compilation for Scotian Shelf and Newfoundland-Labrador

Shelves bioregions, offshore Atlantic Canada; Geological Survey of Canada, Open File 9064, 29 p.

https://doi.org/10.4095/332197 / https://ostrnrcan-dostrncan.canada.ca/handle/1845/267331

Title: Setup of GRASS GIS 7.8.5 Project -- Description: To compile the various base layer rasters, they were first reprojected to a custom Lambert

Conformal Conic (LCC) projection using the Geospatial Data Abstraction Library (GDAL)

(Open Source Geospatial Foundation, 2022) tool gdalwarpa. Bilinear interpolation was

employed during this reprojection process, and the resulting rasters were saved in TIFF

format.

Subsequently, the projected TIFF files were imported into GRASS GIS 7.8.5 (GRASS)

(Open Source Geospatial Foundation, 2022) using the r.in.gdalb program, preserving their

native resolutions. The geographic extent for the imported rasters was defined by the

following corner coordinates:

Northwest Corner: -72.253067, 60.653003

Northeast Corner: -37.352057, 59.414870

Southeast Corner: -45.357224, 38.880454

Southwest Corner: -67.819771, 39.683465

Upon importing all the datasets, they were uniformly resampled to the compilation's base

resolution of 100 meters using the GRASS program r.resamp.interpc. Bilinear

interpolation was applied during this resampling process to ensure accurate representation.

The custom LCC projection parameters utilized in the GRASS GIS project, represented in

Well-Known Text (WKTd) format, were as follows:

+proj=lcc +lat_1=48 +lat_2=52 +lat_0=46 +lon_0=-60 +x_0=0 +y_0=0 +no_defs

+a=678137 +rf=298.257223563 +towg84=0.000,0.000,0.000 +type=crs +to_meter=1

Title: Processing of GEBCO (2020 - 2022) -- Description: The GEBCO-2022 bathymetry grid contains some new data artifacts that required attention. Four problematic sections have been identified, each defined by their geographic bounds in the WGS84 datum format (west/east/south/north):

GEBCO-2022 Scotian Slope: -64.204835 / -63.267644 / 40.845932 / 42.005429

GEBCO-2022 Gulf of St. Lawrence 1: -61.718938 / -61.143213 / 49.763745 / 49.911131

GEBCO-2022 Gulf of St. Lawrence 2: -60.214804 / -60.089132 / 49.897717 / 49.972548

GEBCO-2022 Labrador: -56.100427 / -51.004948 / 53.517117 / 57.861460

For the GEBCO-2022 Scotian Slope section, the GEBCO-2019 grid quality was superior to the GEBCO-2020, so the GEBCO-2019 was clipped by setting the GRASS

computational region to that of the coordinates given above, and running r.mapcalce to clip the corresponding section of bathymetry from the GEBCO-2019 grid into a new raster dataset. This raster was then appended to the Amalgamated Small Surveys already described.

For the two Gulf of St. Lawrence sections, the GEBCO-2020 grid was the best version, so it was clipped using the same procedure described above. In the GEBCO-2022 Labrador section, there wasn't a clearly superior earlier version of GEBCO data to cover the data artefact at this location, so some geoprocessing was done on the GEBCO-2022 data itself to mask the artefacts.

After extensive experimentation, it was determined that a low-pass filter would be the most effective tool for reducing the relatively high-frequency noise associated with the artifacts. To achieve this, the GRASS module r.resamp.filterf was used. The module resamples an input raster, filtering the input with an analytic kernel. After setting the computational region to that of the GEBCO-2022 Labrador section, it was run using the following

command call:

r.resamp.interp input=GEBCO-2022_100m output=GEBCO-2022_100m_gauss

filter=box,gauss radius=1100,1100.

In this command, both a box and Gaussian filter were utilized, each with a radius of 1100meters. This value was determined through trial and error, striking a balance between noise reduction and signal preservation. The Gaussian-filtered section of the GEBCO-2022 data was then overlaid on the unprocessed GEBCO-2022 data within the same area.

General Bathymetric Chart of the Oceans (2019, 2020, 2022)

GEBCO Compilation Group (2020) GEBCO 2020 Grid (doi:10.5285/a29c5465-b138-

234de053-6c86abc040b9)

Title: Smooth Patching of Final Mosaic -- Description: The final step was to choose the drawing order for the rasters. The rule of thumb used was to assign the highest priority in the drawing order to those rasters with the highest native resolution. Other considerations that influenced the drawing order of the rasters were the data extent, quality, and data voids. After experimentation, the following final drawing order for the grid was chosen (top-most raster in the list draws on top of rasters lower in the list):

1. Amalgamated Small Surveys

2. NONNA-100 Bathymetry Grid

3. Amundsen 2013-2021 multibeam amalgamation

4. NOAA NCEI Multibeam Bathymetry

5. GEBCO_2022_Labrador Gaussian-filtered section

6. GEBCO-2022 Native Grid

The GRASS module r.patch.smooth was used to both overlay and patch the six finalized grids, and also to blend the areas of overlap between each pair of overlapping grids, such that the vertical transition between the grid edges was minimized. The following command-line parameters were used:

r.patch.smooth -s input_a=raster1 input_b=raster2 transition_angle=3

parallel_smoothing=11 difference_reach=9

A custom Bash shell script was written to use the command above in the following fashion:

Rasters 1 and 2 were used as the input parameters for r.patch.smooth, and

assigned to variables input_a and input_b, respectively. The module would then smooth the areas of overlap between these two rasters, using a spatially-variable overlap, as well as using the other command parameters specified above.

The output raster of this smooth-patching iteration would then be used as the input for the input_a parameter of r.patch.smooth in the next round of processing, and raster 3 as the input_b parameter. The output of this smooth-patching round would be run again with raster 4, and so on. The end result is a mosaic that has the input rasters drawn in the preferred order, with all of their boundaries smooth-patched.

The final step once all input rasters were smooth-patched into a single mosaic was to clip all land areas from the raster. To do this, a vector polygon shapefile representing Canada's coastline and Exclusive Economic Zone was downloaded from MaritimeRegions.orgo and imported into GRASS, converted to raster format, and then used as a processing mask to exclude all compilation grid values not contiguous with the polygon. A GeoTIFF of the final mosaic was exported at 100 m spatial resolution, using the custom

Lambert Conformal Conic projection specified in the beginning of the lineage section.

1. Amalgamated Small Surveys

2. NONNA-100 Bathymetry Grid

3. Amundsen 2013-2021 multibeam amalgamation

4. NOAA NCEI Multibeam Bathymetry

5. GEBCO_2022_Labrador Gaussian-filtered section

6. GEBCO-2022 Native Grid

Gridded multibeam bathymetry, Placentia Bay 1995, 2004, 2005, 2006, Newfoundland and Labrador, Canada

Data collected over multiple years as follows:

07/01/1995 to 07/05/1995

04/16/2004 to 04/25/2004

05/12/2004 to 05/30/2004

07/08/2004 to 07/12/2004

08/07/2004 to 08/29/2004

07/08/2005 to 08/19/2004

07/18/2005 to 07/27/2005

05/25/2005 to 06/02/2005

07/09/2006 to 07/13/2006

09/15/2006 to 09/28/2006

Gridded multibeam bathymetry, Port au Port 1996 and 1997, Newfoundland and Labrador, Canada

Data collected over multiple years as follows:

08/03/1996 to 08/14/1996

06/15/1997 to 07/01/1997

Title: Resampling NONNA-100 Bathymetry Grid -- Description: The Canadian Hydrographic Service's (CHS) NONNA-100 grids that overlaid the required compilation geographic extents were downloaded as a series of individual TIFF tiles and amalgamated into one raster dataset in GRASS.

The NONNA-100 dataset is comprised of blocks of multibeam bathymetry mosaics

interspersed with older, lower-resolution single beam survey bathymetry. Much of this single-beam data was of very poor quality and too widely-spaced to support resampling of the data voids without introducing serious artefacts into the data, and so the single-beam data were clipped out by running the GRASS module r.resamp.interp, using method=bilinear, which retained only the higher-quality contiguous multibeam bathymetry mosaics. Small data gaps within the remaining NONNA-100 multibeam blocks were filled by using three passes of the GRASS module r.neighborsh module, using a median filter with a 3x3 cell window size.

Title: Importing Additional Multibeam Bathymetry Data -- Description: During the years 2003-2014, the University of New Brunswick's (UNB) Ocean Mapping Group collected arctic multibeam bathymetry, backscatter and sub-bottom data on the CCGS Amundsen during the University of Laval's ArcticNet science programs. This collection of data was manually edited and delivered to NRCan in ArcGIS floating-point grid format. These grid files were in turn imported into the GRASS project using the module r.in.gdal, using the same custom Lambert Conformal Conic projection parameters as given above, at a resolution of 100m.

In addition, UNB also provided multibeam bathymetry data from 2014-2021, albeit in raw format. The data for these years was also manually edited using MB-System to remove data outliers. The cleaned and processed bathymetry was exported year by year as a geotiff, imported into GRASS and mosaiced together to create one raster dataset for all Amundsen surveys.

The UNB CCGS Amundsen dataset was augmented by CCGS Louis S. St Laurent

multibeam data collected in 2019 and two additional CCGS Amundsen

expeditions carried out by NRCan in 2020 and 2021.

Gridded multibeam bathymetry, ArcticNet 2003 to 2013, offshore northern Canada

Data collected by University of New Brunswick's (UNB) Ocean Mapping

Group onboard CCGS Amundsen 2003-2013.

Gridded multibeam bathymetry, ArcticNet 2014 to 2021, offshore northern Canada

Data collected by University of New Brunswick's (UNB) Ocean Mapping

Group onboard CCGS Amundsen 2014-2021.

Gridded multibeam bathymetry, Northeast Newfoundland shelf and slope 2019, offshore Newfoundland, Canada

Data collected onboard CCGS Louis S. St Laurent using the Kongsberg EM122 MBES in 2019.

Gridded multibeam bathymetry and bathymetric LiDAR, South Shore, Nova Scotia, Canada

The compilation includes 24 bathymetric LiDAR data sets and 12 multibeam bathymetry data sets.

Temporal extents for bathymetric LiDAR are as follows:

2011/03/27 to 2011/04/04

2012/03/20 to 2012/03/31

2014/03/12 to 2014/04/08

2016/12/04 to 2017/02/28

Temporal extents for multibeam bathymetry are as follows:

1998/09/04 to 1998/09/18

2005/06/11 to 2005/11/18

2006/09/10 to 2006/11/23

2008/05/18 to 2008/05/20

2008/06/01

2008/07/09 to 2008/07/17

2009/05/16 to 2009/05/18

2011/09/17 to 2011/12/27

2013/07/01 to 2013/07/31

2014/08/12 to 2014/11/13

2015/11/03/ to 2015/11/10

2018/07/21 to 2018/08/11

Gridded multibeam bathymetry and bathymetric LiDAR, Eastern Shore, Nova Scotia, Canada

The compilation includes 10 bathymetric LiDAR data sets. Known temporal extents for bathymetric LiDAR are as follows:

2014/03/29 to 2014/04/06

2014/04/06 to 2014/04/13

2016/07/20

Title: Processing Small Harbour Multibeam Surveys (1 of 5) -- Description: Several of the highest-resolution datasets available for use in this compilation were noncontiguous, high-resolution small harbour or coastal multibeam bathymetry and bathymetic LiDAR surveys of Newfoundland and Nova Scotia that were delivered to NRCan directly from CHS. These datasets were already cleaned and processed, so were used without further modification. The datasets were all reprojected to the LCC projection and resampled to 100m cell resolution using gdalwarp. They were then amalgamated in GRASS to simplify the total number of raster datasets and geoprocessing operations that would have to be done. CHS small harbour surveys, South Shore Bioregion compilation, and Eastern Shore bathymetric LiDAR compilation were already of high quality with no further processing required, so they were amalgamated and treated as one raster for the purposes of overlaying and comparison to other raster datasets

in the compilation; this raster dataset will be referred to hereafter as 'Amalgamated Small Surveys'.

Title: Merging and Interpolation of NOAA National Centers for Environmental Information (NCEI) Multibeam Bathymetry -- Description: The NOAA NCEI bathymetry portal is a web

service that allows users to download swath multibeam bathymetry data free of charge. The portal was accessed to download all multibeam survey data falling with the compilation extents. The downloaded data are stored in MB-System format, an open-source swath bathymetry editing and visualization suite. MB-System was used to manually edit and merge more than 40 surveys' worth of multibeam bathymetry data into one contiguous grid. Other than

manual flagging of bathymetric outliers from the raw data, the only other processing done was to apply spline interpolation to fill grid cells not filled by swath data; the interpolation filled data gaps up to a radius of 4 grid cells away from the cell being examined by the algorithm.

NOAA National Centers for Environmental Information. 2004: Multibeam

Bathymetry Database (MBBDB). [cruise]. NOAA National Centers for Environmental

Information. https://doi.org/doi:10.7289/V56T0JNC. Accessed [date].

Data are publicly available and free to use.

Link: https://www.ncei.noaa.gov/access/metadata/landingpage/

bin/iso?id=gov.noaa.ngdc:G01034