oceanographic surveys target GENERAL OF areas of interest such as trench- es, ridges and . But it THE PACIFIC means the existence of these fea- tures must be known prior to the survey planning and it Because of the increasing pressure requires baseline information on on sedentary marine species — Franck Magron the probable location of unsur- both in coastal areas and offshore Reef Fisheries Information veyed underwater features. seamounts — there is an increas- Manager ing need for mapping the poten- SPC, Noumea In June 1995, the US Navy tial habitat of these resources so New Caledonia declassified Geosat altimeter that fisheries managers and deci- ([email protected]) data (height of the surface) sion makers can make informed which, combined with ERS-1 decisions about coastal resources. Oceanic Commission (IOC) of radar data, allowed Smith and UNESCO digitised available Sandwell (1994) to derive a map Contrary to land resources, which marine charts (contour and track of marine gravity anomalies can be mapped using satellite lines based on soundings, coast- used later to predict the seafloor products, the marine environ- lines) to produce a global 1- depth between the surveyed ment is more difficult to map minute bathymetric grid known bathymetry tracks. In 1997, the because water absorbs visible as the General Bathymetric Chart authors produced a two-minute light and radar frequencies very of the (GEBCO). Because it global map of predicted quickly, and the ocean is basically was produced from marine charts, seafloor for lati- opaque from the sky for depths this grid will not provide more tudes between 72°S and 72°N beyond 50 m, even using airborne information than paper charts, but (Sandwell and Smith 1997). Lidar (laser bathymetry). it is a convenient global digital grid of ocean bathymetry. This initial work was later TRADITIONAL OCEANOGRAPHIC refined and several derived SURVEYS PREDICTED BATHYMETRY products are now available to the general public, combining High-resolution mapping of the Because mapping the ocean the predicted bathymetry with ocean floor requires expensive floor is difficult and onerous, other sources of information equipment (multi-beam and side- scan ), operated from oceanographic ships with high operational costs. For these rea- sons, direct ocean floor mapping is very limited, and traditional marine charts can miss important relief features outside the areas surveyed by hydrographic ships.

Figure 1 depicts the sounding lines used to produce the marine chart NZ14606. It shows that some areas have been intensively surveyed, such as the Tonga Trench or Savan- nah chain (French Poly- nesia), whereas there is much less information for the Cook Islands or Line Islands (Kiribati). This is not necessarily a problem when the area is a large , but it does not allow the systematic inventory of underwater features such as seamounts and ridges.

On a global scale, the International Hydrographic Organization Figure 1: Diagram of sounding line density (IHO) and the Intergovernmental of chart NZ14606 and Pacific Island EEZs

SPC Fisheries Newsletter #117 – April/June 2006 49 GENERAL BATHYMETRY OF THE PACIFIC OCEAN such as GEBCO or Shuttle tory of potential seamounts quence, are often falsely Radar Topography Mission using a filter that detects peaks detected as one or more (SRTM) elevation data, and and searches for local rises of seamounts, as depicted in Figure with various grid sampling 1000 m or more from the . 5 on p. 52. Screening and cleaning (from 30 seconds to 5 minutes). Kitchingman and Lai (2004) used of this data are currently under- the Global Digital Elevation taken by the Pacific Islands While Etopo2 is certainly the Model (Etopo2) grid and identi- Oceanic Fisheries Management best known global seafloor pre- fied between 14,000 and 32,000 project (SPC Oceanic Fisheries dicted bathymetry product, the potential locations of seamounts, Programme/ GEF). original Smith and Sandwell Wessel (2001) used the S&S grid (S&S) data were slightly mis-reg- to extract around 15,000 HIGH-RESOLUTION MAPPING istered in latitude and longitude seamounts locations for the OF SEAMOUNTS when the grid was produced whole world. (Marks and Smith 2004), and this Once seamounts and other shift is apparent when compar- The methodology (filter and thresh- underwater features are identi- ing the predicted bathymetry olds) and source data grid used for fied, either from previous with data (see Figs 3 and 4 the inventory of these potential hydrographic surveys or using on p. 52), or when displaying seamounts have coastlines produced from a significant Landsat 7 ETM+ (visible and impact on the infrared) or SRTM elevation data results and their on top of shaded bathymetric usability. Because data. Note that predicted the Etopo2 grid bathymetry matches here the is mis-registered, sonar data because the very so are potential same sonar data were used to seamounts (the produce the grid (and interpolat- distance between ed using gravity anomaly data). Wessel and Kitchingman The newer S2004 bathymetric seamounts can grid produced by Walter Smith be up to 10 km). from the S&S data and blended with the GEBCO data is properly Moreover, pre- registered. It uses GEBCO data dicted bathym- for depths shallower than 200 m, etry is less reli- and a blend of the two datasets able in areas between 200 and 1000 m. That shallower than mitigates the fact that predicted 200 m, around bathymetry is less reliable for land masses, shallow water and the resultant and is averaged grid is the best global bathymet- because of the ric grid available at the moment initial S&S two- Figure 2: Satellite altimetry used (see Table 1). minute resolu- produce gravity anomaly grid tion. As a conse- (reproduced from Sandwell and Smith 1997) The shaded bathymetry produced from S2004 Table 1 Bathymetric grids and source data for each country can be found on the SPC Grid Resolution Source Comments Mis-registration in latitude and PROCFish Portal in the Etopo2 2 minutes S&S, IBCAO, DBDBV, GLOBE GIS repository1. longitude Correctly registered but smoothing GINA 30 seconds S&S, IBCAO, GTOPO30 effect observed (Marks and Smith 2004) INVENTORYOF GEBCO 1 minute Charts contour lines Chart accuracy, smoothing effect SEAMOUNTS S2004 1 minute S&S, GEBCO Correctly registered The availability of glob- IBCAO: International Bathymetric Chart of the Arctic Ocean al bathymetric grids DBDBV: Digital Bathymetric Data Base - Variable Resolution made it possible to con- GTOPO30: Global 30 Arc-Second Elevation Data Set duct a systematic inven- Globe: Global Land One-kilometer Base Elevation

1 http://www.spc.int/coastfish/sections/reef/PROCFish_Web

50 SPC Fisheries Newsletter #117 – April/June 2006 GENERAL BATHYMETRY OF THE PACIFIC OCEAN predicted bathymetry, they can the mis-registration observed in Marks K.M. and Smith, W.H.F. be fully mapped by oceano- some products and because of 2004. Not all bathymetry graphic vessels using multi- the limited resolution of these grids are created equal. beam and side-scan sonars. grids. Submitted to Marine Figure 6 (p. 52) shows the Capri- Geophysical Researches, corn Seamount as displayed Figure 7 summarises the vari- GEBCO Special Issue with a one-minute resolution ous bathymetric grids and August 25, 2004. In Press. (S2004 grid) and with a 200 m datasets cited in this article and resolution (multibeam data). indicates how they are related. Sandwell D.T. and Smith W.H.F. 1997. Marine gravity from While the multibeam sonar cap- A future article will examine Geosat and ERS-1 Altimetry. tures bathymetry, a side-scan methods that can be used to Journal of Geophysical sonar captures texture and mor- produce shallow water bathym- Research, 102:10039–10054. phology. The side-scan sonar etry maps (between 0 and 50 m). reflects the type of substrate and Smith W.H.F. and Sandwell D.T. habitability of the area for deep REFERENCES 1994. Bathymetric prediction bottom fish species and, when from dense satellite altimetry available, can be mapped on top Kitchingman A. and Lai S. 2004. and sparse shipboard bathy- of the bathymetry on a three- Inferences on potential metry. Journal of Geophysical dimensional model. seamount locations from Research 99:2180–21824. mid-resolution bathymetric High-resolution data for data. pp. 7–12. In: Morato T. seamounts that have already and Pauly, D. (Eds). FCRR been mapped is generally avail- seamounts: Biodiversity and able from the Internet, in particu- fisheries. Fisheries Centre larly from the Seamount Catalog Research Reports. University of the Seamount Biogeosciences of British Columbia 12:7–12. Network (http://earthref.org), from where it is possible to download multibeam data, mixed with predicted bathyme- try. On the PROCFish portal, there is a MapInfo file with the location of seamounts referenced in the catalog with direct links to the seamount pages.

The linkage between the deep bottom fish resources and seamounts is currently a research topic and is one of the topics of the Marine Geological Habitat Mapping (GeoHab)2007 conference that will be held in New Caledonia in 2007.

CONCLUSION

While only a small part of the Pacific Ocean has been thorough- ly mapped by oceanographic vessels, predicted bathymetry is available globally and allows the localisation of underwater fea- tures such as seamounts, which can be surveyed in detail at a later time using multibeam and side-scan sonars. Yet some cau- tion is necessary when using S&S-derived products because of Figure 7: Datasets and their relationships

SPC Fisheries Newsletter #117 – April/June 2006 51 GENERAL BATHYMETRY OF THE PACIFIC OCEAN

Figure 3: S2004 shaded relief and sonar sounding points from NOAA’s National Geophysical Data Center

Figure 4: Extraction of Etopo2 and S2004 values with corresponding sonar data along the transect shown in Figure 3

Figure 5: Kitchingman (triangles) and Wessel (stars) potential seamount locations

Figure 6 Shaded relief of Capricorn Seamount near Tonga trench at one-minute (S2004) and 200 m (sonar) resolution

52 SPC Fisheries Newsletter #117 – April/June 2006