In July Geoscience Australia released “Phase I” of its oceanographic data gathered during the search for the missing Malaysia Airways flight 370 (MH370). This data was gathered by three ships along an arch thought to contain the wreckage of the aircraft. The purpose of these missions was to collect high resolution bathymetric data in the search area so that it could be used for Autonomous Underwater Vehicle operations in Phase II. Before these Phase I surveys this area had very little bathymetric data at all. This article looks into what data is available to oceanographers, and discusses some of the more interesting features found. Don’t miss the youtube video at the end of the blog!
Geoscience Australia has made a story style webpage here. This website has quite a few flybys of the data an excellent overview of the entire survey effort. If you want to take a look at the bathymetry, or download portions of it, Geoscience Australia has made an online GIS portal. This second website is impressive. I assume it is an ArcGIS Online or ArcGIS server interface. I’ve never seen such detailed bathymetry on the web before. I believe they are doing the “pre-rendered raster tile layer trick” as you can see them load quickly, and they progressively get better resolution as you zoom in. Perhaps this is a lesson for us all in future online GIS bathymetry delivery. In addition the rendering of the bathymetry has a pleasing color bar and hillshade. I think its better then the standard Global Mapper examples in this blog!
Figure 1: Overview of the bathymetry and the search and transit areas.
Of more interest to oceanographers is the raw data which is located on Australia’s National Computational Infrastructure website. The data consists of processed backscatter, processed bathymetry, raw multibeam, subbottom, water column data, sound velocity data, and Caris files. Overall the dataset is about 2.1TB, with the raw multibeam being about 580GB, and the water column data at 1.0TB. (1TB = 1012 bytes)
Processed backscatter comes in a variety of formats including xyz (really x,y,amplitude), a geotiff of the backscatter in two flavors, and a set of esri grids. The processed backscatter tiff, mh370_backscatter_30m.tiff, does not contain RGB data and will import with, I assume, the average dB for the cell. Just be forewarned because in Global Mapper it will import this GeoTiff as an elevation geotiff and will look nothing like what you would expect. It does give a good result in ArcGis. The second geotiff, pansharp_backscatter_150m, is a false-color representation of the bathymetry. The processed backscatter is only available for the search area (Figure 2)
Figure 2: Overview of the processed backscatter.
Processed bathymetry is similar with xyz, three geotiffs (elevation, hillshaded, and pansharpened), and esri grids. There are two versions of the bathymetry, an overview binned at 150m, and bathymetry by survey area which has various smaller bin sizes and is separated by UTM zone. This second version of the bathymetry is called “optimal bathymetry”.
The optimal bathymetry is a better product for those of us that don’t have Caris and can’t really dive into the nitty gritty as it’s tighter bin sizes allow for more detail. However the optimal bathymetry data does include csar databases for Caris which will cut down time for those that want to reprocess the data entirely. If you are a bathymetry snob such as myself, there are plenty of kongsberg .all files, along with vessel files, delayed heave,and raw ctd/xbt data if you really want to reprocess the entire dataset.
In this case I loaded all the available optimal bathymetry xyz data in Global Mapper 17 (GM) to see what we were working with. This data spans 7 UTM zones, and came in bin sizes from 25m to 83m. I let GM figure out the bin spacing (it went with 20x30m), and had approximately 3x bin size search radius. Figure 1 shows the resulting bathymetry as seen in GM.
From the bathymetry we can see that the vessels were usually going in and out of Perth and recording multibeam the entire way. Backscatter isn’t processed from these areas. it’s unclear if water column data is recorded during these transits. Backscatter should be able to be recovered from the raw data. For oil and gas interests these transits will be of utmost importance as they pass over the flank of the Naturaliste Plateau and near Mentelle Basin near Perth. Both areas have been considered for oil exploration in the past.
As for the actual seabed morphology in the search area (figure 1) we can see that there are chains of seamounts, big scars in the seafloor, and a strange repetitive pattern in the deep waters. Figure 3 shows an example of one of these submarine volcanos just south of the Broken Ridge/Diatamious Trench.
Figure 3: Location of an example submarine volcano.
Figure 4 shows a close up of the volcano, it is about 3,000m deep at the top, and nearly 1,000m taller than the surrounding seafloor. Clearly visible is the volcano’s crater at the top.
Figure 4: Closeup of the volcano.
Figure 5 shows the backscatter data from this volcano. Near the tops and along the sides of this volcano are spots of high reflectivity. These areas are large but look similar to hardgrounds (chemosynthetic communities) in the Gulf of Mexico. it will be interesting to review the water column data to see if there is any hydrothermal vent activity near these areas. My hypothesis is that these areas are potential active hydrothermal vents.
Figure 5: Backscatter data across the volcano.
Figure 6 is a 3D image of the volcano with backscatter draped over the bathymetry.
Figure 6: A 3D rendering of the volcano.
In the farther southern reaches of the search area the seabed becomes more like what we expect from the abyssal plain (flat), however there is also a large amount of of NW – SW oriented ridges generally with 100m in relief. It’s unclear (to me) what the process is that makes these ridges. I at first thought these ridges were just bad multibeam but that was quickly ruled out. Later I thought they could be very large mega- furrows. A quick look at the literature suggests that thermohaline flow could result in large sediment waves, but these ridges seem to be parallel to the expected direction of flow (like furrows). Also the extent of these features is incredible, stretching just south of Broken Ridge through all the southern search area data and then east into the transit data stopping at the Diamantina Fracture Zone. Figure 7 shows a close up of the features, and a profile. Figure 8 shows how only the deepest ridges show up in the backscatter.
Figure 7: Closeup of the mystery ridges.
Figure 8: Backscatter in the southern search area. Only the very deep ridges are visible.
In the transit data we can see the extreme seafloor of the Diamantina Fracture Zone. It’s in this area that the deepest recorded depths of the dataset are seen (7,018m), with some of the highest gradients. Figure 9 shows a close up of the Diamantina Fracture Zone. The data is incomplete as this is transit data and not the primary search area. Figure 10 shows a plot of the gradients observed with red being over 25°.
Figure 9: Closeup of the Diamantina Fracture Zone.
Figure 10: Gradients from the Diamantina Fracture Zone.
Finally we take a look at the areas that will probably get reviewed by the oil and gas exploration industry, this is the region near Perth and includes Naturaliste Plateau and the basin just west of Perth Canyon. The Naturaliste Plateau has been explored for oil and gas in the past and has been drilled by the Ocean Drilling Program (for scientific research not oil and gas exploration). This part of the transit data is more complete as the ships tracks were all converging on Perth. The northern flank of the Naturaliste Plateau is observed along with a few submarine volcanos. Interestingly it appears that some large sediment waves are on the top of the plateau and are clearly observed in the bathymetry. This adds more proof that this area might have stronger than expected seabed currents.
Figure 11: Naturaliste Plateau and approaches to Perth.
As for the impact on oil and gas exploration this data will have it is unclear. At the time of this writing the extent of the water column data from the dataset is unknown. The dataset does contain a note that says the water column data wasn’t recorded on every line. It would be a shame if the was not recorded while over the Naturaliste Plateau as water column data can indicate the presence of gas seeps. Also the backscatter will have to be processed for the transit data to see if any hardgrounds can be observed near the Naturaliste Plateau or Perth Canyon as hardgrounds can also indicate seabed fluid flow.
Multibeam datasets aren’t typically used for search. Backscatter has always been thought of as second rate compared to sidescan. Sidescan, of course, providing a much closer look at the seabed at much higher resolution. Still, in previous efforts to identify hardgrounds via deep water multibeam backscatter in the Gulf of Mexico I have found that sometimes oilfield infrastructure can be imaged by backscatter. The hard metal of a manifold can reflect enough sonar to show up in the backscatter and shows up as a bright spot in the water column data. In this case it is unclear if a debris field of a plane would have the same effect. It would require an examination of the backscatter at the ping level as well. The processed backscatter is binned at 30m and that would mean the debris field would only be a few pixels at most. It goes without saying that most of this area was later surveyed by sidescan in the second phase of the search and it would be unlikely that the onboard interpreters missed anything.
I have added my Global Mapper Grid (800MB) and a RGB version of the backscatter(350MB) to a Dropbox.
I hope you enjoyed this quick review of the MH370 Phase 1 dataset. For more information you can find me on linked in or email me at huxrules@gmail.com (actual corporate email hidden because of cyber).
Governments of Australia, Malaysia and the People's Republic of China, 2017. MH370 Phase 1 data. doi: 10.4225/25/595d7744b71e2p
Recent Comments