The Hui Feng was a 121-foot longline fishing vessel that grounded on Palmyra Atoll in 1991. Iron from this shipwreck was promoting the growth of an invasive corallimorph that smothered previously healthy corals.. The ship was removed during the winter of 2013-2014. Some articles describing the problem and the ship removal include:
In partnership with USFWS and the Scripps Institution of Oceanography, the University of Miami and University of Girona have created a series of underwater landscape mosaics for the area near the longliner grounding site. Two types of mosaics were made. One type was created from a towed array of gopro cameras. The second type was created from a pair of diver-swum DSLR cameras. The towed array covers a large area (approx. 130 x 300 m) at lower resolution (approx 6 mm/pixel), wheras the diver array covers a smaller area (approx 10 x 20 m) at higher resolution (< 1mm/pixel).
(click on any image for a larger version)
An array of gopro cameras were towed to the northwest of the grounding site during the summer of 2014. These are three views of the mosaic from 2014. Currently the mosaic is rendered in two parts, which is why there is a line through the middle of it. The two parts do overlap, however, and we are working on a seamless verion. The source data for these mosaics was gopro video, from which 336,242 frames were extracted for processing. Out of these 336K frames, 69,416 were selected with at 75% overlap criterion. In other words, the frames that had > 75% overlap were discarded. The original frames had approximately 2-4 mm/pixel resolution depending on the depth of the water at any given location. The mosaics were rendered at 6mm/pix. The total sizes of the mosaics are 29,095 x 29,822 pixels (south block) and 23,825 x 40,048 pixels (north block).
The above images show the southeast corner of the mosaicked area, which includes the area of the grounding scar. The left images shows the wreck visible in imagery from 2009 available on Google Earth. The center image shows the mosaic displayed in 50% transparency over the google earth image to show the alignment of the mosaic and the previous locaiton of the wreck. The right image shows the same area with no transparency, so you can see the area of the scar very clearly in the mosaic.
The full resolution of the mosaics can be explored interactively with the online tool "Zoomify."
The above images show the southeast corner of the mosaicked area, including the scar left by the vessel. A structure-from-motion technique was used to create a digital elevation model (DEM) of the seabed. The DEM is shown here with the image texture draped over it (left) and without texture in a sunshade view (right). Click on any image to see a larger version.
(click on any image for a larger version)
An array of gopro cameras were towed to the northwest of the grounding site during the summer of 2013. These are two views of the mosaic from 2013. Currently the mosaic is rendered in two parts, which is why there is a line through the middle of it. The two parts do overlap, however, and we are working on a seamless verion. The source data for these mosaics was gopro video. 43,501 images were selected for the north block and 29,469 for the south. The original frames had approximately 2-4 mm/pixel resolution depending on the depth of the water at any given location. These 2013 mosaics were rendered at 12mm/pix. The total sizes of the mosaics are 29,215 x 19,952 pixels (south block) and 27,555 x 24,408 pixels (north block).
These images compare the area mapped in 2013 vs 2014. On the left is just the 2013 images. On the right is the 2013 images with the 2014 images superimposed. Note 2013 covers a larger area, but at lower resolution.
Image sizes: 2013 N ~ 330 x 290 m (95,700 m^2) 2013 S ~ 350 x 240 m (84,000 m^2) 2014 N ~ 140 x 240 m (33,600 m^2) 2014 S ~ 174 x 180 m (31,300 m^2)
Note: the GeoTiffs for both the 2013 and 2014 surveys have been rendered in UTM zone 17N by mistake. They should be zone 3 N. The coordinates are correct just the zone is wrong. In many GIS packages you can correct this after loading (just change the projection). If you can't do that in your favorite GIS then you'll have to reproject the images themselves.
