West Coast Multibeam Base Map Series

West Coast Multibeam Base Map Series
a proposal

John Hughes Clarke (1), Kashka Iwanowska (2), Mike Brissette (3) and Doug Cartwright (3)

1: Ocean Mapping Group, UNB

2 : Geological Survey of Canada - Pacific

3 : MosaicHydro, Victoria

Hi all,
I've been trying to rationalise the manner in which the growing GSC/CHS/Neptune Pacific multibeam compilations could be handled. What I've rehashed is an old quad tree model that I designed for the Geological Survey of Ireland to cope with their MB surveys from 5000m to 50m. I should say they didn't adopt it, choosing instead to stay with 1:250,000 base maps irrespective of water depth (reflecting their mandated mapping scale, rather than their data management needs).

I should emphasise that these map "sheets" are really not sheets (i.e.: they would never be printed). They are merely a standard series of terrain model dimensions that reflect the best compromise bewtween sonar resolution and array manipulation. Final products (bathymetry AND backscatter) at user defined projection, resolution and bounds are automatically generated by combining ("patchArea") these terrain models/ backscatter mosaics. To chose a custom area, the user does NOT need to return to the source data (huge volumes, pseudo-random requiring all the beam weighting cosmetology), rather he/she combines the highest resolution terrains/mosaics.

The underlying issue is that swath sonar resolution is inherently depth dependent. To date the multibeam surveys have been presented to the user community as terrain models whose extent reflects the bounds of a particular vessel deployment.The resolution was the finest that could be managed by the gridder. This was acceptable for the first few years, when the coverage is typically sparse and project focussed. But over the years ,what is clearly happening is that successive deployments of multiple platforms are being combined to build up regional coverage. The generalisation inherent in hydrographic chart products is not adequate for continuous surface data where the user is interested in the resolvable morphology rather than the shoallest depth. To produce a new blended compilation of multiple surveys currently requires that they all be brought back on line and regridded from scratch. A time consuming task and one that will grow as more surveys are added.

I see three main scales of mapping active on the West Coast: the shelf work (~80-400m, Barrie and co.), the delta fronts (~5-80m, Hill et al. and Venus) and the Juan de Fuca Plate (500m+, Neptune and others). The three programs could be handled separately, but the reality is that one ends up flowing into the other. So I've tried to come up with a way of seamless jumping between scales of resolution. Hardly new, and more rigorously implemented in Morton coding, etc.., but those models are designed to maintain source point data and seem less suitable for handling image data as opposed to just XYZ's.

20m pixel - > 1500m	10 m pixel < 1500m	5 m pixel < 400m	2.5 m pixel < 100m

40 min lat x 80 min lon	20 min lat by 40 min lon	10 min lat by 20 min lon	5 min lat by 10 min lon

Naturally there could be coarser sheets in the abyss and finer sheets for the beach. These two end members I've chosen to leave out for now (although it is just another line in the script per layer). The chosen pixels sizes are generally optimistic (with the exception of the shallow for now) but reflect the future of larger physical apertures (0.5 degree is approaching 1% of depth) and even synthetic aperture sonars (a dream!).

The mapsheets are subdivided based on the known bathymetry in the area. It is all automated, requiring you to have at least a low-resolution terrain model for the whole west coast offshore. For ease, I've used the global predicted bathy which is rather poor, but works for demo purposes. Note, for example that a series of higher resolution mapsheets are automatically generated for the top of Bowie Seamount. You can always generate a missing sheet if you strongly feel it is needed (a region not resolved by the global data set). I noted that GSC-P have digitised the CHS regional charts and I'd dearly love to get a copy of that data as it it far better represents the real shelf depths (predicted bathy goes walkabout in areas of crustal density variations).

I've putted a gzipped tar bundle of all the Box headers for these 4 base map scales here. You must get an updated version of jview that supports the -specbox option to have a look at them:

>jview background.map -specbox BASES_10mLAT_20mLON/selected/Box.*
(middle mouse click identifies map sheets name).

For now just use the 10x20 maps at 5m resolution as they are ideal for the 100-300m depth range of the Hecate surveys. The actual deliverables can be whatever area and projection Vaughn requests (just patchArea). But these map series will be ready to add in newer surveys as they appear (using the addWG functionality to blend bathy (comboMOS for mosaics) that I described).

And because you demanded it, they are UTM approximations. That is to say they are the closest fitting UTM coordinate rectangle that includes the desired lat - lon dimensions. Bear in mind you have to go through 3 UTM zones (central meridians -123, -129, -135). The map sheets use the zone that is closest to the sheet centre. As each sheet ends up having a a slightly rotated aspect there is actually natural overlap. Frankly it doesn't matter what projection this array series is, as you reproject for final.

You'll notice I've included the entire West Coast EEZ. This is because I see this as a natural way of handling the blue water swath sonar data that the Neptune project are trying to compile (Cheryl Katnick and Peter Phibbs). The neat thing about it for them is that, should they go for AUV surveys, the quad tree can naturally collapse down to include their ridge crest surveys at metre resolution. I'd suggest that the Barkley Sound cable corridor survey be handled with this approach. They may ask for a series of rotated map sheets along the corridor, but that is what patchArea is for.

Just to show you a proto-implementation, I've zoomed in on the Georgia-Juan de Fuca area. I've plotted the nav for the Georgia surveys, and a Tommy Thompson EM300 cruise off the margin.


all quads	40x80	20x40	10x20	5x10

These aren't cast in stoned yet and I'd like to discuss the feasibility of these dimensions and resolutions. But we want to come up with a standard so that whoever batch processes archived, current or future surveys, can clip them all together like lego in as painless a manner as possible.

last modd'ed by JHC, Feb 2005