should we provide a local, high-res topo dataset?

[dankelley]

I was musing (#1755 (comment)) that it might be nice to have a higher-resolution topography dataset.

The problem with that is that even reducing the bin width by a factor of 2 increases the file size by a factor of 4, which is a problem because this is the biggest dataset in oce already.

Another idea might be to provide a high-resolution file for a small region (I would say, the Maritimes), so that folks could see the sort of resolution that is easily downloadable nowadays.

I'm sort of thinking out loud here. This is my first 'discussion' item, and partly I am posting it here and not as an issue because I view 'issues' as 'problems', and my goal here is, literally, to promote discussion.

[richardsc]

I like this idea. I'm pretty used to calling up marmap to get what I want, but it requires a connection (and leaves cast-off files in its wake) and adds a few lines to what should be a simple example. The "Northwest Corner" would definitely be a good candidate.

[dankelley]

@clayton33 -- how does the following look? The Q is whether there is a brownish hue in waters south of Shelburne (southern tip of NS). I don't see that hue there anymore, but I have troubles with some colours, it seems! I'm not worried about a few pixels that don't line up with the coastline (like in Cape Breton); these are inevitable near steep topography unless you zoom in the topo a lot.

This uses 1 minute topo (pixel size 1.8km in north-south direction, and 1.4km in east-west direction, at this lat). The .rda file is 159K, so I think that would be okay to include this.

I think this is a lovely plot, actually. Part of it is that I know the land reasonably well, having driven up the various hills (mountains) and along low-lying areas, too. The familiarity makes it fun to look at. And I want to advertise the Maritimes in this dataset.

I find the detail fascinating, e.g.

• There are some blue smudges in the middle of NB, because the topo shows the bottom of the lakes, not the top, and those lakes are at sea level. (See also https://en-ca.topographic-map.com/maps/qun/New-Brunswick/ ... which has a neat colour scale for land ... oh no, another rabbithole)
• You can see the LaHave river valley going pretty far inland.
• Cape Breton has some high land. I can see why folks from there look down on the rest of us.
• Speaking of being looked-down upon, the pic shows the low land of God's country the Tantramar Marshes clearly.

PS. the code for this is not on GH yet. I would like to get some more eyes on this first (and I apologize for taking up folks' time).

[clayton33]

Oooo, this looks nice.

[richardsc]

Ok yes, that does look pretty good.

I'd like to point out that probably much of this "discussion" should probably have been done inside issue #1757 .... just saying 😄

[dankelley]

The funny thing is that I made that issue, but then wrote comments here. Oh well. This was a test of GH discussions. I like the way it hides stuff in the 'read more' link. That's a form of threading, I guess. But this is not the kind of threading I was really hoping for in GH discussions. I was hoping that e.g. there would be a little triangle symbol for this whole sidetracking. Many issues end up with sidetracking.

Anyway, the new datafile is only 87K so I think that should be fine for adding.

Unless someone objects, I'm going to go ahead and insert that into oce this today at 1300h. I'll write up some docs for it, of course. And I'll see whether some examples can use it, instead of topoWorld. If I were a user, and saw a pretty map like this, I'd be excited to try it myself.

PS. I use download.topo(), which gets a netcdf file that is much smaller than the csv that marmap downloads, and that loads faster, too, because it's already in matrix form. But the server changed, and so the download.topo() from the CRAN version of oce won't work. Yet another reason to consider a new CRAN release in January, I think.

Oh, and the vignette could use it, too. For example, it might be neat to write a bit of code to go along the Halifax line and extract bathymetry along it. Possibly doing such an extraction might be a useful exercise for @j-harbin, who knows people who know the lon-lat of the Halifax line...

[dankelley]

I've added data(topoNS), and example(topoNS) now plots it, first in contour form and then in image form.

[dankelley]

I gave up on this idea because, during the last CRAN release, I wanted to remove every unneeded byte from the package, to get in under the size limit.