WIP: support for a few `naturalearthdata` datasets

haakon-e commented 1 year ago

WIP on support for plotting various datasetes from naturalearthdata.com.

The website provides a plethora of datasets that are very nice to use as background images / overlays on Earth-themed plots.

Some comments:

Technically, this PR primarily adds support for reading certain shapefiles using the Shapefile package.
Some files in the illustrative examples, below, are quite large and you'd likely not want to include them in this package. Some options are:
- Add the data URLs as Artifacts. I don't know much about this option, but it seems like all data is downloaded when the package is installed. This could be nice for small datafiles, but may be inconvenient for files larger than e.g. 100MB.
- Use Downloads (or a similar package) to download these files upon request. This is already done for certain examples in this repo.
- Let the user provide a path to the functions, leaving it in the user's hands to obtain any relevant data.
- This is the wrong repo -- there should be another repo that provides this type of support, and adds relevant methods that GeoMakie/Makie can interface with.

Personally, I'm most interested in plotting more accurate coastlines and ocean bottom, so will probably not work on too much support beyond this.

Note: Just before posting I saw that they may actually have .geojson-type data available for download via github [link], so the Shapefile support may not be needed (at least for these datasets). If the maintainers prefer people to individually download any datasets they may need, I may be happy with adding some info about this to the docs, for others to enjoy.

1. Higher-resolution `coastlines()`

For this example, I assume the relevant data exists in ~/Downloads/ne_10m_coastline/ne_10m_coastline.shp (download here)

julia> using GeoMakie, GLMakie

julia> fig = Figure(; resolution=(1800,1000))

julia> ga11 = GeoAxis(fig[1,1]; coastlines = false, dest = "+proj=longlat", title="World");

julia> ga12 = GeoAxis(fig[1,2]; coastlines = true, dest = "+proj=longlat", title="Norway", lonlims=(2.5, 20), latlims=(57, 65));

julia> ga21 = GeoAxis(fig[2,1]; coastlines = true, dest = "+proj=longlat", title="Greece", lonlims=(19, 32), latlims=(35, 42));

julia> ga22 = GeoAxis(fig[2,2]; coastlines = true, dest = "+proj=longlat", title="Hawaii", lonlims=(-161, -154), latlims=(20, 23));

julia> lines!.([ga11, ga12, ga21, ga22], Ref(GeoMakie.coastlines_10m()); color=:red);

2. Bathymetry contours

Assumes you have the folder ~/Downloads/ne_10m_bathymetry_all/ (download all here)

julia> fig = Figure(; resolution=(1800,1000))

julia> ga = GeoAxis(fig[1,1]; coastlines = true, dest = "+proj=longlat");

julia> bathyplot = lines!(GeoMakie.bathymetry(2000); color=(:red, 0.8));

julia> bathyplot = lines!(GeoMakie.bathymetry(3000); color=(:darkred, 0.8));

(Note: I'm not sure if the long straight-ish lines are due to one-off indexing errors on my part in how I read the raw data or if they're actually a part of the dataset).

3. Detailed ocean bottom

Assumes you have the file ~/Downloads/OB_50M/OB_50M.tif (download here)

julia> fig = Figure(; resolution=(1800,1800))

julia> ga = GeoAxis(fig[1,1]; coastlines = true, dest = "+proj=longlat", lonlims = (-180, -125), latlims=(-40,-15));

julia> ga2 = GeoAxis(fig[2,1]; coastlines = true, dest = "+proj=longlat", lonlims = (-180, -125), latlims=(-40,-15));

julia> img = image!(ga, -180..180, -90..90, rotr90(GeoMakie.earth()); interpolate = false);  # regular earth

julia> img = image!(ga2, -180..180, -90..90, rotr90(GeoMakie.oceanbottom_50m()); interpolate = false);

# Optionally, also add the high-res coastplot

julia> coastplot = lines!(ga2, GeoMakie.coastlines_10m(); color=:black);

julia> coastplot = lines!(ga, GeoMakie.coastlines_10m(); color=:black);

Here the combination of detailed bathymetry and detailed coastplot really shines, and it becomes apparent the regions where the bathymetry permeates the ocean surface as islands.

asinghvi17 commented 1 year ago

I wouldn't mind spinning this out to a NaturalEarth.jl, which downloads on request then caches using Artifacts, and we could depend on that from GeoMakie. What's your take on this idea @haakon-e? It could probably live in JuliaGeo.

haakon-e commented 1 year ago

I wouldn't mind spinning this out to a NaturalEarth.jl, which downloads on request then caches using Artifacts, and we could depend on that from GeoMakie. What's your take on this idea @haakon-e? It could probably live in JuliaGeo.

Yes, I think a separate repo makes more sense. I'll have to learn how to conditionally cache with Artifacts, but I think that'll be a really nice contribution to the ecosystem. I'll see if I get time to think more about this in the next few weeks. Thanks for your feedback!

asinghvi17 commented 1 year ago

Yeah the Artifacts bit shouldn't be too bad - you just have to mark the artifact which you don't want to pre-download as lazy. Presumably we could also specify some to be auto-downloaded.

Natural Earth has a github repo with all their vector files up as GeoJSON or Shapefile...theoretically, we should be able to retrieve that GeoJSON by downloading the raw files from Github, which would be a pretty clean solution to our artifact problem. People could retrieve the dataset they want by name directly (artifact"ne_10m_coastline") or by some easy accessor function in the same vein as what you wrote in this PR.

haakon-e commented 1 year ago

Thanks @asinghvi17 I added some initial commits here: https://github.com/asinghvi17/NaturalEarth.jl/

MakieOrg / GeoMakie.jl