option DCW issue

[liming-he]

Might not be a bug. For the following code, if I add "DCW=(country=:CA, pen=(0.225,:black))", the Canada border will be shown, but the grid line (lat, lon) will not show.

If I remove "DCW=(country=:CA, pen=(0.225,:black))", the grid line shows well. Any reason why the country and grid line repel each other?

Thanks. Liming.

for yy = 1: 1
file_out = @sprintf("Pixel_%s_%04d.png", this_variable, year(year_list[yy]));

file_out = joinpath(data_path1, file_out);
println(file_out)

x.z[:] = reverse(transpose(data[:,:,yy]), dims=1);
x.z[x.z .== 0] .= NaN


# figure 7
topo = makecpt(color=:jet, range= (0, 250, 10), continuous=true) #make colorbar  other options: color=:polar
gmt("set MAP_GRID_PEN_PRIMARY thinnest,.");

grdimage(x, show=false, 
    interp = "n", 
    frame=(axes=(:left_bare,:bottom_bare,:right_bare,:top_bare),  annot=true, ),
    coast=(**DCW=(country=:CA, pen=(0.225,:black)),** frame=(axes="WSEN", grid=10,  annot=10, ticks=2, ), ),
    colorbar=(pos=(anchor=:BC,length=(10.5,0.5), horizontal=true, offset=(0,.8), ), color=topo, show=true, frame=(ylabel=:"mm year@+-1@+",ticks = :auto, annot = :auto) ), 
    par=(MAP_ANNOT_ORTHO=:n, MAP_TITLE_OFFSET = 0.8, ), fmt=:png,
    savefig = file_out, dpi=600);

end

[joa-quim]

Hmm, can you please provide an example that I can reproduce?

[liming-he]

here is testing code.

my intention is to plot figure with both grid lines and CA border.

a file with projection is attached (tiff but zipped).

thanks a lot.

Mascons_JPL_result.zip

using GMT

testing_path = "/mnt/hdd1/GRACE_data/Mascons_JPL";
data_path1 = "/mnt/ssd1/Forcing/EALCO_Output";

input

file1 = joinpath(testing_path, "Mascons_JPL_result.Tiff");
x = gmtread(file1, grid=true, band=:1);

output

file_out = joinpath(data_path1, "test1.png");
x.z[:] .= 1:960*1140;

############### version 1 ####################

this produces figure with Canada border but without grid lines.

topo = makecpt(color=:jet, range= (0, 960*1140, 10), continuous=true) #make colorbar other options: color=:polar
gmt("set MAP_GRID_PEN_PRIMARY thinnest,.");
grdimage(x, show=false,
interp = "n",
frame=(axes=(:left_bare,:bottom_bare,:right_bare,:top_bare), annot=true, ),
coast=(DCW=(country=:CA, pen=(0.225,:black)), frame=(axes="WSEN", grid=10, annot=10, ticks=2, ), ),
colorbar=(pos=(anchor=:BC,length=(10.5,0.5), horizontal=true, offset=(0,.8), ), color=topo, show=true, frame=(ylabel=:"GRID ID",ticks = :auto, annot = :auto) ),
par=(MAP_ANNOT_ORTHO=:n, MAP_TITLE_OFFSET = 0.8, ), fmt=:png,
savefig = file_out, dpi=600);

############# version 2 ###############################

this produces figure with shorelines and grid lines.

file_out = joinpath(data_path1, "test2.png");
gmt("set MAP_GRID_PEN_PRIMARY thinnest,.");
topo = makecpt(color=:jet, range= (0, 960*1140, 10), continuous=true) #make colorbar other options: color=:polar
grdimage(x, show=false,
interp = "n",
frame=(axes=(:left_bare,:bottom_bare,:right_bare,:top_bare), annot=true, ),
coast=(frame=(axes="WSEN", grid=10, annot=10, ticks=2, ), ),
colorbar=(pos=(anchor=:BC,length=(10.5,0.5), horizontal=true, offset=(0,.8), ), color=topo, show=true, frame=(ylabel=:"GRID ID",ticks = :auto, annot = :auto) ),
par=(MAP_ANNOT_ORTHO=:n, MAP_TITLE_OFFSET = 0.8, ), fmt=:png,
savefig = file_out, dpi=600);

[liming-he]

[joa-quim]

Sorry, there seem to be issues here but I cant even reproduce your figures. This is what I get

(which GMT.jl version are you using?)

grdimage(x,  interp="n", frame=(axes=(:left_bare,:bottom_bare,:right_bare,:top_bare), annot=true), coast=(frame=(axes="WSEN", grid=10, annot=10, ticks=2)), colorbar=(pos=(anchor=:BC,length=(10.5,0.5), horizontal=true, offset=(0,.8)), color=topo, show=true, frame=(ylabel=:"GRID ID",ticks = :auto, annot = :auto)), par=(MAP_ANNOT_ORTHO=:n, MAP_TITLE_OFFSET=0.8), show=1)

[liming-he]

Yours read the image file correctly. It is strange to see a different version of grid lines.

Mine:
GMT v1.6
Julia 1.9
Ubuntu 20.04

That single line produced below:

[joa-quim]

I don't have an answer for now. This is a harder case where one have a projected grid and want to overlay coastlines that guess the right projection. This is a case that I've worked a couple times (fixing previous failures) but current version (1.10) seems to be behaving worst than the one you have for this case. I will need more time to understand what/why things are failing the way they are.

[liming-he]

Thanks a lot!

Also to mention, I encountered these warning/errors when using grdimage. But the plot looks nice anyway regardless of these errors.

┌ Warning: Que exagero de cores
└ @ GMT ~/.julia/packages/GMT/Mxc9g/src/gmt_main.jl:1082
ERROR 1: Point outside of projection domain
ERROR 10: Pointer 'hTransform' is NULL in 'OCTTransform'.

ERROR 10: Pointer 'hTransform' is NULL in 'OCTTransform'.

[joa-quim]

First warning (in portuguese, sorry) is due to the fact that your makecpt command generates range= (0, 960*1140, 10) 109440 colors. The others are more complex to figure out but are coming from GDAL complaining that data points are outside domain (part of the problem that needs more understanding).

[joa-quim]

OK, I found the regression and now I'm also able to reproduce your results. The issue seems ti lie deep in the GMT (C lib) interaction with GDAL. If you have a standalone GMT CLI, you'll see that this works (it uses the GMT syntax for establishing the projection)

gmt coast -ECA+p0.225,black -R-122.913612/36.214964/-9.977657/62.430681+r -JL-95/0/49/77/15c -Bpa10f2g10 -BWSEN -Da -png lixo

but this one, that uses the PROJ4 syntax, prints many complains and ends up not plotting the grid lines.

gmt coast -ECA+p0.225,black -R-122.913612/36.214964/-9.977657/62.430681+r -J+proj=lcc+lat_0=0+lon_0=-95+lat_1=49+lat_2=77+width=15c/0 -Bpa10f2g10 -BWSEN -Da -png lixo
ERROR 1: Point outside of projection domain
ERROR 1: Point outside of projection domain
...

[liming-he]

Thanks. Could you share which version of GMT CLI you used?

$ gmt coast -ECA+p0.225,black -R-122.913612/36.214964/-9.977657/62.430681+r -J+proj=lcc+lat_0=0+lon_0=-95+lat_1=49+lat_2=77+width=15c/0 -Bpa10f2g10 -BWSEN -Da -png lixo
gmt [ERROR]: Syntax error: Unrecognized keyword FONT_SUBTITLE.
gmt [ERROR]: Syntax error: GMT_VERBOSE given illegal value (warning)!
gmt [ERROR]: Syntax error: Unrecognized keyword MAP_EMBELLISHMENT_MODE.
gmt [ERROR]: Syntax error: MAP_FRAME_AXES given illegal value (auto)!
gmt [ERROR]: Syntax error: MAP_FRAME_WIDTH given illegal value (autc)!
gmt [ERROR]: Pen attributes not using just - and . for dashes and dots. Offending character --> :

$ gmt coast -ECA+p0.225,black -R-122.913612/36.214964/-9.977657/62.430681+r -JL-95/0/49/77/15c -Bpa10f2g10 -BWSEN -Da -png lixo
gmt [ERROR]: Syntax error: Unrecognized keyword FONT_SUBTITLE.
gmt [ERROR]: Syntax error: GMT_VERBOSE given illegal value (warning)!
gmt [ERROR]: Syntax error: Unrecognized keyword MAP_EMBELLISHMENT_MODE.
gmt [ERROR]: Syntax error: MAP_FRAME_AXES given illegal value (auto)!
gmt [ERROR]: Syntax error: MAP_FRAME_WIDTH given illegal value (autc)!
gmt [ERROR]: Pen attributes not using just - and . for dashes and dots. Offending character --> :

[joa-quim]

Hmm, this makes think you are using an old GMT version

gmt [ERROR]: Syntax error: Unrecognized keyword MAP_EMBELLISHMENT_MODE.

what does this print?

gmt --version

Anyway the Julia wrapper install its own GMT version and that should be GMT6.5.0. The issue you reported roots on GMT itself and I fixed a couple days ago on the master (GMT's, not GMT.jl) version and tried to rebuild the GMT_jl package but the Yggdrasil built is now failing on MacOS and I have no idea why.

[liming-he]

Thanks for helping, - involving to fix problems upstream. I will test it using new version.

name = "GMT"
uuid = "5752ebe1-31b9-557e-87aa-f909b540aa54"
authors = ["Joaquim Luis jluis@ualg.pt"]
version = "1.6.0"

Binary program via Ubuntu 20.04:
GMT 6.0 (in 2019)

[joa-quim]

No, I'm afraid it's not that simple. You cannot test it with neither of the above. Options are:

1- A new GMT_jll build, but that needs understanding/getting support to know why now BinaryBuild fails on Macs

2- Or build GMT yourself from master and follow instructions in https://github.com/GenericMappingTools/GMT.jl/blob/master/src/GMT.jl#L31

[liming-he]

I am able to see the difference between these two lines with gmt v6.4 in another machine .

gmt coast -ECA+p0.225,black -R-122.913612/36.214964/-9.977657/62.430681+r -JL-95/0/49/77/15c -Bpa10f2g10 -BWSEN -Da -png lixo

gmt coast -ECA+p0.225,black -R-122.913612/36.214964/-9.977657/62.430681+r -J+proj=lcc+lat_0=0+lon_0=-95+lat_1=49+lat_2=77+width=15c/0 -Bpa10f2g10 -BWSEN -Da -png lixo

Before a bug can be fixed, is there a way to pass "-JL-95/0/49/77/15c" into grdimage (coast) in gmt.jl? For example: adding proj="L-95/0/49/77/15c" into coast:

grdimage(x, show=false, 
interp = "n", 
frame=(axes=(:left_bare,:bottom_bare,:right_bare,:top_bare),  annot=true, ),
coast=(proj="L-95/0/49/77/15c", DCW=:CA,  frame=(axes="WSEN", grid=10, annot=10, ticks=2),show=false),
colorbar=(pos=(anchor=:BC,length=(10.5,0.5), horizontal=true, offset=(0,.8), ), color=topo, show=true, frame=(ylabel=Symbol(y_label), ticks = :auto, annot = :auto) ), 
par=(MAP_TITLE_OFFSET = 0.8, ), fmt=:png,
savefig = file_out, dpi=600);

However, this did not fix the problem (or see difference except the errors are gone).

julia> x
title: Grid imported via GDAL
Gridline node registration used
x_min: -2.595e6 x_max :3.095e6 x_inc :10000.0 n_columns :570
y_min: 5.705e6 y_max :1.0495e7 y_inc :10000.0 n_rows :480
z_min: 4.999104976654053 z_max :9999.0
PROJ: +proj=lcc +lat_0=0 +lon_0=-95 +lat_1=49 +lat_2=77 +x_0=0 +y_0=0 +datum=NAD83 +units=m +no_defs
480×570 Matrix{Float32}:

should I also change the property of x (x.PROJ)? Is there a way to modify the value of x.PROJ in julia, manually?

Thanks.

[joa-quim]

If you update your GMT.jl all of these problems should be solved now.

gmt coast -ECA+p0.225,black -R-122.913612/36.214964/-9.977657/62.430681+r -JL-95/0/49/77/15c -Bpa10f2g10 -BWSEN -Da -png lixo
.
gmt coast -ECA+p0.225,black -R-122.913612/36.214964/-9.977657/62.430681+r -J+proj=lcc+lat_0=0+lon_0=-95+lat_1=49+lat_2=77+width=15c/0 -Bpa10f2g10 -BWSEN -Da -png lixo

These two now give me the same plot with the foxed GMT (C), the same you get when you update GMT.jl

Before a bug can be fixed, is there a way to pass "-JL-95/0/49/77/15c" into grdimage (coast) in gmt.jl? For example: adding proj="L-95/0/49/77/15c" into coast:

Yes but do not add the figure size. That is, this works proj="L-95/0/49/77"

grdimage(x, show=false, 

No need to use show=false. That is the default for all modules except viz|imshow

should I also change the property of x (x.PROJ)? Is there a way to modify the value of x.PROJ in julia, manually?

No, but you can if you want. x is a struct with the fields described in the GMTgrid docs, so you can do
x.proj4 = "+proj=longlat" or just skip the "+proj=" part that should be added automatically when needed (but not 100% sure it happens all the times that is should)