Dashed contours in a .eps figure appear solid

I am producing some figures for a publication and I am looking to create a contour plot with positive values in solid contours, negative values in dashed contours, and then shading for 95% significant values. I have set up the code to do this, but the dashed contours do not appear in the figure on screen or in the saved .eps file.

Here is the code I have used:

figure(1);
colormap(getpmap(4)); %This is a special colormap I have.
contourf(sst.lon,sst.lat,siglev,0.99:0.01:1.01);
contour(sst.lon,sst.lat,siglev,0.99:0.01:1.01);
contour(sst.lon,sst.lat,sst.cCPW(:,:,13).*sst.mask,...
        -0.9:0.2:-0.1,'k--','LineWidth',2);
contour(sst.lon,sst.lat,sst.cCPW(:,:,13).*sst.mask,...
        0.1:0.2:0.9,'k-','LineWidth',2);
print('SST_Correlation.eps','-depsc','-loose','-r1200');

Of course the problem is that the dashed lines appear as solid. I have read that this is an issue with the painters renderer, but when I switch to another renderer (zbuffer or opengl) and try to render a .eps, the file is gigantic in size. I tried rendering as a .jpg and .png but the quality of the image is not that great.

Can anyone help me with getting a .eps file (of reasonable size) with the dashed contours showing correctly?

[Oliver Woodford]

Jason, no need to put "Help" in your question. Every question is asking for help! Also, your code cannot be run by anyone - we don't have getmap, stt or siglev. Can you change it for a simple script that exhibits the same problem and can be run by anyone with a standard MATLAB installation?

Answer 1

[Bjorn Gustavsson]

Ok, I just happened to have to deal with this ordeal. Here is a solution that gets the job done:

% Make the contours (and let matlab draw them):
[c1,h1] = contour(x,y,Z,'k-');
% Take all the info from the contourline output argument:
i0 = 1;
i2 = 1;       
while i0 <  length(cc1)

    i1 = i0+[1:cc1(2,i0)];
    zLevel(i2) = cc1(1,i0);
    hold on
    % And plot it with dashed lines:
    ph(i2) = plot(cc1(1,i1),cc1(2,i1),'k--','linewidth',2); 
    i0 = i1(end)+1;
    i2 = i2+1;

end
% Scrap the contourlines:
delete(h1)

Then continue with personal decorations et al.

HTH.

Answer 2

[Jan]

You can insert NaNs in your data. The corresponding points are not drawn:

a = peaks;
a(rand(size(a)) < 0.1) = NaN;
[c, h] = contour(a);

For complicated data, this might be impossible, because it is hard to find suiting positions for the gaps.