Extract XData and YData from contourm hggroup handle

Hello,

I would greatly appreciate some assistance to extract the XData and YData arrays, from the hgroup type handle of the contourm function:

[SC,h]=contourm(LatGrid,LonGrid,Cgrid,V,'Fill','on');

I require the XData and YData arrays as inputs to the following polybool function:

[xtrim, ytrim] = polybool('&', XData, YData, STlong, STlat);

thanks,

Mark

Answer 1

[Kelly Kearney]

Do you want to extract the latitude and longitude values, or the projected X/Y coordinates?

If the former, the coordinates are in the SC matrix in your example. Its format is a bit obscure, so I use a slightly-modified version of this contourcs function to extract it.

Below is an example that I wrote a while ago that I think does the exact thing you're attempting (masking a contour plot to show within a polygon only).

If you really want x/y coordinates, then you'll need to add

for ii = 1:length(Cout)
  [Cout(ii).X2, Cout(ii).Y2] = mfwdtran(Cout(ii).Y, Cout(ii).X);
end

after you extract the contours.

------------

The full example:

    % Load coastlines (replace with your polygon)

    latlim = [23 50];
    lonlim = [-127 -65];

    Usa = shaperead('landareas', 'usegeo', true, 'bounding', [lonlim' latlim']);
    [latusa, lonusa] = maptrimp(Usa(1).Lat, Usa(1).Lon, latlim, lonlim);

    % Create contours (replace with your gridded data)

    n = 50;
    xdata = linspace(lonlim(1), lonlim(2), n);
    ydata = linspace(latlim(1), latlim(2), n);
    zdata = peaks(n);

    figure;
    usamap('conus');
    plotm(latusa, lonusa, 'k');
    [C,h] = contourm(ydata, xdata, zdata);

    % contourcs-stolen code

    K = 0;
    n0 = 1;
    while n0<=size(C,2)
       K = K + 1;
       n0 = n0 + C(2,n0) + 1;
    end

    el = cell(K,1);
    Cout = struct('Level',el,'Length',el,'X',el,'Y',el);

    n0 = 1;
    for k = 1:K
       Cout(k).Level = C(1,n0);
       idx = (n0+1):(n0+C(2,n0));
       Cout(k).Length = C(2,n0);
       Cout(k).X = C(1,idx);
       Cout(k).Y = C(2,idx);
       n0 = idx(end) + 1; % next starting index
    end

    % Trim to coastlines

    [xc, yc] = deal(cell(length(Cout),1));
    for ii = 1:length(Cout)
        [xc{ii}, yc{ii}] = polybool('&', Cout(ii).X, Cout(ii).Y, lonusa, latusa);
    end

     % Plot

    figure;
    usamap('conus');
    plotm(latusa, lonusa, 'k');
    cellfun(@(x,y) plotm(y,x,'r'), xc, yc);

[Mark Cejas]

Excellent code Kelly and thank you so much for your assistance.

I had to slightly modify it to work with my unprojected data, but it worked. However, there is one problem. I need the contour layer to be filled. Is there away to retain the filled contour information in the polybool clipping process?

thanks you so much,

Mark

[Kelly Kearney]

Hmm, may have to experiment on that one...

The following example uses a slightly different approach so we can preserve the color data in the contours of a contourm plot. It uses the x/ydata (projected) instead of the data in the C matrix.

 % Load coastlines (replace with your polygon)

 latlim = [23 50];
 lonlim = [-127 -65];

 Usa = shaperead('landareas', 'usegeo', true, ...
                 'bounding', [lonlim' latlim']);
 [latusa, lonusa] = maptrimp(Usa(1).Lat, Usa(1).Lon, latlim, lonlim);
 [latusa, lonusa] = interpm(latusa, lonusa, 0.5); % Fixing north border

 % Create contours (replace with your gridded data)

 n = 50;
 xdata = linspace(lonlim(1), lonlim(2), n);
 ydata = linspace(latlim(1), latlim(2), n);
 zdata = peaks(n);

 figure;
 usamap('conus');
 plotm(latusa, lonusa, 'k');
 [C,h] = contourm(ydata, xdata, zdata);

 hpatch = get(h, 'children');
 [xusa, yusa] = mfwdtran(latusa, lonusa);

 for ii = 1:length(hpatch)
     xp = get(hpatch(ii), 'xdata');
     yp = get(hpatch(ii), 'ydata');
     zp = get(hpatch(ii), 'zdata');
     [xp, yp] = poly2cw(xp, yp);
     [xnew, ynew] = polybool('&', xp, yp, xusa, yusa);

     set(hpatch(ii), 'xdata', xnew, ...
                     'ydata', ynew, ...
                     'zdata', ones(size(xnew))*zp(1));
 end

Getting this to work with contourfm is trickier, since it would involve breaking some of the patches into multiple ones (any NaN-delimited patch that represents two separate polygons will result in color spilling out, because, well, patches in a map axis are just horribly inconvenient like that...). Do you need filled contours?

[Mark Cejas]

Great work Kelly. Although I am having trouble integrating this into my data.

Would you be able to look it over and provide some guidance?

If so, I can send you the data by email, since this information cannot be shown over the inet.

thanks a ton,

Mark

[Mark Cejas]

Yes, my goal is to visualize the surface as filled contours.

thanks,

[Kelly Kearney]

No promises, but if you contact me by email I can see if I have time in the next few days (not going to put my email on this forum, but you can contact me via my File Exchange author page).

[Mark Cejas]

thanks for the consideration Kelly.

[Mark Cejas]

Hi Kelly. Better that I send it this way. thanks again for your consideration.

Answer 2

[per isakson]

This approach should do it (despite I don't know neither contourm nor polybool )

• find the handles of the line-objects
• use get( line_handle, 'Ydata' );
• etc

Hint:

    >> plot( magic(5) )
    >> fh = gcf;
    >> lh = findobj( fh, 'Type', 'Line' );
    >> y_data = get( lh(1), 'Ydata' );
    >> y_data = cat( 1, y_data, get( lh(2), 'Ydata' ) );
    >> y_data = cat( 1, y_data, get( lh(3), 'Ydata' ) );
    >> y_data
    y_data =
        15    16    22     3     9
         8    14    20    21     2
         1     7    13    19    25