Further question about using pcolor

Dear MATLAB wizard,

I have read and used the answers that you gave to Alan B's question of a few days ago. I, too, am attempting to put concentrations on a map. I have uploaded my data from Excel, and split the latitude (LatDIC), longitude (LongDIC) and concentration (DIC) data into separate matrices that are one column of data each. I keep getting an error because pcolor does not seem to like that my data is not in a wide matrix.

figure(1); pcolor(LongDIC,LatDIC,DIC), shading flat, colorbar, title('DIC');

I have been trying to average the data using some code sent to me by someone else, but I am having problems because I don't know what their data set looked like to begin with. I have copied their code below.

% calculate out averages for each latitudinal band
    % discard any cells where any data is not available
            if isfinite(gd_co3a(ii,jj,1))
                atl_co3_gd(ilat) = atl_co3_gd(ilat) + gd_co3a(ii,jj,1);                         
                c_atl_co3_gd(ilat) = c_atl_co3_gd(ilat) + 1.0;
            end;
            if isfinite(pred_co3(ii,jj,1))
                atl_co3_p1(ilat) = atl_co3_p1(ilat) + pred_co3(ii,jj,1);
                c_atl_co3_p1(ilat) = c_atl_co3_p1(ilat) + 1.0;
            end;
            if isfinite(pred_co3_2(ii,jj,1))
                atl_co3_p2(ilat) = atl_co3_p2(ilat) + pred_co3_2(ii,jj,1);
                c_atl_co3_p2(ilat) = c_atl_co3_p2(ilat) + 1.0;
            end;
            if isfinite(woa_sal(ii,jj,1))
                atl_sal(ilat) = atl_sal(ilat) + woa_sal(ii,jj,1);
                c_atl_sal(ilat) = c_atl_sal(ilat) + 1.0;
            end;

or this code:

% calculate averages for each latitudinal band
  if (c_co3_gd(ilat) > 5)
      co3_gd(ilat) = co3_gd(ilat) / c_co3_gd(ilat);
  else
       co3_gd(ilat) = NaN;
  end;
  if (c_co3_pred1(ilat) > 5)
      co3_pred1(ilat) = co3_pred1(ilat) / c_co3_pred1(ilat);
  else
       co3_pred1(ilat) = NaN;
  end;

I think that I might need to initialise the original matrices to zeros(181,361) with the data in the first row.

Am I on the right track? Any suggestions would be appreciated.

Many thanks!

Answer 1

[Kelly Kearney]

How is your data distributed, geographically? Is it located on a grid? If so, you just need to rearrange the data into the appropriate grid before feeding it to pcolor.

If the data is scattered, you can either plot discrete points or interpolate onto a grid and plot via pcolor. A few helpful functions

reshape

vec2grid

scatter

TriScatteredInterp

griddata

[Jen]

Thanks, Kelly. You are right--they are not on a grid. I have tried to grid the data using griddata:

DICLat=data(:,1);

DICLong=data(:,2);

DIC=zeros(83);

DIC=data(:,13);

XIDIC=[-3:0.5:10];

YIDIC=[49:0.5:62];

YIDIC2(:,1)=YIDIC(1,:);

ZIDIC=griddata(DICLat,DICLong,DIC,XIDIC,YIDIC2);

but it comes up with a matrix that is entirely NaN! Oh no! There is no error message. I was planning then to do:

figure(1); pcolor(XIDIC,YIDIC,ZIDIC), shading flat, colorbar, title('DIC');

Any thoughts on why my ZIDIC matrix is NaN?

Thanks again for your help!

[Jen]

Ah..I needed to use meshgrid. I have sorted it out now and got both pcolor and contour maps. Many thanks for your help! This forum is an invaluable resource.

Answer 2

[Fangjun Jiang]

To visualize a series of (x,y,z) data, I made the following example. The first part is just to come up with some data. The second part is the visualization.

Note that the pcolor() in figure 1 has 9x9 patch of colors. The 2-D view of the scatter3() in figure(3) has 10x10 squares of colors. So ignore the first row and last column of figure(3) you will get a 100% match of colors. Adjust the size of the square to achieve the desired results.

close all;
N=10;z=rand(N);
x=repmat(1:N,N,1);
y=repmat((1:N)',1,N);
X=x(:);Y=y(:);Z=z(:);

CircleSize=1500;
figure;pcolor(z);
figure;h1=stem3(X,Y,Z,'filled');
figure;h2=scatter3(X,Y,Z,CircleSize,Z,'s','filled');
view(2);