Scatter plot data different resolution satellite images

I have two satellite images with different resolutions and I would like to do a scatter plot comparing pixels

for the same latitude and longitude. I have both images in netcdf format, with their correspondent latitude, longitude and band information (in this case Reflectance at 859 nm). Any ideas on how I could do this in matlab? I have inserted an image showing what I want to obtain (Vanhellemont & Ruddick 2015).

[Walter Roberson]

Is the scale large enough that curvature of the Earth must be worried about, or can it be treated as flat rectangular?

[SnovG]

Yes, it can be treated as rectangular. Is not large.

Answer 1

[Chad Greene]

If both images are the same size, and each pixel in image A corresponds to the same lat/lon of the same pixel in image B, You can get a scatter plot by

plot(A(:),B(:),'b.')

Does the color in your example image correspond to density of scattered points? If so, something like this might help.

Here's an example with some random data:

% Create lat/lon grid: 
[lon,lat] = meshgrid(-180:180,-90:90); 

% Create two sample data images: 
A = 50*cosd(lat) + 5*sind(lon); 
B = A+30*randn(size(A))./(A.^1.2);
A = A/2e3; 
B = B/2e3; 
B(B<0) = NaN;  

% Set ocean data to NaN: 
ocean = ~landmask(lat,lon); 
A(ocean) = NaN; 
B(ocean) = NaN; 

figure
subplot(211) 
pcolor(lon,lat,A); 
colormap(brewermap(256,'*BrBg'))
shading flat
axis tight
borders('countries','color',rgb('dark gray'),'nomap')
title 'Image A'

subplot(212)
pcolor(lon,lat,B); 
shading flat
axis tight
borders('countries','color',rgb('dark gray'),'nomap')
title 'Image B'

And to compare the two images:

figure
plot(A(:),B(:),'b.')
axis([0 0.03 0 0.03])
hold on
polyplot(A(:),B(:),'k-')
xlabel 'Rrs 859 nm (MODIS-Terra)'
ylabel 'Rrs 865 nm (OLI)'

Above, landmask, brewermap, borders, rgb, and polyplot are all available on the File Exchange site.

[SnovG]

Thank you for your answer. The only thing is that since they have diferent resolutions, I think I need to degrade the MODIS data to have both grids overlap. Or I wonder if I should coarse the grid of OLI 8, o make the MODIS grid finer, to have the same amount of pixels to compare.

[Chad Greene]

Ah, you mean horizontal spatial resolution. I'd use interp2 to downsample the higher resolution image. But note that simply using interp2 may result in slightly aliased data from high-spatial-frequency information, so you may want to do a low-pass filter before interp2. If you have the image processing toolbox, imresize is another option.