How to plot regression coefficients on a map?

I have attached a zipped netcdf file ('ssh.nc'). I wanted to do a corellation test between variables, 'zos' and 'bottomT' which have the following dimensions of longitude,latitude,time. After this I wanted to plot the correlation coefficients on a map.

I tried the following code but get an error about the rows. I find it weird since the dimensions are the same.

>> lat = ncread('ssh.nc','latitude');
>> lon = ncread('ssh.nc','longitude');
>> ssh = ncread('ssh.nc','zos');
>> t = ncread('ssh.nc','bottomT');
>> nx=length(lon);
>> ny=length(lat);
>> for i=1:nx
for j=1:ny
[r,s] = corr(t,ssh(i,j,:));
rxy(i,j)=r;
sxy(i,j)=s;
end
end

I get the error:

Error using corr (line 106)

X and Y must have the same number of rows.

I also tried correlate, corr2 but get errors.

Would be grateful to receive help in this, preferably with the correct codes, since I am going wrong somehere. I also want to know what's wrong in the above.

Answer 1

[KSSV]

YOu need to calculate the coeffecenits for 3D matrix.

lat = ncread('ssh.nc','latitude');
lon = ncread('ssh.nc','longitude');
ssh = ncread('ssh.nc','zos');
t = ncread('ssh.nc','bottomT');
nx=length(lon);
ny=length(lat);
rxy = zeros(nx,ny) ; 
sxy = zeros(nx,ny) ;
for i=1:nx
    for j=1:ny
        [r,s] = corr(squeeze(t(i,j,:)),squeeze(ssh(i,j,:)));
        rxy(i,j)=r;
        sxy(i,j)=s;
    end
end

[Keegan Carvalho]

Thanks KSSV, It worked. However, the result is in the form of a film. Probably because it shows the correlations for each time value.

How do I get the R2 value for the entire time as an average (just one value for the entire time range)? So that I can store it as an image.

Regards

[KSSV]

Hey I relaized that..my last answer is irelevant to your question...now I have modified the answer..I guess this works according to your old code.

[KSSV]

rxy and sxy are matrices......what you want exactly? The old code or the one you psoted int he question?

[Keegan Carvalho]

Sorry for the confusion; ignore that. Since I got two matrices rxy and sxy, I used the code to create the map:

pcolor(lon,lat,rxy) ;
shading interp
colorbar
drawnow

Also tried for sxy.

But am a little confused. Which one should I plot? rxy OR sxy?

Because the values are different in both matrices

[KSSV]

Off course rxy. Read the docmumentation. You can simply use corr like below:

for i=1:nx
    for j=1:ny
        r = corr(squeeze(t(i,j,:)),squeeze(ssh(i,j,:)));
        rxy(i,j)=r;
    end
end

[Keegan Carvalho]

Thank you KSSV :)

Appreciate your continued guidance.

[Aristo Pacino]

Hi, @KSSV, I have used the same code for calculating correlation coefficient at grid points using the above code but I am getting NaN values. Please find attached data.

Some specification:

1. Lat =0:10:360
2. Lon = -180:10:180
3. t and ssh are 37-by-37 matrix