How can I calculate ensemble average of 6 2D matrices with nanmean function?

Dear All,

I have 6 2D matrices. Both of them have 10958 rows and 30471 columns. The rows are the time and the columns are the space. I would like to calculate the ensemble time average of these martices using with nanmean function. I tried the following way:

pr_h1=repmat(NaN, 10958);
pr_h=repmat(NaN,size(pr_loci_MPI_ESM_LR_CLM));
>> for j=1:10958
for  i=1:30471
pr_h1(j)=[pr_loci_MPI_ESM_LR_CLM(j,i), pr_loci_CM5A(j,i), pr_loci_CNRM_ALADIN(j,i), pr_CNRM_CCLM(j,i), pr_loci_CNRM_CM5_RCA4(j,i), prM_loci_EC_EARTH_RACMO(j,i)];
pr_h(j,:)=nanmean(pr_h1(j));
end
end

but I got the error:

In an assignment  A(:) = B, the number of elements in A and B must be the same.

I would appreciate if someone helped me!

Answer 1

[Star Strider]

I can’t follow your code.

See if this does what you want:

M1 = randi(9,10, 15);
M2 = randi(9,10, 15);
M3 = randi(9,10, 15);
M4 = randi(9,10, 15);
M5 = randi(9,10, 15);
M6 = randi(9,10, 15);

MC = cat(3, M1, M2, M3, M4, M5, M6);

EnsAvg = nanmean(MC, 3);

[Szabó-Takács Beáta]

I tried this way but I got the message:

Out of memory. Type HELP MEMORY for your options.

Error in nanmean (line 25)
x(i) = zeros(size(i));

Can you suggest me a solution with for loop?

[Star Strider]

Yes, but it’s not pretty:

Mlist = {'M1','M2','M3','M4','M5','M6'};                            % Matrix Names Cell Array
EnsAvg = zeros(size(M1));
for k1 = 1:6
    Mcat = cat(3, EnsAvg, eval(Mlist{k1}));
    EnsAvg = nansum(Mcat,3);
end
EnsAvg = EnsAvg/6;

This avoids creating the (10958 x 30471 x 6) concatenated matrix at the expense of using eval, that not everyone here likes. (It has its uses, though, if it’s not abused.)

I checked with my test matrices and my previous code, and the ‘EnsAvg’ results are identical.

Note — This only works with the mean. It would have to be modified significantly to calculate the variance, and I’m not certain that would even be possible in the presence of NaN values.