Documentation of eof_nan

Global Index (short | long) | Local contents | Local Index (short | long)

Function Synopsis

[pc,eof_reg,sing]=eof2(input_array,n,nya)

Help text

 function [pc,eof_reg,sing]=eof2(input_array,n,nya)

 Computes the EOF of the input_array by formulating
 the covariance matrix in time. PCs are normalized; 
 EOF's are regression maps upon the normalized PC's; SING
 is the singular values.
 
 INPUTS:
 input_array
 N : number of regression maps to compute
 NYA is the latidue axis of the maps (from top to bottom)

Cross-Reference Information

Listing of function eof_nan

function [pc,eof_reg,sing]=eof2(input_array,n,nya)

[ts,gs]=size(input_array);

cov_matrix=nan*ones(ts,ts);

% lets make a covariance matrix in time 
disp 'making cov matrix'
for i=1:ts
if (i/10)==round(i/10)
 ts-i
end
% only go from i to the end of the matrix
 for j=i:ts
  cov_matrix(i,j)=cosp(input_array(i,:),input_array(j,:),...
                              nya,0);
  cov_matrix(j,i)=cov_matrix(i,j);
 end
end

[left_vec,sing,right_vec]=svd(cov_matrix,0);
sing=diag(sing);

for eof_nu=1:n
 pc(eof_nu,:)=left_vec(:,eof_nu)';
 pc(eof_nu,:)=pc(eof_nu,:)./std2(pc(eof_nu,:));
end

% reg 
for eof_nu=1:n
eof_nu
  tmp=pc(eof_nu,:)';
for grd=1:gs
  eof_reg(eof_nu,grd)=cov5(input_array(:,grd),tmp);
end

end