How to vectorise or speedup the code

Hello. I have Lat(2689x1) Lon(2689x1) arrays that coordinaines coordinates of meteostations, and lat (1200x200) lon (1200x200) arrays with coordinates of satellite's measurements M(1200x200). I need to find elements of satelletes measurements, that located closer then 0.25 deg to meteostations. What i've did:

k=1;
for m=1:length(Lat)
	x=lat-Lat(m);
	y=lon-Lon(m);
	[a,b]=find(abs((x))<=0.25 & abs((y))<=0.25);
        for n=1:length(a)
            Mes(k,n)=M(a(n),b(n))
        end
        St_N(k)=m;
        k=k+1;
end;

Because of many loops it works very long. Is there any way to vectorise or speedup this code?

Answer 1

[darova]

Try pdist2

D = pdist2([Lat(:) Lon(:)],[lat(:) lon(:)]);        % create every possible combinations of distances
[i,j] = find(D<=0.25);                              % find every distance satisfies condition
% i - station of (Lat,Lon)
% j - station of (lat,lon)
[i1,j1] = ind2sub(j,size(lat));                     % return to 2D matrix
Mes = M(i1,j1);                                     % stations that close enough

[Eugene Pashinov]

Thanks, but i think it does not work, because:

min(abs((Lat-lat(i1(1),j1(1))))+abs((Lon-lon(i1(1),j1(1)))));

should be less then 0.25, but it is 19

[darova]

hm. Can you attach the data?

[Eugene Pashinov]

found a mistake, you mixed up places of j and size(lat), correct is:

[i1,j1] = ind2sub(size(lat),j);

Thank you very much for help!

[darova]

Try without this line:

% [i1,j1] = ind2sub(j,size(lat));                     % return to 2D matrix

only

min(abs(Lat-lat(j(1)))+abs(Lon-lon(j(1))))

[darova]

Don't use ind2sub. It's not correct in this case

% [i1,j1] = ind2sub(j,size(lat));               % return to 2D matrix

Just one index

Mes = M(j);                                     % stations that close enough