Given a anomaly time series, for which time series need to be computed by removing annual cycle or 1st harmonic or any particular harmonic. Is the method of computing FFT and making ith & (n+1-i)th coefficients making zero and then iFFT, will give the required result or anything need to be taken care?
t=0:0.2:4*pi+.1; l = 64 ;
X = fft(x);
X2=X; X2(3)=0; X2(64-1)=0;
x2 = ifft(X2);
plot(x,'r'); hold on; plot(x2, 'g');