Welch's Overlapped Method

I am measuring wind speed using a number of different anemometers. These anemometers are placed on a structure which exhibits sinusoidal motion. Thus, the signal of the anemometers contains both the wind speed as well as the sine wave to which they were exposed.

I wish to accurately determine the amplitude of the sine wave from the anemometer signal. Thus, I decided to use a flat-top window when calculating the FFT. However, since the signal is 'noisy' the resulting amplitude of the sine wave is not correct. Sometimes, the amplitude given on the spectrum is greater than the actual amplitude of the sine wave!!

I am considering using Welch's overlapped method (pwelch in MATLAB). Can someone please tell me whether such a method is suitable for my application? I have read that this method should only be used with stationary signals? I am confused whether my signal is considered as stationary. Can someone please help me? Thanks.

• I wouldn't say your signal is stationary. However, you are interested in a sine wave interference which is stationary. The Welch method will yield a better interference to noise ratio. If your interference is strong enough then you wouldn't gain much. – ThP Sep 23 '14 at 19:56
• @ThP: Thanks for your reply. When using pwelch in MATLAB I can see a peak at the frequency of my sine wave. Considering my situation, do you think a psd spectrum makes sense or should I just calculate fft? Thanks in advance. – user10881 Sep 23 '14 at 21:30