I have made a test run using fake input data: 210 Hz sine wave starting with 0 phase of amplitude 1.0 where I used different methods of windowing data.
The FFT size was 1800 seconds, consisting of thirty 60 second segments.
All plots are logarithmic.
| Sh[210hz] | Window type | Comment |
| 29.211485 | Tukey 20480 | Hann window of size 2*20480, split in half and applied to the tails |
| 28.430417 | Tukey 40960 | |
| 26.868335 | Tukey 81920 | |
| 22.181985 | Tukey 204800 | |
| 15.347736 | Tukey 384000 | |
| 11.247189 | Hann | Same as Tukey 491520 |
| 15.996011 | Welch | |
| 8.797347 | Papoulis | |
| 22.119478 | Overlapped Hann | added segments shifted by 30 seconds |