The input resistor 909Ohm of AD602 was shorted. I've confirmed that the gain (= attenuation by voltage division) was increased by a factor of 10.
This modification was done for WFS2-I1 and WFS2-Q1. Also the thick film resistors for the WFS2-I1 channel was all replaced with thin film resistors.

Attachment 1 shows the comparison of the noise levels. The curves were all calibrated referred to the response of the original whitening filter configuration.
(i.e. measurement done after the gain change was compensated by the factor of 10.)

Now the AF chain is not limited by the noise in the whitening filter board. (Brown)
In fact, this noise level was completely identical between I1 and Q1. Therefore, I don't think we need this resistor replacement for the whitening filter board.

We can observe the improvement of the overall noise level below 10Hz. (Comparison between green and red/blue)
As the signal level goes up, the noise above 100Hz was also improved.

Now we need to take care of the n x 0.7Hz feature which is in the demod board...