I am performing a site response analysis (SRA) for a deep, stiff soil site where Vs is about 800 ft/s for hundreds of feet (site Class D), and I am using a non linear effective stress code.  There are many opinions regarding the development and implementation of input ground motions.

I am using ground motions recorded on site class B (since these are generally considered to be un-modified) and scaling/modifying the ground motion so that the spectrum from the recorded ground motion is similar to the Site Class D, MCE 2007 CBC response spectrum at the specific location.

I am then inputting the ground motion at a depth of about 100 feet into the code and assigning the ground motion a Vs of 800 ft/s, similar to the site specific Vs at the site and depth.

Any comments on the methodology are appreciated.  I am aware of the NEHRP commentary and realize my method is different.

Thanks