Figure 15. Prediction of MIK east–west motions using GSA east–west as input motion. Motions recorded at GSA are convolved with an impulse response at a particular frequency. AMBIENT_SDOF uses the fundamental frequency as determined from FFTs of pre-event data, FORCED_SDOF uses the fundamental frequency as determined from forced vibration tests. (a) M 2.0 18 June 2003 Pasadena earthquake ( = 5 km). A nonyarying second-mode SDOF response is included for both ambient and forced models to include the high-frequency components of the motions. During this very small amplitude motion, the natural frequency determined using the ambient data models the observations at MIK better than the forced vibration result. (b) M 5.4 22 February 2003 Big Bear earthquake ( = 119 km). Neither the AMBIENT_SDOF or FORCED_SDOF response model the MIK observation well. Another model, FFT_SDOF, with natural frequency at 1.06 Hz, as determined by the FFT peak of the strong-motion record is included. This leads to a somewhat improved model of the observed motion. Clearly, even during the moderate shaking generated from this relatively small and distant earthquake, a linear building model is inappropriate. A model that can have evolving natural frequencies will best represent the observed motions.