The concept of developing a measure that evaluates the degree of system observability for nonlinear power systems modeled as DAEs has been introduced by the authors in the past with an emphasis on the dependence of this measure to the available measurement sets and changes in the inherent differentiation indices. This work extends this analysis by investigating the sensitivity of the observability measure to system initial conditions and dynamic parameter variations. In particular, the sensitivity of the measure to the generator parameters (inertia and damping) and variations in the system load trajectories (PV-curves) is investigated. The presented results indicate that the degree of observability is robust to small deviations of the system initial conditions, and dependent on the generator damping and inertia coefficients.