The use of the ratio of microtremor spectra, as computed by the Nakamura's technique, was recently proved successful for the evaluating the thickness of sedimentary covers laying over both shallow and deep rocky bedrocks thus enabling bedrock mapping. The experimental success of such application and its experimental uncertainties are today reported in many publications. To map bedrock, two approaches exist. The first is to assume a constant shear wave velocity profile of the sediments. The second, and most preferable, is Ibs-von Seht and Wohlenberg's, based on correlating Nakamura's curves main peak and wells information. In the latter approach, the main sources of uncertainty addressed by authors, despite the lack of formal proof, comprise local deviations of the subsurface from the assumed model. I first discuss the reliability of the simplified constant velocity approach showing its limitations. As a second task, I evaluate the uncertainty of the Ibs-von Seht and Wohlenberg's approach with focus on local subsurface variations. Since the experimental basis is well established, I entirely focus my investigation on numerical simulations to evaluate to what extent local subsurface deviations from the assumed model may affect the outcome of a bedrock mapping survey. Further, the present investigation strategy suggests that modeling and inversion, through the investigation of the parameters space around the reference model, may reveal a very convenient tool when lateral variations are suspected to exist or when the number of available wells is not sufficient to obtain an accurate frequency-depth.