The application of the sensitivity vector fields (SVFs) for tip-sample interactions in the context of multi-mode tapping dynamics of atomic-force microscopes (AFM) is presented. SVFs represent a novel approach to accurately determine simultaneous parameter variations by exploiting the geometric features of observed chaotic dynamics. The paper also demonstrates that, in certain operating conditions, higher modes are essential to correctly predict the AFM dynamics, and they cannot be neglected. The accuracy of the SVF approach is discussed as applied to a multimode AFM model where mode shapes vary due to multiple parameter variations. The identifiability of various parameters based on SVF reconstruction is investigated. Several calibration issues for parameter reconstruction are observed and resolved by a specialized sample filtering and a novel correction factor.