This paper develops a novel framework for analysing dynamics of vibrating structures using spatial maps of frequency-domain features, represented with Gaussian processes. A signal model is used to transform dynamic features of vibration from the time- to frequency domain. Frequency-domain characteristics of vibrations are then represented continuously over space at separate harmonics using Gaussian process regression to form a spatial map. To demonstrate the utility of this method, we provide an example using data from single frequency excitation of a composite plate, in both healthy and damaged conditions. The results show that different structural conditions of composite plates can be described and analysed in the frequency-domain, motivating the development of future structural health monitoring schemes based on the signal model/Gaussian process framework presented here.