Drawing on a 50-member ocean ensemble hindcast, the magnitude and characteristic temporal and spatial scales of intrinsic and forced sea surface height (SSH) variability are evaluated over a 37-year period. The intrinsic and forced contributions derived from the ensemble are found to have similar temporal spectra, but different characteristic spatial scales. These results suggest that, with an appropriate choice of cutoff scales, simple spatial filtering can be used to estimate the forced and intrinsic contributions given either a single model run, or an observational data set. The method is tested using a single member drawn from the ensemble, before being applied to the observed altimetric record. Two sample applications with relevance to large-scale climate are used to illustrate the method's potential utility. Firstly, the long-term trends calculated from the total and recreated forced components using the altimetric record are compared and local differences highlighted. Second, the recreated forced SSH is shown to covary with the North Atlantic Oscillation at seasonal time scales in regions where no such influence can be found using the original SSH signal. Some limitations and uses for which the method may prove unsuitable are also briefly considered.