This paper proposes a simple method for scale- and flow-dependent calibration of ensemble spread to account for excessive damping in a numerical weather prediction model. It is hypothesized that relationships between spatial variance and error growth can be applied to individual forecast periods. The relationships are transformed to spectral space and a simple example is used for concept demonstration. Applicability to individual forecasts is then tested on six independent cases by introducing numerical damping to the Weather Research and Forecasting (WRF) model, the reference model, to create an imperfect model. The results show that error growth estimated by the ensemble of imperfect forecasts can be calibrated to agree with the reference model. The empirical, scale-dependent, correction factor is a function of the two model forecasts and the flow of the day. Finally, the limitations of the calibration are demonstrated by comparing against a third model with very different error properties, and it is argued that the calibration provides a measure of the effect of those differences on ensemble spread. The calibration approach has potential application to ensemble forecasting systems, estimates of predictability limits, model-error diagnosis, and modern data assimilation systems.