In order to minimise the differences in noise exposure between forecast and control in specific control points, a new forecast model is described, based on historical noise data at the control points. By clustering the noise data of individual flights as function of aircraft type (or category), runway, route and flight procedure, energetically averaged noise levels are obtained. From traffic forecasts the number of flights for the same cluster of parameters can be extracted. By combining the traffic forecast and the averaged noise levels, a consistent noise exposure forecast in Lden is obtained for each control location. First verifications of this concept show a significant reduction of the differences between forecast and enforcement.