Recent results (Kuypers 2000, Nuss et al. 2000) suggest that fine resolution mesoscale model forecast error has similar characteristics independent of the random sampling strategy employed in defining the initial atmospheric state. Each simulation showed that fine resolution mesoscale model forecast error was primarily due to subtle errors in the definition of relevant synoptic-scale features in both a winter and summertime case study. Current operational methods which attempt to define the important relevant synoptic-scale features on a given day typically use a tangent linear model to deduce volumes of the atmosphere where the model is most sensitive at the time of initialization given a chosen forecast parameter of interest. This presentation will focus on a simpler approach whereby the important relevant synoptic-scale structure is defined as the locations required mathematically to reasonably define the three dimensional atmospheric structure to within some predefined degree of accuracy. A comparison will be made between the mesoscale model forecast error using random sampling and the simplified targetting approach.