The potential value of targeted observations collected by mobile platforms such as UAS for numerical weather prediction of storm-scale phenomena is explored using ensemble sensitivity analysis (ESA). ESA has been demonstrated for use in guiding targeted observations of synoptic scale and mesoscale phenomena but this is the first time that it has been applied on the storm-scale. An idealized supercell is used in this examination with a horizontally homogeneous initial state. The strongest correlations at 20, 40, and 60 minute lead times were found outside the core of the supercell, though most of the domain showed a statistically significant relationship between the perturbations and forecast responses. The results show that, for this case, greater value would likely be obtained at those lead times by targeting observations that are representative of the storm environment rather than within the actual storm. The storm environment and the impacts of the storm on the far field appear to be the most important predictors of future storm strength for this supercell case.