The ensemble transform technique is used to attempt to distinguish between supplemental observation sites that will primarily influence hurricane track forecasts from those that will primarily influence hurricane intensity forecasts. The ensemble transform technique uses linear combinations of ensemble perturbations to simulate the effect of observations on forecast uncertainty. It identifies optimal supplementary observation sites by comparing the effects on forecast uncertainty of supplemental observations in each observation site that is reachable by a dropsonde equipped aircraft. The accuracy with which it can identify optimal sites depends on the number of ensemble perturbations, the realism of the numerical simulations on which the ensemble is based and the level of sophistication of the scheme which is used to assimilate the supplemental observations. Whether the technique identifies observation sites that will improve track or intensity forecasts depends on how one chooses to measure forecast uncertainty. The ensemble transform technique allows virtually any measure of forecast accuracy to be used so that, for example, observation sites that minimize the uncertainty in forecasts of the coordinates of the hurricane center can be identified. Alternatively, observation sites that minimize the uncertainty in the minimum pressure of a hurricane can be identified. By comparing the differences in the positions of optimal sites for track and intensity one can use the technique to explore and test theories of hurricane intensification and movement. The talk will outline our methodology and illustrate it with some preliminary examples from ensembles based on the GFDL, VICBAR, NCEP and ECMWF ensembles