Factors controlling tropical cyclone (TC) track and intensity change, as well as transitions of cyclones from tropical to extratropical are poorly understood and represented in models. Some important influences are due to weather systems (including jet streaks, upper troughs, and cyclones) that are inadequately represented, but the most influential yet poorly represented parts of these weather systems, in terms of their affect on TC behavior, remain unquantified. Furthermore, it is difficult to distinguish forecast errors due to poor initial specification of the cyclone itself from errors in the specification of the environment.

Adjoint-derived forecast sensitivity analyses have been widely applied to problems related to extratropical cyclone intensity, but few extant studies have used adjoint output to understand the sensitivity of TC track, the dynamics of extratropical transition (ET), or the dynamics of tropical transition (TT). In this presentation, following the review of adjoint methodology, the results of adjoint-derived forecast sensitivity studies will be used to diagnose the sensitivity to initial conditions (and model forecast state) of the ET of Atlantic Hurricane Floyd (1999) and the formation of Subtropical Storm Andrea (2007). A potentially computationally efficient, adjoint-derived approach to targeted observing of TC track, ET, and TT is proposed.