Adjoint-derived forecast sensitivity study of hurricane track and extratropical transition

Factors controlling tropical cyclone (TC) track and intensity change 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 track errors due to poor initial specification of the cyclone itself versus errors in the environment. Additionally, extratropically transitioning cyclones have been shown to have a large effect on hemispheric weather due to the excitation of Rossby wave energy that propagates rapidly downstream, yet numerical prediction of transition is often poor for reasons that remain obscure.

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 and the dynamics of extratropical transition (ET) to the model forecast state. In this presentation, the results of an adjoint-derived forecast sensitivity study will be used to diagnose the sensitivity to initial conditions (and model forecast state) of both TC track and of the structure and intensity of a subsequent ET. Different response functions will be chosen and moist adjoints of both NOGAPS and MM5 will be examined. Cases examined include Hurricane Floyd (1999) and Hurricane Katrina (2005).