Sensitivities of an intense Mediterranean cyclone: Analysis and validation

Abstract

AbstractOn 10 and 11 November 2001 a deep cyclone moved northward across the western Mediterranean. Severe floods affected Algeria on 10 November and a mesoscale‐sized region of strong damaging winds occurred over the Balearics and eastern Spain during the first hours of 11 November. These large intense cyclones, originating over north Africa and moving northward, are occasionally observed in the region. Numerical simulations of these types of events are potentially hampered by the lack of observations over the Mediterranean Sea, north Africa and the Atlantic Ocean. To evaluate more accurately the regions in which the model simulations are influenced by this lack of data, the MM5 adjoint system is used to determine the most sensitive areas within the initial conditions of the simulation of this 10–11 November event. Limitations of available adjoint models, such as their linear character, suggest that a test of the applicability of MM5 to the case under analysis is needed. In this study, the evaluation is performed by means of the tangent linear model and, despite finding that the adjoint has an acceptable accuracy, important nonlinear effects are found and attributed to the moist processes. The study tracks backward in time the sensitivities shown at different simulation times using parameters chosen to characterize the cyclone's intensity at 0000 UTC 11 November. Results reveal that the areas that show the largest sensitivities are located over north Africa for the 12 h and 24 h simulations, whereas south‐western and western Europe emerge as areas with important sensitivities for the longer 36 h and 48 h simulations. Subsynoptic details regarding the shape and intensity of an upper‐level trough, as well as a low‐level cold front, are highlighted by the adjoint runs as the structures which influence most strongly the baroclinic development of the intense Mediterranean cyclone and the damaging surface winds it produces. The usefulness of the sensitivity fields in the nonlinear simulations is confirmed by perturbing the control model's initial conditions using the adjoint results as guidance. This analysis helps to quantify the limitations of the linear estimation when applied to the full nonlinear model, focusing on the nonlinearities introduced by the moist physics parametrizations. Copyright © 2004 Royal Meteorological Society