BRED VECTOR APPLIED TO THE ATMOSPHERIC DYNAMICS
Luis Fernando Salgueiro Romero1; Sandra Sandri1; Haroldo Campos Velho1; Rosângela Cintra1; Saulo Freitas2
1 INPE; 2 CPTEC-INPE
Resumo
One remarkable conquer of science was the numerical weather prediction. However, some dynamics processes are not well represented in the simulation, generating a degradation of the forecasting. In addition, a chaotic atmospheric dynamics is also noted, with strong dependency of the initial condition. The initial condition is never known in a complete way. Indeed, the procedure to determine the initial condition in the operational prediction centers is called data assimilation, where the background fields (forecasted atmospheric state) are combined with observations for producing the analysis - the new initial condition. The bred vector approach can be employed to investigate the sensitivity of a certain process in the atmospheric dynamics simulation, detecting the fastest-growing instabilities. The latter scheme is also applied to evaluate the goodness of the predicted dynamics, i.e., the breeding process is employed to evaluate the predictability. The bred vector is computed as the difference between reference and perturbed simulations for the dynamical system after a time period of integration. After the mentioned time integration period, the bred vector is scaled at the same size as the initial perturbation, and a new perturbed initial condition is added to restart the process.
Palavras-chave: Climate Dynamics; Numerical Simulation; Atmospheric dynamics; Global circulation model; Bred Vector