Predictability of 3D Isotropic Turbulence —Effect of Data Assimilation—
In numerical simulation of turbulent flows, it is usually difficult to obtain exact initial conditions and it is well known that initial uncertainty in high wavenum-bers spreads toward large scale through nonlinear dynamics. These imply the limitation of the numerical prediction. We present here a series of numerical experiments to study the effect of data assimilation on the predictability of isotopic 3D turbulence. In the experiment we have two parameters K c and T to control the data assimilation, where K c is the maximum of the assimilated wavenumber range and T is the time interval of the assimilation. It is found not only that we can suppress the growth of uncertainty by data assimilation but also that we can reduce it under appropriate conditions on K c and T. It is also found that there exists a critical K c * such that if K c < K c * the uncertainty grows for any T.
KeywordsData Assimilation Difference Spectrum Numerical Weather Prediction Initial Uncertainty Large Wavenumbers
Unable to display preview. Download preview PDF.