Non-Linear Prediction Algorithm of Satellite Cloud Images

In order to solve the problem of low prediction accuracy of conventional prediction algorithm of satellite cloud images. The spectrum-space reconstruction and phase-space prediction algorithm are proposed, the algorithm takes full account of that the cloud moves, grows and dissipates in the non-linear and nonstationary process, the mathematical principles of the algorithm and the process of numerical realization is given. Numerical experiment results show that the prediction algorithm for satellite cloud images is feasible when the forecast time length is less than two hours, the forecast result is reliable and stable, as the forecast time length increasing, the forecast error rapidly increase, when more than two hours, forecast result is unstable.