Interpolation of missing data using nonlinear and chaotic system analysis

A new method for interpolating missing data in a time series is introduced. The method uses local flow approximation for interpolation and is based on a suitable modification of some recent developments in predicting nonlinear and chaotic time series. The method has been tested for various geophysical and climatic time series. We find that in many cases the method gives better results than the cubic spline. The real power of this method is seen when there are several consecutive missing values in the series.