An interpolation method combining Snurbs with window interpolation adjustment

The presence of missing value in time series increase the difficulty of data analysis. The traditional interpolation method is mainly aim at simple numeric data missing, which without considering the correlation among data points of actual time series, so it is easy to distort the hidden relationship among data points. Therefore, based on the Snurbs (series-Non-Uniform Rational B-Spline) interpolation method, we present a Snurbs interpolation method, which further combines window interpolation adjustment based on the time series data correlation. The new algorithm was applied to two real world hydrological time series, which represent the periodic and aperiodic time series respectively. The experiment results showed that our method is better than other traditional methods, such as Linear interpolation, Pchip interpolation and Cubic Spline interpolation. We can always make interpolation curve better approximate to actual time series curve by i) selecting suitable weight for data points, ii) and further doing window interpolation adjustment.