Development and application of ergodicity model with FRCM and FLAR for hydrological process (Q1045274)

The use of power-weighted Markov chain techniques to investigate the prediction of summer monsoon intensity has lead to some researchers to conclude the ergodicity of the monsoon intensity. However the ergodicity of Markov chain where limited states are converted into each other is not identical to the usual ergodicity of time series. In general, the ergodicity of time series refers to the ergodicity of a stationary process, which means that the process average over time is identical to the average over process states. In this paper a new ergodic property property analysis model of hydrological process is proposed based on fuzzy-rough c-means clustering, autocorrelogram, and fuzzy least absolute regression. A precipitation time series (1951--2004) from Shanghay Hydrology Station is then analyzed with the model. The results show that the precipitation time series of April, May, June and September has ergodic property. We conclude that in the long run, the precipitation of these months will not keep decreasing and it will converge to its mean value in some period.