Rainfall prediction is a very challenging task due to its dependence on many meteorological parameters. Because of the complex nature of rainfall, the uncertainty associated with its predictability continues to be an issue in rainfall forecasting. The Hurst exponent is considered as a measure of persistence and it is believed that if a time series has persistence (as reflected by a Hurst exponent value greater than 0.5) it is also predictable. However, very limited studies have been carried out to demonstrate this hypothesis. This study, through experimental work on hypothetical data as well as real data, demonstrates that the Hurst exponent can be taken as an indicator for predictability provided that the exponent values at “smaller levels” of the time series are also significantly greater than 0.5 together with the Hurst exponent of the overall time series. It is also shown that it is better to predict the “similarity” aspect associated with a time series (and derive the predicted rainfall) than to predict the rainfall directly.

No abstract

Temporal resolution of rainfall series needs to be necessarily less to use it in many engineering applications. But most of the simulated and observed rainfall series are coarser than 3hours. Hence, it is imperative to disaggregate coarser rainfall to finer. The quantum of necessary fineness depends on application in which the rainfall data is going to be used. In this paper, the competency of Artificial Neural Network to disaggregate 3 hour rainfall into hourly, in case of limited data is verified. It is found that the disaggregation is viable with the constraint of limited data also. The rainfall is disaggregated using three models, of which, performance of the second model is much better than the others. Nonetheless the constraint of limited data, the rationale behind the better performance of the second model, is clearly discussed

Rainfall prediction is a challenging task due to its dependency on many natural phenomenon. Some authors used Hurst exponent as a predictability indicator to ensure predictability of the time series before prediction. In this paper, a detailed analysis has been done to ascertain whether a definite relation exists between a strong Hurst exponent and predictability. The one-lead monthly rainfall prediction has been done for 19 rain gauge station of the Yarra river basin in Victoria, Australia using Artificial Neural Network. The prediction error in terms of normalized Root Mean Squared Error has been compared with Hurst exponent. The study establishes the truth of the hypothesis for only 6 stations out of 19 stations, and thus recommends further investigation to prove the hypothesis. This concept is relevant for any time series which need to be used for real time process control.