SA-1 : Time series modelling

Abstract

Hydrologic variables might vary in space, as well as time and are represented by either continuous or discrete series. Much of the statistical methodology is concerned with models in which the observations are assumed to vary independently. However, a great deal of the data in water resources planning and management occur in the form of time series where observations are dependent and the nature of this dependence is an important characteristic of time series.

The report gives a short review on hydrologic time series with particular emphasis on river flow time series modelling. The analysis of deterministic and stochastic components of time series has been discussed in the beginning and further extended to stochastic models. Short memory models like autoregressive models, moving average models, autoregressive moving average models, autoregressive integrated moving average models have been discussed in the light of identification of the model, parameter estimation and diagnostic checking. Long memory models like fractional Gaussian noise models , fast fractional Gaussian noise models and broken line models have also been described. Details of multisite short memory and long memory models have been provided. Final remarks including areas in which further research is needed and extensive list of references is given at the end of the report.