RN-18 : Time series analysis models

Abstract

Time series analysis belongs to major statistical techniques used in the extraction of information on hydrologic and water resources random variables from observed data.-This report gives a brief review on time series models and steps used for time series modelling. Various criteria for the classification of time series models are presented and described. Available time series models are explained in the light of short memory models and long memory models. Short memory models include autoregressive (AR), moving average (MA), autoregressive moving average(ARMA), and autoregressive integrated moving average (ARIMA)models. Long memory models such as fast fractional Gaussian noise, filtered fractional Gaussian noise and broken line models are then described. Generation of daily data by shot noise model has been given. In the end disaggregation model and multisite models have been explained.

Some of the areas in which further study and research are needed have been identified by the review of literature. These include (i) time series analysis of water quality and quantity to meet the solution of complex environmental problems, (ii) development of more comprehensive families of time series models, (iii) physically based time series models, (iv) development of daily flow generating models with lesser parameters, (v) differential persistence and (vi), application of time series models (after modification) to Indian rivers as many of them have nearly zero flows during non-monsoon season (Nov. May).