CHAPTER I INTRODUCTION 1.1 Background At present, stochastic modelling of seasonal hydrological time series is an accepted approach for data generation and forecasting of hydrological events. Although almost from the beginning of the century the importance of statistical approaches for analysing hydrological data was evident, it was not until the mid 1950’s that formal development of stochastic modelling started with the introduction and application of autoregressive models for monthly rainfall (Hannan, 1955). Since then, extensive research efforts have been towards improving the early concepts and models, providing physical (conceptual) justification of some models, introducing alternative models and developing and applying improved estimation procedures as fitness tests for such models. About thirty-five years has elapsed since Hannan first suggested the lag-one auto regressive model for the modelling and simulation of monthly rainfall. Since then significant steps in this area are represented by the works of Thomas and Fiering (1962) for modelling annual and seasonal streamflows, Yevjevich (1963) and Roesner and Yevjevich (1966) for modelling annual runoff and monthly precipitation and streamflow, respectively; Matalas (1967) for multivariate lag-one autoregressive modelling and Yevjevich (1972a) for extending the lag-one autoregressive models to multilag with seasonal parameters. About the same time, the book of Box and Jenkins (1970) stirred interest in the application of a new class of models, namely the 2 autoregressive-moving average (ARMA) models which have gained wide application in stochastic hydrology (Carlson et al., 1970; O’Connell, 1971; McKerchar and Delleur, 1973; Hipel et al., 1977; Lettenmaier and Burges, 1977; Curry and Bras, 1978; Salas et al., 1980; Cooper and Wood, 1980; Rao et al., 1982; and Stedinger and Taylor, 1982). A landmark in the modelling of seasonal hydrology time series was the development of the disaggregation model (Valancia and Schaake, 1973), which allows the generation of seasonal hydrological samples by preserving statistics at both annual and seasonal time scales. Improvements and applications of disaggreregation models have also been made (see for instance, Mejia and Rousselle, 1974; Todini, 1980; Lane, 1982; Santos and Salas, 1983; Valencia et al., 1983 and Stedinger aand Vogel, 1984). Synthetic hydrology is an accepted practice for the planning and management of the water resources systems. This study will focus in simulation of rainfall. Depending on the particular system considered either seasonal or annual synthetic rainfall might be required. Stochastic simulation of water resources times series in general and hydrologic time series in particular has been widely for several decades for various problems related to planning and management of water resources systems (Salas et al, 1980; Loucks et al, 1981). Stochastic simulation of hydrologic time series such as rainfall is typically based on mathematic models. For this purpose a number of stochastic models have been suggested in literature (Salas, 1993; Hipel and Mcleod, 1994). The solution from many hydrological problems is intimately related to a better knowledge of the rainfall process such as the development of reliable streamflow and groundwater models, the prediction of floods and droughts, the design of the data collection networks among others. In this study, the disaggregation methods were used to generate accurate synthetic rainfall data to predict the accurate runoff. In general, all three disaggregation models satisfactorily preserve annual and seasonal (monthly) rainfall generation. 3 1.2 Statement of Problems Stoshastic data generation aims to provide alternative hydrologic data sequences that are likely to occur in future. These data sequences, particularly monthly time series are widely used in water resources planning and operation studies to assess the reliability of alternative system designs and policies and to understand the variability in future system performances (Loucks et al., 1981). However, for valid and realistic results is necessary that these synthetic monthly data sequences should preserve both monthly and annual statistical parameters of historical data especially mean, variance, correlation and skewness. As monthly data generation models such as the Thomas-Feiring model (Thomas and Feiring, 1962) rarely produce monthly data sequences that preserve both monthly and annual parameters of the historical record, disaggregation models are currently being widely used to generate such data sequences. Disaggregation models produce monthly data sequences by disaggregation annual data that have been generated by suitable annual data generation model. The purpose of rainfall generation is to generate accurate, reliable synthetic rainfall data for predict accurate runoff from this generation. Disaggregation method is uses to preserve the statistical characteristics such as mean, standard deviation, skewness and season to season correlation. This study will prove that synthetic rainfall generation can obtained a good model. These models can preserve the statistical characteristic of historical data because synthetic runoff generation not easily preserved the statistical characteristic. This problem happened with many reasons such as infrastructure that build at river basin, economic and etc. In this study, the problem will solve when the rainfall generation applied. 4 1.3 Objectives In general, this study focuses the stochastic simulation of rainfall time series for seasonal data. The objectives of this study are; 1. To apply the disaggregation models in the rainfall simulation. 2. To evaluate the model performance in the rainfall simulation. 3. To identify the best model in the rainfall simulation. 1.4 Scope The study covers the application disaggregation methods. Those methods composed of basic model (Valencia and Schaake, 1973); extended model (Mejia and Rousselle, 1973) and the condensed model (Lane, 1979; Grygier and Stedinger, 1991). The data obtained from the JPS Hydrology and Water Resources Division, one of division in Department of Irrigation and Drainage of Malaysia (DID). The utility of this approach is to demonstrate through the application of mean monthly rainfall from 5 sites in Johor measured in millimetre (mm). The data modelled were the data rainfall for 52 years from 1949 until 2000. The study uses the disaggregation method to the monthly (seasonal) and annual rainfall data for single site multi season case.