Abstract:
Two types of data that are usually used of for analysing flood magnitude and frequency are the Annual Maximum Series (AMS) and the Partial Duration Series (PDS). This research study uses both of them to demonstrate the processes of flood magnitude and frequency analysis for each return period and to compare their efficiency by using different types of data series. Daily peak discharges in the Upper Ping River Basin are used as based data in order to study the condition and the appropriateness by using both types of data to predict and evaluate flood magnitude and frequency. The study covers methods/processes for choosing based flood (Ob), testing of data independency, parameters estimation, computing annual discharge and its variance, and comparing efficiency of data used by taking the ratio of the discharge variation. Different techniques such as: Exact Theoretical Approach (Rv,1), Approximately Theoretical Approach (Rv,2), Empirical Approach (Rv,3) and Mean Square Error Approach (Rm) are employed. After computing the differences of annual discharge and analyzing regional flood frequency, the results are then compared. This process involves the selection of frequency distribution function, i.e. Gumbel Function for AMS, Exponential and Poisson Functions for explaining the average magnitude and for number of exceedances of PDS, respectively. The results of analysis show that different methods of parameter estimation, i.e. Maximum Likelihood (ML) and Moment Method (MM), give similar results in comparing efficiency of data used. However, most of parameters from ML give less maximum difference between flood frequency from the Plotting Position method and the selected probability distribution function than from MM. Therefore, it can be concluded that ML is more suitable for parameter estimation than MM. Different techniques for comparing efficiency of data are employed. Rv,1 gives the most distinct result in which the PDS gives less discharge variation than the AMS, if it has the average number of exceedances at least in the range of 1.65-1.70. Other techniques such as: Rv,2 Rv,3 and Rm provide inconclusive outcomes because the recorded data is rather short. For the relation between annual discharge and its average at the return period of about 2.5, the estimation from the PDS gives higher annual discharge than the AMS, therefore it can improve reliability or provide less risk for designing hydraulics structures or flood relief measures. Furthermore, flood estimation from the PDS provides much more confidence in designing because it takes into account the stage and discharge from the physical condition thus the designing is in line with the natural condition. Data used by the PDS is firstly selected from the based flood and tested to ensure that the data in used is independent. In the case of the high based flood whereby there will be small data in the PDS series, the use of the AMS series in designing is inevitable. Other factors involving in this analysis such as the location of the recording stations, rainfall and the hydraulics structures in the area, contribute only little effect on the variation while the most important factor is the length of recorded data.