Unlike small-scale hydraulic structures, large-scale hydraulic structures such as dam have been designed based on extreme floods. PMF(Probable Maximum Flood) is mostly applied for the designs of large-scale hydraulic structures and it estimated by computing the runoff hydrograph where PMP(Probable Maximum Precipitation) is inserted as design rainfall input parameter. PMF determines the design flood discharge of large-scale hydraulic structures and the estimated PMF greatly depends on PMP. Therefore, in order to estimate the PMF with high accuracy, it is very important to improve the accuracy of PMP.
The existing PMP is estimated by transferring the heavy rainfall from
all watersheds of korea to the design watershed. Thus, there may be concerns of overestimation because it does not properly reflect the characteristics of the design watershed. Therefore, in this study, PMP was analyzed by selecting only rainfall events occurred in the design watershed. In order to accurately analyze the characteristics of the design watershed, the Average Point-tracking method which can reflect spatio-temporal characteristics of the rainfall was utilized and multiply with the moisture maximizing rate for estimation of PMP.
Therefore, this study proposed a new PMP estimation methodology based on watershed with the procedures as above mentioned. After that, the CSEM model was used to estimate the PMF based on the estimated PMP. In the end, the PMF and the sediment-runoff yield for Yongdam-dam basin were estimated.

The summary of process and result is as following.

1. The existing PMP estimtation method and the PMP estimation method proposed in this study were presented and the differences between these two methods were evaluated. In order to compare the estimation method proposed in this study, the design watershed was selected as the Yongdam-dam watershed. From 2002 to 2014, a total of 18 heavy rainfall events exceeding the standards of heavy rain warning and heavy rain advisory were selected.

2. The DADs analyzed in all events were multiplied by the moisture maximizing rate, and PMP was estimated by listing the maximum values for each area and duration. As a result, the 24-hour PMP of the YD1 basin was estimated to be 366 mm, the 24-hour PMP of the YD2 basin was 359 mm, the 24-hour PMP of the YD3 basin was 401 mm and the 24-hour PMP of the YD4 basin was 398 mm.

3. The CSEM model was used to simulate the rainfall-runoff and sediment-runoff behavior of the estimated PMP in the Yongdam-dam basin. A total of nine parameters were selected to determine the surface runoff and sediment runoff and they were calibrated using SCE-UA method. The simulation results indicated that the PMF corresponding to the estimated PMP was 7,799 m3/s and the sediment yield was estimated as 696 mg/l.

4. Comparing with the PMF estimated by the lumped model, the PMF estimated using distributed model was 1,333 m3/s higher than PMF using the lumped model. After that, the PMF estimated in this study compared with the existing PMF and the 200-year frequency design flood discharge. Consequently, the existing PMF was 1.38 times higher as compared with the PMF estimated using watershed-based PMP. It is due to different deriving procedures used in storm area based approach and watershed based approach for estimating PMP. After reviewing the results, The PMF estimated in this study is considered to be out of the risk of underestimation because it was estimated to be 1.75 times higher than the 200-year frequency design flood discharge.

PMP and PMF according to the existing guidelines may be reasonable in areas with high flooding probability and frequent meteorological changes, but there is a possibility of overestimation in areas where flood risk is relatively lower, and abnormal climate changes is relatively rare. Therefore, application of the method proposed in this thesis along with the existing method for respective flood risk of a watershed could reduce the possibility of PMF overestimation. However, it is necessary to evaluate the applicability of the proposed methodology in several watersheds other than Yongdam-dam basin and carry out a comparative study using rainfall data observed using radar.