| Summary: | This thesis focuses on hydrological study of selected watersheds on the Island of Oahu, Hawaii. The study includes rainfall frequency analysis for six selected rain gages situated within different watersheds, namely, Hawaii Kai Golf Course 724.19, Kailua Fire Station 791.3, Kahuku 912, Makaha, Pupukea Height 896, and Waimea 892. Runoff prediction based on three different methods for Kamananui Stream found within Waimea Watershed is also presented in this study. The GIS/ArcInfo software package is used in this study as a supporting tool. For rainfall frequency analysis, Gumbel Distribution and Log Pearson Type III Distribution for extreme values are applied to predict the return intervals of different rainfall intensities based on 24-hour duration for the selected watersheds according to long-term hourly rainfall data that were available up to year 2002. The calculated results are compared with the predictions from two earlier studies published in 1962 and 1984. The objective is to examine the effect of newer and longer rainfall data on the frequency analysis results. Our results show that the rainfall intensity of different return intervals predicted in the present study using the newer and longer data records in general agrees quite well with the results predicted by the two earlier studies. A further study on the effect of record length on the accuracy of rainfall frequency prediction is carried out and the results show that the average error in predicting rainfall frequency based on 15-year and 20-year records is similar to that based on longer records (e.g., 30- and 40-year records), however, longer records provide more consistent predictions with smaller uncertainty. This result confirms that for predicting a value of return period Y based on the Gumbel distribution, a data record longer than OAY years (40% of recorded data) is preferable. In the present study, intensity-duration-frequency (IDF) curves are also developed for Kaelepulu and Waimea Watersheds based on long-term I5-minute rainfall data. The conversion factor for converting 1-hour rainfall intensity to rainfall intensity of other periods based on the IDF curves is determined. The present result is compared with the empirical conversion factor used in engineering design. Our results show that there are some noticeable differences between the empirical values and the actual values of the conversion factor as indicated in the City and County of Honolulu Storm Drainage Standards Manual. However, the differences are relatively small and the empirical factor is acceptable for engineering practice. For stream runoff prediction, three different methods, namely, the rational method, the Soil Conservation Service's (SCS) (now known as Natural Resources Conservation Service (NRCS)) TR-20 method, and the USGS regression method, are used to predict runoff amounts for gaged Waimea Watershed. The objective of this part of the study is to examine the consistency and validity of the three different methods for predicting stream flow in Hawaii. Our results show that all three methods give quite consistent predictions for runoff of different return periods in Waimea Watershed despite the fact that the area of the watershed is relatively large and exceeds the upper limit for the area for the rational method. The predicted runoff results also show good agreement with the measured results based on the stream gage record. === xii, 110 leaves
|
|---|
