Distributed runoff simulation of extreme monsoon rainstorms in Malaysia using TREX
Date
2013
Authors
Abdullah, Jazuri, author
Julien, Pierre Y., advisor
Bledsoe, Brian P., committee member
Venayagamoorthy, Subhas K., committee member
Wohl, Ellen E., committee member
Journal Title
Journal ISSN
Volume Title
Abstract
Malaysia has a monsoon climate and most areas receive more than 2,500 mm of rainfall every year. For the past five years, the frequency and magnitude of floods in Malaysia have been relatively high. Floods have become the most significant type of natural disaster for Malaysia in terms of the population affected, financial losses and adverse socio-economic impact. This study uses the distributed two-dimensional TREX model to simulate infiltration, overland runoff and channel flow during extreme rainfall events. The main objective is to calibrate the distributed hydrological model to simulate monsoon floods. The second objective is to determine the affected flooding area under different rainfall events (i.e., large and extreme rainfall events). Large rainfall events cover return periods ranging from two to one hundred years. Extreme rainfall events include both the PMP and the world's largest rainfall events. The third objective is to examine the effect of rainfall duration on the magnitude of peak flood discharge as a function of watershed size. Finally, determine and produce graphs for the relationships between peak specific-discharge and watershed sizes. Three different sizes of watersheds are considered: Lui (small - 68 km2), Semenyih (medium - 236 km2) and Kota Tinggi (large - 1,635 km2). Generally, the topography of these watersheds is steep, except for the large watershed. The TREX model calibration and validation have been done using field measurements during several storm events. The performance of the model to find peak discharge, time to peak, and volume has been tested using three metrics: Relative Percentage Difference (RPD), Percentage Bias (PBIAS) and Nash-Sutcliffe Efficiency Coefficient (NSEC)) comparison. On average, the model performance was good for small (RPD - 7%, PBIAS - 14% and NSEC - 0.4) and medium watersheds (RPD - 14%, PBIAS - 28% and NSEC - 0.7). The RPD (4%), PBIAS (2%) and NSEC (0.8) for the large watershed shows that the model performance was very good. The spatial and temporal runoff distribution for overland and channel flows were successfully visualized in 3D. Both small and medium watersheds were not flooded by large events, except in the main channel. The flow depth reached 1.72 m in the valley of the small watershed only during extreme events. It was estimated that about 24% (±10%) and 83% (±5%) of the valley area exceed a flow depth of 1.72 m during PMP and world's largest events, respectively. For the medium watershed, the valley area was covered with water in excess of 4.49 m under the world's largest events. The visualization tool shows that the valley areas are prone to severe flooding (in excess of 4.49 m of flow depth) under this event (±5%). For the large watershed, the low land areas (i.e., along the tributaries and channels) are more likely to be flooded during large and extreme events. The water depths covered more than 2.8 m in these areas. The maximum estimated discharges (MED) for large rainfall events were highest for rainfall durations of 3 to 5 hours on small watersheds. However, the MED values for medium watersheds were obtained for rainfall durations between 5 and 12 hours. The MED values for extreme rainfall events were highest for rainfall durations between 10 and 13 hours on both watersheds. For the large watershed, the MED values of large and extreme events were obtained for a rainfall duration of 168 hour. The main conclusions of this study are: (1) rainfall intensity (i.e., hourly data) is one of the main factors that contribute to the magnitude of flooding on small and medium watersheds (watershed size less than 1,000 km2). The flooding events on large watersheds (watershed size more than 1,000 km2) result from longer rainfall durations (i.e., multi-day rainstorms), (2) for all size watersheds, the average magnitude of peak discharge for the PMP and the world's largest events are approximately 5 and 12 times larger than a 100-year rainfall event, (3) the peak specific-discharge (cms/km2) decreased as the watershed size (km2) increased, and (4) the runoff coefficient C increased significantly (i.e., a factor of three) from the 100-year rainfall event to the PMP and the world's largest events for all watersheds (CPMP,CWGR > 0.7).
Description
Rights Access
Subject
extreme monsoon rainstorms
TREX model
runoff simulation
Malaysia