Single-site modeling of rainfall based on the Bartlett-Lewis rectangular pulse model
Sub-daily time-scale data such as hourly base are important for the purpose of modeling the urban system. However, as similar data may not be readily available, a stochastic rainfall model is mandatory to generate reliable rainfall series that have similar properties as those of the observed in orde...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2010
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/id/eprint/16334/7/LauKimSoonMFSA2010.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Sub-daily time-scale data such as hourly base are important for the purpose of modeling the urban system. However, as similar data may not be readily available, a stochastic rainfall model is mandatory to generate reliable rainfall series that have similar properties as those of the observed in order to estimate the input for design work in the future. In this study, one of the famous models that applied the Poisson clustered point process is the Bartlett-Lewis Rectangular Pulse Model (BLRPM) will be used to access a 10-year hourly rainfall data from Station Tele Ulu Remis, Johore, Malaysia. This model applies a flexible fitting procedure to match approximately to the historical data by an optimization technique called as Shuffle Complex Evolution (SCE). SCE algorithms is chosen for the parameters estimation by minimizing an objective function with six parameters ?, ?, ?, ?x, ? and ?. The SCE algorithm performs very well in obtaining the global optimum value and the time to get the optimum value is fast. The estimated parameters for the month of November and December were compared with Powell method for validity. Subsequently, an hourly and daily rainfall simulation based on BLRPM is carried out. The performance of the BLRPM is then evaluated on a monthly basis in term of its ability to preserve statistical and physical properties. The said properties involve rainfall series over time-scales of 1-hr and 24-hr. Results from the model evaluation have indicated that BLRPM is capable to reproduce most of the statistical and physical properties of the historical data. There are also some properties that are failed to be preserved accurately. However, the BLRPM is still capable to preserve the trend of the observed properties. |
|---|