dc.contributor.author	Harisaweni
dc.date.accessioned	2024-08-21T15:08:16Z
dc.date.available	2024-08-21T15:08:16Z
dc.date.issued	2015
dc.description	Thesis (PhD. (Civil Engineering))
dc.description.abstract	Short-time rainfall data is often used in many hydrological risk assessments. Unfortunately the availability of those data is scarce because of high capital and maintenance costs, especially in the less developed countries. Disaggregation techniques therefore have considerable appeal to solve this problem due to its ability to generate short-time rainfall data from a long-time scale. This study applied the Bartlett Lewis Rectangular Pulse (BLRP) methods in the hyetos program and a newly developed program using Artificial Neural Network method, namely MultiLayer Perceptron with Back Propagation (MLP-BP). Both methods were used to disaggregate daily rainfall data into hourly data. Three regions in Peninsular Malaysia were selected for this study, namely Damansara, Kelantan River Basin and South Johor. From those three regions, a total of 22 rainfall stations were selected for analysis. Based on the graphical findings of historical data over simulated results, the MLP-BP model outperformed the BLRP model in terms of its data synchronisation with the extreme values. In addition, the MLP-BP performed better for disaggregating rainfall data that fits the Generalized Pareto, Weibull, Gamma, Log Pearson 3 and Normal distributions. The BLRP methods performed better only in gamma and exponential distributions. Visually, the mean and standard deviations of the disaggregated rainfall that are obtained by both methods revealed similar patterns and are comparable with the historical records. However, statistical test showed that the simulated hourly rainfall using MLP-BP for 14 out of 22 rainfall stations closely matched the historical series. This means that the developed program of MLP-BP technique is better than the BLRP technique for disaggregating daily rainfall. In addition, the MLP-BP method can be applied for many types of rainfall distributions.
dc.description.sponsorship	Faculty of Civil Engineering
dc.identifier.uri	https://openscience.utm.my/handle/123456789/1313
dc.language.iso	en
dc.publisher	Universiti Teknologi Malaysia
dc.subject	Rain and rainfall—Mathematical models
dc.subject	Rain and rainfall—Measurement
dc.subject	Hydrologic models
dc.title	Disaggregation of daily into hourly rainfall using bartlett lewis rectangular pulse model and multi layer perceptron
dc.type	Thesis
dc.type	Dataset

Disaggregation of daily into hourly rainfall using bartlett lewis rectangular pulse model and multi layer perceptron

Files

Original bundle

License bundle

Collections