1. UKM Learning and Research Repository
2. PHD Thesis / Tesis PHD
3. Faculty of Engineering and Built Environment / Fakulti Kejuruteraan dan Alam Bina

Title:	Hybrid wavelet-artificial intelligence models for hydrometerological drought forecasting
Authors:	Md. Munir Hayet Khan (P80412)
Supervisor:	Nur Shazwani Muhammad, Ir. Dr.
Keywords:	Universiti Kebangsaan Malaysia -- Dissertations Dissertations, Academic -- Malaysia Drought forecasting Artificial intelligence
Issue Date:	4-Feb-2019
Description:	In the past few decades, Malaysia has been experiencing warm climates causing droughts. Therefore, drought forecasting is an essential tool for effective water resource management as well as mitigating some of the more adverse consequences of drought. This study attempts to assess and forecast meteorological and hydrological droughts. Trend analysis was done to detect upward or downward trend of drought indices. Artificial intelligence, data-driven models were used to forecast future drought events. Drought assessment were done using meteorological drought indices such as the Standardized Precipitation Index (SPI), Rainfall Anomaly Index (RAI) and Standard Index of Annual Precipitation (SIAP) and hydrological drought indices such as Streamflow Drought Index (SDI), Standardized Streamflow Index (SSFI), and Standardized Water Storage Index (SWSI). The analyses were done for the Langat river basin, Selangor, Malaysia using rainfall, streamflow and water level data for 30 years (1986 to 2016). These indices indicate that moderate to severe droughts occurred between 1998 and 1999, caused by El Nino. The results of trend analysis supported the findings in drought assessment by drought indices (DIs), where it was observed that Station 3 by SPI and Stations D, E, F by SWSI exhibited downward or decreasing trend. Four artificial neural network (ANN) models (Input model No. 1, 2, 3 and 4) were developed using SIAP and SPI meteorological DIs and four more ANN models (Input model No. 5, 6, 7 and 8) were developed using SWSI and SDI hydrological DI. The ANN model developed by SWSI, a hydrological drought index achieved an overall correlation coefficient (R) of 0.968, and the ANN model of SIAP, a meteorological drought index, achieved an overall R value of 0.899. Then, wavelet-based ANN models were developed using the drought index values of SIAP, SPI, SWSI and SDI. For meteorological drought (SIAP) prediction, the highest R value was 0.899. After preprocessing data by discrete wavelet transform (DWT), Wavelet-ANN model increased the R value to 0.940. Similarly, for the case of hydrological drought, the R value obtained by ANN model was 0.968, and with W-ANN SWSI model the value increased to 0.973. Finally, a discrete wavelet transformation-based hybrid ANN-ARIMA (W-2A) model was developed and showed that the input model 6 obtained from SWSI hydrological drought index performed better than all other W-2A models in terms of R2 (0.876) and RMSE (0.380). This study helps to understand the history of drought conditions of the past 30 years, exploring new methods to forecast droughts and assist in the management of water resources more effectively., Certification of Masters / Doctoral Thesis is not available,Ph.D.
Pages:	193
Call Number:	QC929.24.K483 2019 3 tesis
Publisher:	UKM, Bangi
Appears in Collections:	Faculty of Engineering and Built Environment / Fakulti Kejuruteraan dan Alam Bina

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.