SWAT and ANN model hydrological assessment using Malaysia soil data / Khairi Khalid

Khalid, Khairi (2017) SWAT and ANN model hydrological assessment using Malaysia soil data / Khairi Khalid. PhD thesis, Universiti Teknologi MARA.


Flood is a primary hazard affecting Malaysia, often responsible for loss of lives and a severe threat to infrastructure and environment. Activities in flood plain and catchment such as land clearing for other developments effort may increase the magnitude of a flood. The problem of flood management cannot be solved simply by providing more construction of dams and reservoirs. The adoption of a strategic approach is needed for planning and managing that flood management in any watershed. The evolution of distributed watershed models has been established for more accurate representation of the hydrological system by considering the spatial variability of model parameters and inputs. The research aims to evaluate the performance and hydrological response of the Soil Water Assessment Tool (SWAT) process-based model in tropical river basin using Malaysia soil data. This study was specific to the upper part of Langat River Basin (UPLRB) in the context of Greater Kuala Lumpur Plan in the southern region. The research also provides a streamflow prediction using the Artificial Neural Network (ANN) method as another tool for assessment. ArcSWAT2009.93.b, which is embedded in ArcGIS10, has been selected for this study, and the model requires comprehensive data on topography, soils, land use and daily weather data within a watershed. SWAT-CUP, which links SUFI-2 algorithm to SWAT models, has been utilized in the study for the calibration of SWAT models. There were two sets of algorithms in developing the UPLRB ANN model and every algorithm set consisted of model inputs data preparation, neural network script and neural network error checking measures. All the processes for ANN model were conducted in MATLAB software. The study found that five SWAT input parameters were required to show the most stable and sensitive outcome using both local and global sensitivities analysis techniques, inclusive of CN2.mgt, GW Delay.gw, SLOPE.hru, SOL_AWC.sol and SOL_K.sol. SWAT model performed better during the validation period compared to the calibration period in simulating streamflow at UPLRB. The runoff values were low at the upstream, and the evaluated value was increased from upstream sub-basin towards downstream sub-basin of the river basin with the exception of sub-basin 4 and sub-basin 10. The grand average of the surface runoff output for the period of the study ranged from 18.621 mm to 113.293 mm. The month of November experienced the highest monthly runoffs for all three different settings. The study also successfully produced two distinct sets of Neural Network Scripts to predict streamflow at UPLRB. Both models produced good results in predicting streamflow, and the existing AWC soil data in the ANN model did not significantly change the value of the simulation output. A comparison between the simulated streamflow by ANN model and SWAT model proved that the coupling of the outputs improved the results of the streamflow, mostly at the peak value of the monthly streamflow. It is hoped that the study can contribute to the improvement of integrated river basin management in tropical river basins.


Item Type: Thesis (PhD)
Email / ID Num.
Khalid, Khairi
Subjects: T Technology > TC Hydraulic engineering. Ocean engineering > Technical hydraulics
T Technology > TD Environmental technology. Sanitary engineering > Special types of environment. Including soil pollution, air pollution, noise pollution > Malaysia
Divisions: Universiti Teknologi MARA, Shah Alam > Faculty of Civil Engineering
Programme: Doctor of Philosophy (Water Resources and Environmental System)
Keywords: Hydrological, Soil data, Flood
Date: 2017
URI: https://ir.uitm.edu.my/id/eprint/27794
Edit Item
Edit Item


[thumbnail of TP_KHAIRI KHALID EC 17_5.pdf] Text

Download (130kB)

Digital Copy

Digital (fulltext) is available at:

Physical Copy

Physical status and holdings:
Item Status:

ID Number




Statistic details