UTAR Institutional Repository

Geographical Weighted Regression (GWR) Rainfall Spatial Distribution and Variability In Malaysia

Chew, Kim Soon (2021) Geographical Weighted Regression (GWR) Rainfall Spatial Distribution and Variability In Malaysia. Final Year Project, UTAR.

Download (17Mb) | Preview


    The aims of this study are to compare the spatial interpolation methods and study the rainfall patterns in Peninsular Malaysia. Geographical Weighted Regression (GWR) and Multiscale Geographical Weighted Regression (MGWR) were used to analyse the monthly rainfall data. The GWR is an extension of the traditional linear regression model as it is using the spatial (x, y) coordinates to build up a relationship between location and other parameters. The MGWR is a further improvement of the model of GWR; in which it removes the constraint of all analysis and is modelled using different spatial scale with different bandwidth. Moreover, MGWR allows the range of databorrowing to vary across the parameter surfaces so that the scale of the independent variable, and dependent variable will not be inconsistent across the analysis. The daily rainfall data for Peninsular Malaysia during 1988-2017 was acquired from the Department of Irrigation and Drainage (DID) for the analysis. The missing rainfall data was repaired in order to increase the estimation accuracy of both methods. The monthly rainfall data, number of wet days and maximum daily rainfall were extrapolated from the daily rainfall data. After that, the rainfall data was broken down into 6 sub-parts, with each part being of a length of 5 years. The root means square error (RMSE), mean absolute error (MAE) and coefficient of determination (R2 ) were used to evaluate the GWR and MGWR to study the accuracy for predicting the monthly rainfall. Besides, the rainfall stations were zoned into four regions such as northern region, east coast region, southern region and central region, in order to further study the accuracy of both methods in different regions. The results shown that the MGWR has a better performance compared to GWR in the estimation of rainfall data of Peninsular Malaysia as a whole, or as zoned into the four regions, as MGWR has higher R2 and lower RMSE and MAE compared to GWR in all cases. During the peak of the Northeast Monsoon, which are November, December and January, the high average Number of Wet Days and Maximum Daily Rainfall contributed to high average Monthly Rainfall at the northeast region of Peninsular Malaysia. From May to September, the average Monthly Rainfall was high at the northern region of Peninsular Malaysia due to the contribution of high Number of Wet Days. The vi rainfall was found only be concentrated at the northern region of Peninsular Malaysia during the Southwest Monsoon, due to most of the rainfall being block by the island of Sumatra, Indonesia. The average Monthly Rainfall, Number of Wet Days and Maximum Daily Rainfall from January to December were also divided into 5-year periods. It was noticed that the major issue was the effect of increasing average Monthly Rainfall becoming significant in the month of December and January in the year 2013-2017, when comparing with sub-period of year 1988-1992, 1993-1997, 1998-2002, 2003-2007 and 2008- 2012.

    Item Type: Final Year Project / Dissertation / Thesis (Final Year Project)
    Subjects: T Technology > TA Engineering (General). Civil engineering (General)
    Divisions: Lee Kong Chian Faculty of Engineering and Science > Bachelor of Engineering (Honours) Civil Engineering
    Depositing User: Sg Long Library
    Date Deposited: 12 Jun 2021 05:02
    Last Modified: 10 Aug 2023 20:27
    URI: http://eprints.utar.edu.my/id/eprint/4065

    Actions (login required)

    View Item