Title:	SPATIO-TEMPORAL REGRESSION ANALYSIS ON SHORELINES DATA EXTRACTED FROM HISTORICAL SATELLITE IMAGES
Author(s):	Syaifulnizam Abd Manaf, Norwati Mustapha, Md Nasir Sulaiman, Nor Azura Husin, Helmi Zulhaidi Mohd Shafri and Mohd Radzi Abdul Hamid
ISBN:	978-989-8533-82-1
Editors:	Pedro Isaías and Hans Weghorn
Year:	2018
Edition:	Single
Keywords:	Shoreline Prediction, Shoreline Change, Satellite Images, Regression Analysis, Change Analysis
Type:	Full Paper
First Page:	223
Last Page:	233
Language:	English
Cover:
Full Contents:	click to dowload
Paper Abstract:	With the increasing number of sensors of the Earth observation satellites, data were collected at all times all over our planet. The longest satellite mission was Landsat, with the first operation starting from early 1970’s and still operations until today. Landsat satellites collect a vast amount of images with the latest Landsat 8 satellites alone collect 1,200 images per day at approximately 1 GB per image. However, acquiring the suitable images that cover the area of study was not directly proportional to the number of images produced. It depended on the chosen area and the weather conditions while capturing the images. The chosen images considered cleaned from thick cloud and haze that required minimal pre-processing tasks. In the case of coastal zone monitoring, the information about conditions of coastal areas provided whether affected by natural or human activities. Monitoring coastal zones could be applied by monitoring shorelines, which are an important criterion to measure boundary of a country. In this research, Pontian district was chosen as the area of study. It has been identified as one of the most vulnerable places in Malaysia due to sea-level rise. For extraction, satellite image classification using Artificial Neural Network was employed to classify land and water classes. Then, historical shorelines were extracted for the year 1977, 1989, 2000, 2008, 2014 and 2017 covering a medium-term of 40 years’ period. Rates of change statistics were calculated using the End Point Rate method. Approximately 753 transects were cast at left and right of shorelines along the entire coast at 50 m interval. The prediction of year 2017 were assessed using the Linear Regression and the End Point Rate methods. Among the two, Linear Regression was used to project the future positions based on historical shoreline data because it attained the lowest error after validating model of 2017 of 1977 to 2014 data with 2017 extracted data. Using the same prediction model, future shoreline positions were projected from 2020 to 2100 using all historical shorelines data.
	Go Back