DETERMINING THE ACCURACY OF GLOBELAND30 LAND COVER MAP USING COMPARATIVE ANALYSIS WITH ERROR MATRIX, UNBIASED AREA ESTIMATION AND GROUND TRUTHING: A CASE STUDY OF MALAYSIA’S NORTHERN HEMISPHERE

• Main Author: MOHD SAID, SYAZMEER ARRAFI
• Format: Final Year Project
• Language: English
• Published: Universiti Teknologi PETRONAS 2018
• Subjects: TA Engineering (General). Civil engineering (General)
• Online Access: http://utpedia.utp.edu.my/22629/1/Dissertation_Syazmeer_19872.pdf; http://utpedia.utp.edu.my/22629/

Land cover maps (LCM) have the potential to greatly enhance the way the lands are managed and helps to predict the effects of human–nature interactions such as urban planning, predict and assess impacts of flood or even studying land change over time. GlobeLand30 (GL30) is one such land cover map that provides 30-m spatial resolution datasets which are relatively high compared to its other counterpart such as the CCILC. In order to utilize the high resolution map GL30 provides, it is important to validate its accuracy. This project validates the accuracy of GL30 land cover map using three method. The first one is the error matrix analysis where the GL30 land cover map is compared to a sattelite imagery map such as Google Map. The second method is extending the error matrix with the use of Unbiased Area Estimation that takes into consideration the LC type pixel count and in turn the area cover as well.