Land cover change (LCC) has been a hot topic for decades. LCC is caused by not only human disturbances but also natural forces. Considering the advantages of the remote sensing data such as satellite imagery and aerial photos which can be applied to monitor some hard-to-reach locations, they can be used to extract land cover information in the hilly southern Guam. Therefore it is dispensable to find some remote sensing data available to Guam to derive land cover information, and to determine land cover change in space and time if temporal data are available. Thanks to the long term service of Landsat satellites since 1972, Landsat imagery has been utilized for resource management, environmental monitoring, coastal management, extraction of land cover and land cover change, just name a few here.

This study is based on a previous research, which was focused on detection of land cover change in watersheds in southern Guam. Two landsat images were used to determine watershed land cover change. These images are Landsat Multi-Spectral System (MSS) image of November 14, 1973, and Landsat Thematic Mapper (TM) image of March 15, 2001. Supervised classification was used to derive land cover information from these satellite images. When a classification process is done, it's essential to conduct an accuracy assessment to compare the classification to ground truth or other data. However, land cover classification accuracy has not been assessed yet. Therefore, the main purpose of this study aims to conduct land cover classification accuracy assessment. Historical GIS data, and higher resolution satellite images with acquisition dates close to those of the Landsat images are used as reference data for land cover accuracy assessment. The assessment indicates that both land cover results for 1973 and 2001 are satisfactory; however, land cover classification accuracy is affected by image resolution and data quality.