Comparison of OLI and TM Multi-spectral Satellite Imagery Land-use and Land-cover Mapping Using Hierarchical Concept of Earth Surface Matrix

Abstract

The present study compares the capabilities of Thematic Mapper (TM) and Operational Land Imager (OLI) sensors of Landsat satellite and analyzes the results of image classification on their multi-spectral data. To achieve this, LANDSAT 5-TM (2011) and LANDSAT 8-OLI (2016) imageries were used to map the land-use and land-cover for a study area located on Pelasjan sub-basin, in Isfahan, Iran. First, radiometric and atmospheric corrections were performed, and then the overall status of the area was determined by reviewing topographic maps, visual interpretation of the satellite imageries and field studies.  Consequently, a three-level land matrix hierarchy including 1) General level, 2) Mid-level, and 3) The level of details was established. Land matrix hierarchy maps were produced with proper methods using hybrid classification. The comparative analysis in this study showed that the hybrid classification method generates accurate results from the OLI sensor data in comparison to TM imageries. This was particularly evident for residential areas, irrigated agriculture, rain-fed agriculture, sparse, and dense rangelands. Although the results of image classification showed more accuracy for the OLI imagery, the error matrix in Z-test did not identify any statistically significant difference between the two datasets. This highlights the importance of image classification method selection, which can overcome the possible limitations of satellite imageries in land-use and land-cover mapping.

Keywords: OLI and TM Sensors, land matrix hierarchy, hybrid classification, LULC, error matrix.