Land cover and land use change analysis using multi-spatial resolution data and object-based image analysis

Remote sensing data and techniques are reliable tools for monitoring and studying urban land cover and land use (LCLU) change. Fine spatial resolution (FRes) commercial satellite image in conjunction with geographic object-based image change analysis (GEOBICA) methods have been used to generate detailed and accurate urban LCLU maps. The integration of a backdating approach improves LCLU change classification results for the first date of a bi-temporal image sequences. Conversely, moderate spatial resolution satellite images such as those from Landsat sensors may not allow for detailed urban land use and land cover mapping. The objective of this study is to test a new bi-temporal change identification approach that integrates image classification of fine spatial resolution satellite imagery at time-2 and moderate spatial resolution satellite imagery (Landsat) at time-1, in a backdating and GEOBICA framework for mapping urban land use change. We compare the results from this approach to those of a GEOBICA approach based on fine spatial resolution imagery in both periods. The overall accuracy of the time-1 Landsat image classification is 0.82 and that of the fine spatial resolution image is 0.87. Moreover, the overall accuracy of the areal change data estimated from the pixel-wise spatial overlay of bi-temporal FRes LCLU maps is 0.80 while that from overlaying a time-2 FRes-generated map to that from a Landsat time-1 image is 0.81. The proposed method can be used in areas that lack FRes data due to limited coverage in the early 2000s.