Land cover change detection with change vector in the red and near-infrared reflectance space

An enhanced land cover change indicator product is produced using the two 250-m spatial resolution bands of the moderate resolution spectroradiometer (MODIS) of the NASA Earth Observing System. The rationale for creating the 250-m resolution land cover change product is that a very high proportion of land cover changes occur at the finest MODIS spatial resolutions. Multiple change detection algorithms are employed for the product generated because different algorithms detect different types of change. Among these algorithms there are two using the change vector in the red and near-infrared reflectance space. An early example of using change vector analysis for change detection is in Malila (1980). A more recent example is Johnson and Kasischke (1998). In these examples the general concepts of multispectral change vector analysis were described and applied to detect vegetation changes for the cases in specific locations and seasons. In this paper, the change vector in the red and near-infrared (NIR) reflectance space is analyzed for different types of relevant land cover changes. The algorithms using the change vector characteristics obtained with NOAA's Advanced Very High Resolution Radiometer (AVHRR) data to detect the changes are then described. Validations of these two algorithms with the simulated MODIS data from Landsat Thematic Mapper (TM) images of two different areas are presented.