Author :
Chini, M. ; Emery, W.J. ; Pierdicca, N.
Abstract :
The goal of this study is to detect and map the land use and land cover changes of the rocky flats nuclear facility between 2002 and 2005, using high resolution QuickBird satellite images provided by DigitalGlobe. The dataset consists of four multi-spectral QuickBird images, one image per year, with a spatial resolution of 2.4 meters. The evolution of the changes that occurred between these years has been mapped showing changes between different land covers, such as man made buildings, water and soil. We have used a maximum likelihood supervised classifier to map the changes between two different periods using as input all eight bands of each couple of images, plus their respective normalized difference vegetation index (NDVI) coefficients. In addition, it has been necessary to use mathematical morphology operators to correct some false alarms in the classification which are due to objects, such as cars, which are classified like a surface change, but are actually only temporary features. We have clearly documented the demolition of all of the buildings, the conversion of asphalt parking lots to soil, and the creation of drainage channels. We have also monitored the change in water bodies from the earlier distribution to 2005.
Keywords :
cartography; image classification; DigitalGlobe; asphalt parking lots; drainage channels; high resolution QuickBird satellite images; land cover change detection; land use mapping; mathematical morphology operators; maximum likelihood supervised classifier; multispectral QuickBird images; normalized difference vegetation index; rocky flats nuclear site demolition; spatial resolution; Event detection; Gas detectors; Image resolution; Maximum likelihood detection; Monitoring; Remote sensing; Satellites; Soil; Spatial resolution; Urban areas;