Normalized Difference Reflectance: An Approach to Quantitative Change Detection

Remote Sensing is usually used in practice as a tool for describing and comprehending the land cover status and the physical properties of Earth’s surface, but not only. The increasing pressure of human activities on environment and the need to forecast, monitor and manage natural hazards - often due to climate change - have forced the scientific community to focus its attention on the study of environmental dynamics and timeseries, thus giving great importance to Change Detection Techniques. This paper introduces a simple and effective approach to Change Detection from satellite images: the Normalized Difference Reflectance (NDR), a multi-spectral quantitative way to describe radiometric variation of surface features, variation that has to be interpreted then in order to distinguish real changes in land cover from changes related to other variations. Pre-processed multi-spectral and multi-date ASTER and ETM+ data were subjected to parametric radiometric normalization and hence converted to NDR values; these values were used then as a base for applying supervised or unsupervised classification methods to produce change maps of the areas investigated. A case study regarding a flood event and one other dealing with urban expansion were analyzed and assessed.