Remote Sensing Data Normalization

In this chapter, the terms remote sensing data and remote sensing images are used interchangeably. In order to be able to use data in any combined fashion, each individual dataset needs to have the same reference that allows quantitative comparisons. Ÿe most generic term for this is data normalization and for remote sensing data, it is known as radiometric normalization. Depending on the data source, the processing and necessary corrections di§er. Although the terms and de¦- nitions used for remote sensing data from di§erent sources might vary, there are two main categories of radiometric normalization (Bao et al., 2012): absolute normalization (methods based on radiative transfer methods that account for atmospheric, illumination, and sensor di§erences) and relative normalization (techniques that minimize the e§ects of changing atmospheric and solar conditions in one or a series of images, relative to a standard image). For optical and thermal imagery, radiometric corrections need to be applied to account for atmospheric conditions, solar angle, or sensor view angle (Chen et al., 2005; Du et al., 2002), in addition to the sensor prelaunch and postlaunch calibration. Ÿese corrections help to convert the raw signal recorded at the sensor to physically meaningful and measurable values, such as ground re©ectance or ground temperatures. Ÿe quantitative use of synthetic aperture radar (SAR) data requires calibrated imagery (Freeman, 1992). Speci¦cally, the SAR processor used for the image generation needs to be calibrated and the calibration parameters then need to be applied to the data to generate calibrated images.