Hyperspectral remote sensing of vegetation and agricultural crops

There are now over 40 years of research in hyperspectral remote sensing (or imaging spectroscopy) of vegetation and agricultural crops (Thenkabail et al., 2011a). Even though much of the early research in hyperspectral remote sensing was overwhelmingly focused on minerals, now there is substantial literature in characterization, monitoring, modeling, and mapping of vegetation and agricultural crops using ground-based, platform-mounted, airborne, Unmanned Aerial Vehicle (UAV) mounted, and spaceborne hyperspectral remote sensing (Swatantran et al., 2011; Atzberger, 2013; Middleton et al., 2013; Schlemmer et al., 2013; Thenkabail et al., 2013; Udelhoven et al., 2013; Zhang et al., 2013). The state-of-the-art in hyperspectral remote sensing of vegetation and agriculture shows significant enhancement over conventional remote sensing, leading to improved and targeted modeling and mapping of specific agricultural characteristics such as: (a) biophysical and biochemical quantities (Galvão, 2011; Clark and Roberts, 2012), (b) crop type\species (Thenkabail et al., 2013), (c) management and stress factors such as nitrogen deficiency, moisture deficiency, or drought conditions (Delalieux et al., 2009; Gitelson, 2013; Slonecker et al., 2013), and (d) water use and water productivities (Thenkabail et al., 2013). At the same time, overcoming Hughes’ phenomenon or curse of dimensionality of data and data redundancy (Plaza et al., 2009) is of great importance to make rapid advances in a much wider utilization of hyperspectral data. This is because, for a specific application, a large number of hyperspectral bands are redundant (Thenkabail et al., 2013). Selecting the relevant bands will require the use of data mining techniques (Burger and Gowen, 2011) to focus on utilizing the optimal or best ones to maximize the efficiency of data use and reduce unnecessary computing