Spatial uncertainty analysis when mapping natural resources using remotely sensed data

Combining field plot observations and remotely sensed data has been widely used to generate spatially explicit estimates of natural resources such as leaf area index, forest biomass/carbon, biodiversity, soil types, and soil erosion (Wang et al. 2007, 2009b; Mascaro et al. 2011a; Lu et al. 2012). The used data and information possess a large amount of uncertainties and errors (Wang et al. 2009b, 2011). The procedures to generate maps of natural resources also lead to uncertainties (Wang et al. 2005). Therefore, natural resource estimates are associated with uncertainties (Gertner et al. 1995, 1996, 2002a, 2002b; Fang et al. 2002; Sierra et al. 2007; Larocque et al. 2008; Nabuurs et al. 2008; Mascaro et al. 2011b). Using the obtained maps in decision supports will definitely result in numerous risks (Heath and Smith 2000). Thus, how to quantify and reduce uncertainties of natural resource estimates is becoming very important.