Geostatistical Solutions for Super-Resolution Land Cover Mapping

Scale in GIS: An overview

“…Recently, Boucher et al (2008) have shown how geostatistics can be used to support the challenging task of downscaling fields. In remote sensing, for example, the researcher is often constrained by the spatial resolution of the instrument used to create images of the Earth's surface.…”

Object-based sub-pixel mapping of buildings incorporating the prior shape information from remotely sensed imagery

“…Many different prior models have been proposed for SPM, and these can be classified to H-and L-resolution models in accordance with the relationship between the spatial resolutions of pixels in remotely sensed images and land cover patches (Atkinson, 2009). The Lresolution model features cases wherein pixels are much larger than the objects of interest; in this model, the spatial pattern of land cover classes is represented by spatial statistics models, such as the indicator variogram, two-point histogram, and multi-point histogram (Tatem et al, 2002;Boucher et al, 2008;Boucher, 2009). In the H-resolution model, the pixels are smaller than the objects of interest.…”

Soft-then-hard sub-pixel mapping with multiple shifted images

“…downscaling continua) is a critical step. Such a step can be realized by kriging (Verhoeye and Wulf 2002), backpropagation neural network (Mertens et al 2004;Gu, Zhang, and Zhang 2008;Nigussie, Zurita-Milla, and Clevers 2011;Zhang et al 2008), and indicator cokriging (Boucher, Kyriakidis, and Cronkite-Ratcliff 2008;Jin, Mountrakis, and Li 2012;Wang, Shi, and Wang 2014b). However, back-propagation neural network and indicator cokriging require prior spatial structure information at the target fine spatial resolution, which is always difficult to obtain in practical applications.…”