Comparison between sensors with different spectral resolutions, relative to the sumbandila satellite, for assessing site quality differences, in eucalyptus grandis plantations

Abstract

The majority of earth observation studies have for years made use of data from multispectral spaceborne sensors. More recently though, both airborne and spaceborne narrowband hyperspectral sensors have come to the fore and are able to provide more detailed spectral information. Narrowband sensors, with their many contiguous bands, have proved useful in discriminating between vegetation states, e.g. water stress and nutrient deficiencies. However, hyperspectral remote sensing has a number of disadvantages related to the cost of data collection (especially airborne data), the low signal-to-noise ratios of spaceborne hyperspectral data, and data redundancy given the large number of contiguous bands. These challenges, exacerbated by known broadband sensor limitations, have prompted research into the reduction of hyperspectral data dimensionality towards the identification of application-specific spectral features, which ultimately could lead to the definition of new application-centric multispectral sensors. SUNSPACE, in collaboration with the University of Stellenbosch and the South African Department of Science and Technology, aim to launch a high resolution multispectral micro satellite (SumbandilaSat) in 2009. The satellite will be equipped with a multispectral sensor, with a 6.25m spatial resolution and 6 spectral bands. Two of these spectral bands have been strategically placed given a priori correlations between the vegetation signal in these regions and vegetation state, that is the red edge band (690-730nm) and xanthophyll band (520-540nm). The objective of this study was to investigate whether fewer and strategically placed multispectral wavebands could provide similar, and/or more cost effective information compared to common hyperspectral sensors. Canopy-level ASD data were collected and resampled to various sensor resolutions, that is Hyperion, HyMap, AVIRIS and SumbandilaSat