C. Gomez

The Use of Hyperspectral Imagery for Digital Soil Mapping in Mediterranean Areas

Hyperspectral imagery is considered as a promising source of data to overcome the lack of soil information that often hamper digital soil mapping. We have tested it in the vineyard plain of Languedoc (southern France) using an 5×5 m resolution HYMAP image and 52 calibration-validation points. Satisfactory predictions of clay content and calcium carbonate (CaCO3) content were first obtained from HYMAP spectra over bare soils, partial least-squares regression performing better than continuum removal technique. These predictions were however less precise than using laboratory spectra. An examination of the possible factors that could explain this decrease showed that calibration uncertainties of the HYMAP sensor and of atmospheric effects were largely predominant. Secondly, since the HYMAP image was largely covered by vegetation with few pure bare soil pixels, an interpolation-aggregation procedure was proposed to obtain a 100×100 m digital soil map of the whole study area from a set of scattered bare soil fields with hyperspectral soil characterization. Interpolation was performed by a conditional simulation algorithm to estimate the within pixel soil pattern parameters. Validation results showed that satisfactory estimates of local means can be obtained whereas the variations of local variances were only partly represented.

The role of atmospheric correction algorithms in the prediction of soil organic carbon from Hyperion data

ABSTRACT In this study, the role of atmospheric correction algorithm in the prediction of soil organic carbon (SOC) from spaceborne hyperspectral sensor (Hyperion) visible near-infrared (vis-NIR, 400–2500 nm) data was analysed in fields located in two different geographical settings, viz. Karnataka in India and Narrabri in Australia. Atmospheric correction algorithms, (1) ATmospheric CORection (ATCOR), (2) Fast Line-of-sight Atmospheric Analysis of Spectral Hypercubes (FLAASH), (3) 6S, and (4) QUick Atmospheric Correction (QUAC), were employed for retrieving spectral reflectance from radiance image. The results showed that ATCOR corrected spectra coupled with partial least square regression prediction model, produced the best SOC prediction performances, irrespective of the study area. Comparing the results across study areas, Karnataka region gave lower prediction accuracy than Narrabri region. This may be explained due to difference in spatial arrangement of field conditions. A spectral similarity comparison of atmospherically corrected Hyperion spectra of soil samples with field-measured vis-NIR spectra was performed. Among the atmospheric correction algorithms, ATCOR corrected spectra found to capture the pattern in soil reflectance curve near 2200 nm. ATCOR’s finer spectral sampling distance in shortwave infrared wavelength region compared to other models may be the main reason for its better performance. This work would open up a great scope for accurate SOC mapping when future hyperspectral missions are realized.

Vis-NIR-SWIR Remote Sensing Products as New Soil Data for Digital Soil Mapping

Since the early ages of soil surveys, air photographs have been intensively used by soil surveyors for depicting the soil variations across landscapes. The variations of soil surfaces, specifically color and ratio of vegetation cover, that were revealed by this early remote sensing product were a great help for interpolating the scarce soil observations and for delineating the soil class boundaries. This was further transposed in digital soil mapping (McBratney et al. 2003), thanks to the large availability of remote sensing images provided by the emerging spatial data infrastructures. Up to now, digital soil mappers have mainly used remote sensing images as spatial data inputs for representing the landscape variables that are related with soil, such as vegetation and parent material (the soil covariates). Boettinger et al. (2008) reviewed the main indicators that could be retrieved for estimating these soil covariates, using multispectral data acquired in the visible near-infrared and short-wave infrared (VIS, 400–700 nm; NIR, 700–1100 nm; SWIR, 1100–2500 nm) spectral domain. After a spatial overlay with the sparse sets of observed and measured sites collected in a given area, the indicators derived from remote sensing have been used as independent variables in regression-like models or as external drift in geostatistic models (McBratney et al. 2003, Chap. 12 of this book).

Hybrid atmospheric correction algorithms and evaluation on VNIR/SWIR Hyperion satellite data for soil organic carbon prediction

ABSTRACT Visible near-infrared and shortwave infrared data acquired by spaceborne sensors contain atmospheric noise, along with target reflectance that may affect its end applications, e.g. geological, vegetation, soil surface studies, etc. Several atmospheric correction algorithms have been already developed to remove unwanted atmospheric components of a spectral signature of Earth targets obtained from airborne/spaceborne hyperspectral image. In spite of this, choosing of an appropriate atmospheric correction algorithm is an ongoing research. In this study, two hybrid atmospheric correction (HAC) algorithms incorporating a modified empirical line (ELm) method were proposed. The first HAC model (named HAC_1) combines (i) a radiative transfer (RT) model based on the concepts of RT equations, which uses real-time in situ atmospheric and climatic data, and (ii) an ELm technique. The second one (named HAC_2) combines (i) the well-known ATmospheric CORrection (ATCOR) model and (ii) an ELm technique. Both HAC algorithms and their component single atmospheric correction algorithms (ATCOR, RT, and ELm) were applied to radiance data acquired by Hyperion satellite sensor over study sites in Australia. The performances of both HAC algorithms were analysed in two ways. First, the Hyperion reflectances obtained by five atmospheric correction algorithms were analysed and compared using spectral metrics. Second, the performance of each atmospheric correction algorithm was analysed for prediction of soil organic carbon (SOC) using Hyperion reflectances obtained from atmospheric correction algorithms. The prediction model of SOC was built using partial least square regression model. The results show that (i) both the hybrid models produce a good spectrum with lower Spectral Angle Mapper and Spectral Information Divergence values and (ii) both hybrid algorithms provided better SOC prediction accuracy, in terms of coefficient of determination (R2), residual prediction deviation (RPD), and ratio of performance to interquartile (RPIQ), with R2 ≥ 0.75, RPD ≥ 2, and RPIQ ≥ 2.58 than single algorithms. HAC algorithms, developed using ELm technique, may be recommended for atmospheric correction of Hyperion radiance data, when archived Hyperion reflectance data have to be used for SOC prediction mapping.

Surface soil clay content mapping at large scales using multispectral (VNIR–SWIR) ASTER data

ABSTRACT The potential of Visible Near-Infrared and Short-Wave Infrared (VNIR–SWIR, 400 nm–2500 nm) hyperspectral imagery for use in multivariate approaches and geostatistical techniques for mapping topsoil properties has been previously demonstrated. However, the use of VNIR–SWIR hyperspectral imagery remains costly, which limits the spatial scales over which it can be applied. This paper aims to evaluate the potential for substituting the more accessible Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) VNIR–SWIR multispectral data for hyperspectral imagery in mapping surface soil clay contents. This study used ASTER multispectral data (nine bands in the VNIR–SWIR spectral domain) acquired over the Cap-Bon region in Tunisia (2000 km2) and 262 surface soil samples collected within the ASTER scene that were subjected to laboratory analysis of the clay fraction (soil particles less than 2 µm). The approach followed two steps: i) estimation of surface soil clay contents for bare soil areas using a Multiple Linear Regression (MLR) model built from the 9 ASTER VNIR–SWIR bands and ii) spatial interpolation (co-kriging) of the soil sampling of measured points and the ASTER-estimates over the whole study area. The MLR model for estimating clay contents using ASTER multispectral data performed correctly ( = 0.60). In addition, this performance is only slightly lower than that obtained using hyperspectral imagery (specifically, an Airborne Imaging Spectrometer for Applications (AISA-DUAL) dual hyperspectral sensor) in a previous study. Moreover, the co-kriging process appeared to yield encouraging results for capturing the large range of variability of clay content values, although it was not able to represent the short scale variability ( = 0.43). Finally, the ASTER multispectral data, despite being underused in the mapping of soil properties, may open up new ways to perform more extensive mapping of surface soil properties in semi-arid contexts characterized by extensive bare and dry soil surfaces.

Semi-blind source separation for estimation of clay content over semi-vegetated areas, from vnir/swir hyperspectral airborne data

The applicability of Visible, Near-Infrared and Short Wave Infrared (VNIR/SWIR) hyperspectral imagery for soil property mapping decreases when surfaces are partially covered by vegetation. The objective of this research was to develop and evaluate the performance of a “double-extraction” technique for clay content estimation over semi-vegetated surfaces using VNIR/SWIR hyperspectral imagery. The “double-extraction” technique consists of 1) an extraction of a soil reflectance spectrum ŝsoil, using a Semi-Blind Source Separation (SBSS) technique applied to couples of semi-vegetated spectra, and 2) an extraction of clay content from the soil reflectance spectrum ŝsoil, by a classical multivariate regression method. Semi-Blind source separation approach profited by known information about our context (presence of soil and green vegetation).

Clay contents predicted from hyperspectral VNIR/SWIR imagery, under different atmospheric conditions and spatial resolutions

Visible, Near-Infrared and Short Wave Infrared hyperspectral satellite imaging is one of the most promising tools for soil property mapping. The objective of this study was to test the sensitivity of soil property prediction results to atmospheric effects and to degradation in image spatial resolutions, to offer a first analysis of the potential of future hyperspectral satellite sensors for Soil applications (HYPXIM, PRISMA, Shalom, ENMAP and HyspIRI). Our results showed that (i) regression methods have robust performances from images from 5 to 30m and are inaccurate from images at 60 and 90m; (ii) when a correct compensation of the atmosphere effects is done, no differences are detected between the soil property maps retrieved from airborne imagery and the ones from spaceborne imagery; (iii) the spatial aggregation of the images induces a loss of the variance of the soil property prediction from 15 m of spatial resolution and a loss of information on soil spatial structures from 30 m of spatial resolution.