Information depth of NIR/SWIR soil reflectance spectroscopy
Introduction
Recent advances in proximal sensing of soils have provided rapid and cost-effective methods for retrieving various soil properties in laboratory settings. For example, significant correlations between soil reflectance spectra in the optical domain (300–2500 nm) and soil mineralogy and chemical composition (Clark et al., 1990; Kruse et al., 2003; Mulder et al., 2013; Omran, 2017), soil organic matter content (Daniel et al., 2004; He et al., 2009; Ingleby and Crowe, 1999; Leue et al., 2017; Yuan et al., 2020), soil water content (Lobell and Asner, 2002; Sadeghi et al., 2015; Whiting et al., 2004; Zeng et al., 2016), soil particle size distribution (Bänninger et al., 2006; Hermansen et al., 2017; Sadeghi et al., 2018a), and soil hydraulic properties (Babaeian et al., 2015a; Babaeian et al., 2015b) have led to the development of promising data-driven or physics-based methods for soil property estimation.
Despite salient success of soil characterization with proximal sensing techniques, an inherent limitation of these methods is the shallow light penetration into the soil. In addition, the soil information depth, i.e., the maximum soil depth contributing to the measured signal, is not well understood. Consequently, it is challenging to determine the soil volumes that the estimated properties are representative for, which is key for many soil studies.
Light penetration into a particulate opaque medium such as soil is governed by two processes, light absorption and light scattering by particles. Models that explain these processes are mainly derived from approximations of the radiative transfer equation (Philips-Invernizzi et al., 2001). A main challenge of these models, when applied to natural soils, is their calibration for the soil optical properties. This requires information of not only the reflectance (i.e., reflected light intensity at the surface), but also the light intensity at a specified depth of the medium (Kortüm, 2012), which is challenging to measure in soils (Tian and Philpot, 2018; Baranoski et al., 2019). In addition, the inversion of some of the models, e.g., the beam-tracing model of Bänninger et al. (2006), are computationally - very demanding.
A number of researchers have previously attempted to indirectly determine the light penetration depth in soils via studying light-dependent soil chemical or biological activities at various soil depths (Balmer et al., 2000; Hebert and Miller, 1990) or germination of light-sensitive seeds planted at different depths (Benvenuti, 1995; Bliss and Smith, 1985; Woolley and Stoller, 1978). Ciani et al. (2005) were among the first to provide direct estimates of the information depth in the visible light range (300–700 nm). In this approach, the Kubelka-Munk two-flux radiative transfer model or its variants (Kubelka and Munk, 1931; de la Osa et al., 2020) can be used to relate the information depth to the soil optical properties determined with a dilution method, where the measured reflectance spectra of various soil mixtures with a white reference powder (e.g., barium sulphate) having known optical properties are analyzed. Although this method is computationally efficient, showing great potential for soil information depth estimation, it requires laborious soil measurements, and therefore is not feasible for large soil datasets.
In this paper we propose a novel physics-based analytical model that provides new insights about the information depth of NIR/SWIR soil spectroscopy. While the model is not capable of explicitly quantifying information depth, it estimates the contributions of surface reflectance (i.e., reflectance originating from the first layer of particles at the surface) and volume reflectance (i.e., reflectance originating from the underlying soil layers) to the total soil reflectance. The new model is based on the radiative transfer model introduced in Sadeghi et al. (2018a), and it is proposed for longer optical wavelengths (i.e., 700–2500 nm), which are of interest for several soil properties such as water content (Sadeghi et al., 2015) or particle size distribution (Sadeghi et al., 2018a). The proposed model requires easy-to-measure basic soil information including reflectance spectra and particle size distributions, and hence, it may be efficiently applied to large soil datasets.
In the next section, we briefly review the Sadeghi et al. (2018a) model for the reflectance-particle size functional relationship, R(D), which is valid for uniform soils (i.e., soils with a narrow particle size distribution). We then extend its applicability to nonuniform soils to allow partitioning of soil reflectance spectra into surface and subsurface contributions. In Section 3, we present experimental soil reflectance spectra and particle size distribution (PSD) data used for model evaluation. In Section 4, we introduce and discuss modeling results that unravel the contributions of surface and volume components to the total soil reflectance. A summary and conclusions are provided in Section 5.
Section snippets
Uniform soil model
Sadeghi et al. (2018a) developed a physics-based analytical model for elucidating the inverse correlation between soil particle/aggregate size and reflectance, R. In this model, the total reflectance is calculated by coupling a surface reflectance model, derived from Stokes' (1862) solution, and a volume reflectance model that tracks the light path through a hypothetical soil with a distinct structure. The exact definitions for surface and volume reflectance are provided in Fig. 1, Fig. 3 of
Experimental data
To evaluate the performance of the proposed model, we relied on six well-characterized Arizona source soils, including AZ4B (loamy sand), AZ7 (sandy loam), AZ11 (loam), AZ13 (sandy clay loam), AZ15 (silt loam), and AZ18 (clay), covering a wide range of textures and mineral compositions (see Table 2 and Fig. 4 in Sadeghi et al., 2018a). Two major soil properties used are the soil PSDs and reflectance spectra.
Soil optical properties
Fig. 3 depicts a comparison of the ρ and k soil optical properties determined with the proposed new inverse method and the reference method. As evident, the inverse method was capable of effectively retrieving the optical soil properties without relying on the additional R(D) relationship required for the reference method. This success is attributable to the fact that ρ and k are interrelated, and hence, a nearly unique combination of ρ and k is expected to match Eq. (6) with the measured
Conclusions
A new model for quantification of the surface and volume reflectance contributions to the total reflectance was introduced for the NIR/SWIR bands of the electromagnetic spectrum. The model provides new insights about the soil information depth of optical soil spectroscopy. In addition to the reflectance spectra measurements, information about the soil particle size distribution is required for application of the new method. The soil optical properties ρ and k, which are obtained via inverse
CRediT authorship contribution statement
Sarem Norouzi: Conceptualization, Methodology, Data curation, Formal analysis, Writing - original draft. Morteza Sadeghi: Conceptualization, Methodology, Writing - review & editing. Abdolmajid Liaghat: Supervision, Writing - review & editing. Markus Tuller: Conceptualization, Data curation, Resources, Writing - review & editing. Scott B. Jones: Resources, Writing - review & editing. Hamed Ebrahimian: Supervision, Writing - review & editing.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
The authors gratefully acknowledge funding from the United States Department of Agriculture (USDA) – National Institute of Food and Agriculture (NIFA) grant #2020-67019-31028, and from the USDA NIFA Hatch/Multi-State project #ARZT-1370600-R21-189. We would like to thank Mohaddese Effati and Mohammad R. Gohardoust for collecting the PSD data.
References (39)
- et al.
Estimation of complex refractive index of soil particles and its dependence on soil chemical properties
Remote Sens. Environ.
(1991) - et al.
Characterizing regional soil mineral composition using spectroscopy and geostatistics
Remote Sens. Environ.
(2013) - et al.
A linear physically-based model for remote sensing of soil moisture using short wave infrared bands
Remote Sens. Environ.
(2015) - et al.
Particle size effects on soil reflectance explained by an analytical radiative transfer model
Remote Sens. Environ.
(2018) - et al.
Predicting water content using Gaussian model on soil spectra
Remote Sens. Environ.
(2004) - et al.
A physicoempirical model to predict the soil moisture characteristic from particle-size distribution and bulk density data 1
Soil Sci. Soc. Am. J.
(1981) - et al.
Towards retrieving soil hydraulic properties by hyperspectral remote sensing
Vadose Zone J.
(2015) - et al.
A comparative study of multiple approaches for predicting the soil–water retention curve: Hyperspectral information vs. basic soil properties
Soil Sci. Soc. Am. J.
(2015) - et al.
Photolytic transformation of organic pollutants on soil surfaces an experimental approach
Environ. Sci. Technol.
(2000) - et al.
Modeling light scattering at soil surfaces
IEEE Trans. Geosci. Remote Sens.
(2004)
Modelling the effect of particle size, shape and orientation of light transfer through porous media
Eur. J. Soil Sci.
On the light penetration in Natural Sands
Quantitative remote sensing of soil properties
Adv. Agron.
Soil light penetration and dormancy of jimsonweed (Datura stramonium) seeds
Weed Sci.
Penetration of light into soil and its role in the control of seed germination
Plant Cell Environ.
Light penetration in soil and particulate minerals
Eur. J. Soil Sci.
High spectral resolution reflectance spectroscopy of minerals
J. Geophys. Res. Solid Earth
Analysis of VNIR (400–1100 nm) spectral signatures for estimation of soil organic matter in tropical soils of Thailand
Int. J. Remote Sens.
Attenuation measurements of monochromatic light in marine-sediments
Oceanol. Acta
Cited by (17)
Electrochemical framework for dynamic tracking of Soil Organic Matter
2024, Biosensors and Bioelectronics: XTopsoil vertical gradient in different tillage systems: An analytical review
2024, Soil and Tillage ResearchA new variant of the optical trapezoid model (OPTRAM) for remote sensing of soil moisture and water bodies
2023, Science of Remote Sensing