Soil organic carbon assessment by field and airborne spectrometry in bare croplands: accounting for soil surface roughness
Introduction
The monitoring of soil attributes and their evolution over time as well as the development of pedological models rely on the availability of accurate and extensive soil data. The high spatial variability of soils arising from both local and global factors of soil formation requires generally collecting soil information from a very dense network of sites. Diffuse reflectance spectroscopic techniques, and in particular Visible, Near and Short Wave Infra Red (VNSWIR) spectroscopy (350 nm to 2500 nm), allows rapid sampling and instantaneous determination of many soil properties, at field and regional levels (in remote sensing mode). This technique can provide in a cost effective way the large quantity of spatial data required in soil monitoring or modelling studies like the monitoring of decline of soil organic matter in the topsoil.
Spectral libraries and multivariate modelling are often used to predict soil attributes of unknown samples. In the laboratory, such approach has proven to provide accurate determinations of SOC (Viscarra Rossel et al., 2006). When using the same approach, field spectroscopy and hyperspectral remote sensing, however, may fail to produce reliable and robust determinations due to uncontrolled measuring conditions and spatial variation in surface soil properties. According to Atzberger (2000), the main factors affecting the soil reflectance are soil water content, vegetation residues and surface roughness.
For the purpose of this study, field spectroscopic measurements were taken over bare and dry soils. Hence soil roughness remained the main disturbing factor and other influences have not been taken into consideration.
The effect of roughness on soil reflectance has been addressed in several studies. Arnfield (1975) showed that, for a relatively rough soil surface, soil albedo is generally lower than for a corresponding smooth surface due to self shadows. Atzberger (2000) simulated the influence of soil roughness on reflectance by using the SOILSPEC model. He found that, when the soil becomes smoother, due to decreasing micro-shadow effects, soil reflectance increases throughout the visible range.
Geometrical models have been developed to simulate bidirectional reflectance of light from rough soil surface based on the assumption that reflection is strongly correlated with the area of shadowed soil as well as on illumination and viewing geometry. Even though these models have been validated (Cierniewski, 1987, Cierniewski and Verbrugghe, 1997) their application in field conditions is not trivial, since many input parameters have to be considered which are quite difficult to assess in practical cases.
Several geometrical models (Cierniewski and Verbrugghe, 1994, Cooper and Smith, 1985, Irons et al., 1992, Norman et al., 1985) predict soil reflectance based on the assumption that shadowing of soil aggregates or clods has a greater influence than the scattering properties of a soil (Cierniewski and Verbrugghe, 1997). This study is also based on this principle and therefore the influence of roughness on soil reflectance is estimated by assessing its shadowing effects. This approach has been recently validated by (García Moreno et al., 2008).
The purpose of this study is to propose a new method to identify and reduce the effect of soil Relative Shadow (RS, the percentage of shadowed soil of the studied surface) on the assessment of SOC content from VNSWIR hyperspectral (350–2500 nm) field and airborne spectroscopic data.
First a method to measure RS and to correct its impact on field reflectance, measured with an Analytical Spectral Devices (ASD) spectroradiometer and the Airborne Hyperspectral Sensor (AHS), is proposed. Secondly the impact of RS on reflectance and SOCa accuracy is studied under laboratory conditions. Then SOC content is predicted by using uncorrected and corrected field reflectance values to evaluate the improvement in SOCa accuracy achieved with this method. The proposed method is finally compared with well-known mathematical pre-processings that intend to enhance SOCa accuracy.
Section snippets
Study area
The study area consisted of a north-south transect of ~ 7 km width and ~ 60 km length (NW corner: 50°03′N 6°03′E; SE corner: 49°33′N 6°12′E), crossing 4 of the 5 agro-geological regions of the Grand Duchy of Luxembourg.
The Grand Duchy of Luxembourg is characterised by a large variability of soils on a relatively small area (2586 km2). The four agro-geological regions in the study area are:
- •
The North, called Ardennes or Oesling, is a relatively homogeneous area consisting of plateaus and dissected
Method
The logic of the method aiming at enhancing SOCa by reducing RS effect is described below and consists of four steps:
- 1.
Computing shadow correction factors “K” for field reflectance from collected data
- 2.
Correcting field reflectance (ASD and AHS) with the “K” produced
- 3.
Estimating whether corrected reflectance improve SOCa
- 4.
Comparing the performance of the proposed correction method with existing methods
The global workflow of this study is presented in the Fig. 2 (steps 1–3).
Impact of soil Relative Shadow (RS) on reflectance and Soil Organic Carbon assessment (SOCa) accuracy based on laboratory data
Fig. 5 shows the reflectance decrease with increasing RS on one soil sample studied under laboratory conditions. Table 1 presents the impact of RS on SOCa accuracy. Generally increasing soil RS decreases SOCa accuracy. In particular the range of RS measured during the field campaign, as modelled in laboratory, decreases SOCa accuracy significantly. An increase of 20% of RMSECV is observed for the SOCa realised under the field RS range (RMSECV of 3.75) compared to the one realised under full
Discussion and conclusion
This study is based on the idea that the soil shadow, which is due to soil roughness in field conditions, decreases the accuracy of the SOCa from soil reflectance. This has been confirmed by laboratory experiments that showed that the accuracy of SOCa from prepared soil samples decreases when the RS level applied on these soil samples increases. As a result the purpose of this study was to find a method that reduces the RS effect on soil reflectance in order to enhance SOCa.
A new method for
Acknowledgements
The research in this paper is funded by the Belgian Science Policy Office and the National Research Fund of Luxembourg in the framework of the STEREO II program – Project “Monitoring soil organic carbon in croplands using imaging spectroscopy” (SR/00/110). We thank also the Vlaamse Instelling voor Technologische Onderzoek (VITO) at Mol (Belgium) for the organization of the flight campaign and geometric/atmospheric corrections of the images and Damien ROSILLON who elaborated part of the method
References (25)
A model for soil surface roughness influence on the spectral response of bare soils in the visible and near-infrared range
Remote. Sens. Environ.
(1987)- et al.
Shadow analysis: a method for measuring soil surface roughness
Geoderma
(2008) Laboratory, field and airborne spectroscopy for monitoring organic carbon content in agricultural soils
Geoderma
(2008)Measuring soil organic carbon in croplands at regional scale using airborne imaging spectroscopy
Geoderma
(2010)- et al.
Colour space models for soil science
Geoderma
(2006) - et al.
Predicting water content using Gaussian model on soil spectra
Remote Sens. Environ.
(2004) - et al.
Accounting for surface roughness effects in the near-infrared reflectance sensing of soils
Geoderma
(2009) A note on the diurnal, latitudinal and seasonal variation of the surface reflection coefficient
J. Appl. Meteorol.
(1975)Systematic evaluation of factors interfering with soil colour retrieval from space
- et al.
Standard normal variation transformation and de-trending of near-infrared diffuse reflectance spectra
Appl. Spectrosc.
(1989)
Infrared spectroscopic analyses on the nature of water in montmorillonite
Clay Clay Miner.
A geometrical model of soil bidirectional reflectance in the visible and near-infrared range
Cited by (36)
Estimation of agricultural soil surface roughness based on ultrasonic echo signal characteristics
2024, Soil and Tillage ResearchAssessing the effectiveness of ground truth data to capture landscape variability from an agricultural region using Gaussian simulation and geostatistical techniques
2021, HeliyonCitation Excerpt :These studies resulted in spatially mapped distribution of soil properties and quantified uncertainties and correlations in spatial models (Eze and Kumahor, 2019). The need for hyperspectral and high-spatial resolution satellite imagery continues to increase because of its potential for producing reliable maps of soil properties (Denis et al., 2014; Franceschini et al., 2015). In addition, information derived from remote sensing including crop and soil-related spectral indices, could provide sufficient information that is highly correlated with the target farm soil properties and characterize the spatial heterogeneity of the landscapes.
NASA's surface biology and geology designated observable: A perspective on surface imaging algorithms
2021, Remote Sensing of EnvironmentRapid appraisal using landscape sustainability indicators for Yaqui Valley, Mexico
2020, Environmental and Sustainability IndicatorsCitation Excerpt :A target could be application of conservation agriculture on all soils con sales and compactado so that management practices reduce risks for compaction and salinization. As remote sensing methods evolve, there may be opportunities to estimate soil organic matter (SOM) across landscapes, though presently these methods involve specialized airborne sensors and substantial calibration efforts (Denis et al., 2014; Hbirkou et al., 2012; Lacoste et al., 2014) and, thus, were not considered for the present rapid appraisal approach. Diverse management factors are known to influence soil SOM changes in the Yaqui Valley.
Hyperspectral remote sensing applications in soil: A review
2020, Hyperspectral Remote Sensing: Theory and Applications