High resolution topsoil mapping using hyperspectral image and field data in multivariate regression modeling procedures

doi:10.1016/j.geoderma.2006.03.050

Geoderma

Volume 136, Issues 1–2, 1 December 2006, Pages 235-244

https://doi.org/10.1016/j.geoderma.2006.03.050 Get rights and content

Abstract

The spatial variability of within field topsoil texture and organic matter was studied using airborne hyperspectral imagery so as to develop improved fine-scale soil mapping procedures. Two important topsoil variables for precision farming applications, soil organic matter and soil texture, were found to be correlated with spectral properties of the airborne HyMap scanner. The percentage sand, clay, organic carbon and total nitrogen content could be predicted quantitatively and simultaneously by a multivariate calibration approach using either partial least-square regression (PLSR) or multiple linear regression (MLR). The different topsoil parameters are determined simultaneously from the spectral signature contained in the single hyperspectral image, since the various variables were represented by varying combinations of wavebands across the spectra. The methodology proposed provides a means of simultaneously estimating topsoil organic matter and texture in a rapid and non-destructive manner, whilst avoiding the spatial accuracy problems associated with spatial interpolation. The use of high spatial resolution and hyperspectral remotely sensed data in the manner proposed in this paper can also be used to monitor and better understand the influence of management and land use practices on soil organic matter composition and content.

Introduction

Topsoils frequently show a fine tessellated pattern and heterogeneity across fields indicated by e.g. color, roughness, infiltration, erosion and surface sealing phenomena. These characteristics cause differences in crop germination, nutrient and water uptake and thus markedly influence crop growth processes. This has implications for the pattern and the spatial extent of appropriate land use management practices and soil conservation strategies including site specific management in precision agriculture systems (Runge and Hons, 1998, Bullock and Bullock, 2000). To optimize crop growth, the various cropping practices, including soil tillage, seed bed preparation, fertilization and herbicide use must be adapted to the local topsoil properties. However, lack of high spatial resolution topsoil data is a serious limitation to the establishment of sub-field soil and crop management.

There is at present no effective way to map fine-scale soil heterogeneity so as to derive site specific data about topsoil physical/chemical characteristics. Several authors established relationships between soil spectral reflectance data and organic matter characteristics (Dalal and Henry, 1986, Ben-Dor et al., 1997, Ingleby and Crowe, 2000, Reeves et al., 2002, Udelhoven et al., 2003) and soil texture (Al-Abbas et al., 1972, Ben-Dor and Banin, 1995, Thomasson et al., 2001, Ben-Dor et al., 2002, Cozzolino and Morón, 2003). Soil attributes, soil texture and soil organic matter, all play an interdependent and decisive role in assessing topsoil characteristics e.g. soil aggregation, aggregate stability and resistance to water and wind erosion (Neemann, 1991). As a consequence it would be an advantage to be able to map both sets of physical characteristics from the one set of image data.

The aim of the work reported here was to develop a method of mapping fine-scale topsoil organic and texture parameters from a combination of field and hyperspectral image data. The work investigated the use of both multiple linear regression (MLR) and partial least-square regression (PLSR) to construct the model necessary to estimate the soil physical/chemical variables from the image data. Field data were combined with the image data for the construction of independent models derived using both MLR and PLSR. This innovative approach to digital soil mapping achieved gain from the simultaneous estimation of a suite of topsoil parameters. Additionally, the use of high spatial resolution remotely sensed data provides a more detailed pattern recognition of the soil's heterogeneity. Generally in soil mapping a soil data set is available that is restricted to only few sampled locations. The approach reported here avoids the attendant problems of accurate and reliable spatial prediction of spatial interpolation that has to deal with a small soil data set that does not uncover the fine tessellated pattern of soils. Moreover, as a methodology that is applicable over large areas at high spatial resolution this approach can be expected to support change detection research by mapping of past and future land use effects on soil organic matter accumulation and decomposition and contributes to sustainable and site adapted soil management.

Section snippets

Geographical survey, geology and terrain

The East German study area Wulfen (11°55′E, 51°49′N) is located in the federal state Sachsen–Anhalt, and stretches over 20 km from the district town Köthen in the south to the Elbe river in the north. It is characterized by three soilscapes. The southern part is a slightly undulated tertiary plain at 70 m altitude that is covered by a thin Loess layer up to 1.2 m deep. The northern part is the alluvial plain (glacial valley) of the river Elbe at 50 m altitude that served as the origin for the

Calibration procedure by PLSR

PLSR reduces the whole reflectance spectra to a few relevant factors and regresses them to the measured parameter of a given sample. While doing so, a calibration design is recommended that covers the whole range of possible values (Brereton, 2000). It is also reported that a certain redundancy within the spectra is useful to stabilize PLSR models against noise (Martens and Næs, 1989). When this is done, PLSR models are considered to be more robust than MLR calibration models. This might be

Conclusion

This is supposed to be the first study that employed the HyMap sensor and multivariate regression modeling for indirect topsoil parameter measurement. This remote sensing approach shows the potential benefits of using image data with carefully located in situ field data in digital soil mapping. The results also indicate that soil mapping procedures must be adapted to the soil parameter of interest and that multivariate calibration techniques allow calibration modeling as a generic procedure.

Acknowledgment

This paper draws on work from research projects funded by the German Federal Ministry of Education and Research and the Bavarian Research Foundation.

References (28)

E. Ben-Dor et al.
The reflectance spectra of organic matter in the visible, near infrared and short wave infrared region (400–2500 nm) during a control decomposition process
Remote Sens. Environ.
(1997)
L.S. Galvao et al.
Variations in reflectance of tropical soils: spectral–chemical composition relationships from AVIRIS data
Remote Sens. Environ.
(2001)
R. Adams et al.
Seeded region growing
IEEE Trans. Pattern Anal. Mach. Intell.
(1994)
A.H. Al-Abbas et al.
Relating organic matter and clay content to the multispectral radiance of soils
Soil Sci.
(1972)
M.F. Baumgartner et al.
Reflectance properties of soils
Adv. Agron.
(1985)
E. Ben-Dor et al.
Near-infrared analysis as a rapid method to simultaneously evaluate several soil properties
Soil Sci. Soc. Am. J.
(1995)
E. Ben-Dor et al.
Mapping of several soil properties using DAIS-7915 hyperspectral scanner data — a case study over clayley soils in Israel
Int. J. Remote Sens.
(2002)
R.G. Brereton
Introduction to multivariate calibration in analytical chemistry
Analyst
(2000)
D.S. Bullock et al.
From agronomic research to farm management guidelines: a primer on the economics of information and precision technology
Prec. Agric.
(2000)
T. Cocks et al.
The HyMap airborne hyperspectral sensor: the system, calibration and performance

D. Cozzolino et al.

The potential of near-infrared reflectance spectroscopy to analyse soil chemical and physical characteristics

J. Agric. Sci.

(2003)

R.C. Dalal et al.

Simultaneous determination of moisture, organic carbon, and total nitrogen by near infrared reflectance spectrophotometry

Soil Sci. Soc. Am. J.

(1986)

P. Dardenne

Stability of NIR spectroscopy equations

NIR News

(1996)

B. Deronde et al.

Sand dynamics along the Belgian coast based on airborne hyperspectral data and lidar data

EARSeL Proc.

(2004)

Cited by (220)

Deep learning techniques for hyperspectral image analysis in agriculture: A review
2024, ISPRS Open Journal of Photogrammetry and Remote Sensing
In recent years, there has been a growing emphasis on assessing and ensuring the quality of horticultural and agricultural produce. Traditional methods involving field measurements, investigations, and statistical analyses are labour-intensive, time-consuming, and costly. As a solution, Hyperspectral Imaging (HSI) has emerged as a non-destructive and environmentally friendly technology. HSI has gained significant popularity as a new technology, particularly for its promising applications in remote sensing, notably in agriculture. However, classifying HSI data is highly complex because it involves several challenges, such as the excessive redundancy of spectral bands, scarcity of training samples, and the intricate non-linear relationship between spatial positions and spectral bands. Notably, Deep Learning (DL) techniques have demonstrated remarkable efficacy in various HSI analysis tasks, including those within agriculture. As interest continues to surge in leveraging HSI data for agricultural applications through DL approaches, a pressing need exists for a comprehensive survey that can effectively navigate researchers through the significant strides achieved and the future promising research directions in this domain. This literature review diligently compiles, analyzes, and discusses recent endeavours employing DL methodologies. These methodologies encompass a spectrum of approaches, ranging from Autoencoders (AE) to Convolutional Neural Networks (CNN) (in 1D, 2D, and 3D configurations), Recurrent Neural Networks (RNN), Deep Belief Networks (DBN), Generative Adversarial Networks (GAN), Transfer Learning (TL), Semi-Supervised Learning (SSL), Few-Shot Learning (FSL) and Active Learning (AL). These approaches are tailored to address the unique challenges posed by agricultural HSI analysis. This review evaluates and discusses the performance exhibited by these diverse approaches. To this end, the efficiency of these approaches has been rigorously analyzed and discussed based on the results of the state-of-the-art papers on widely recognized land cover datasets. Github repository.
Prediction of soluble solid content in Nanfeng mandarin by combining hyperspectral imaging and effective wavelength selection
2024, Journal of Food Composition and Analysis
Soluble solid content (SSC) is among primary evaluation indicators of fruit quality and a key factor influencing consumer purchasing decisions. The research utilized hyperspectral imaging (380–1030 nm) to forecast SSC in Nanfeng mandarin. After a series of preprocessing methods, partial least squares regression (PLSR) and least squares support vector machine (LSSVM) were adopted to build prediction models. Combining multiplicative scatter correction and Savitzky-Golay smoothing was more effective compared to other preprocessing methods. Effective wavelengths (EWs) were selected by using bootstrapping soft shrinkage (BOSS), competitive adaptive reweighted sampling (CARS), iteratively retaining informative variables (IRIV) and their combinations. The BOSS-CARS-PLSR model performs optimally in prediction with the $R_{p}^{2}$ , RMSEP and RPD being 0.9376, 0.3986 and 4.0542, respectively. Additionally, the spatial distribution of the SSC in Nanfeng mandarin was visualized using the optimal model. Results show that combining hyperspectral imaging and EWs selection offers a rapid and intuitive approach that can non-destructively evaluate internal quality of Nanfeng mandarin.
A novel semi-empirical soil multi-factor radiative transfer model for soil organic matter estimation based on hyperspectral imagery
2023, Geoderma
Soil organic matter (SOM) concentration is an important factor affecting soil quality, and rapid and wide-scale monitoring of SOM concentration is a key step toward sustainable agriculture. Hyperspectral technology is widely used in soil composition monitoring, due to its rich spectral information. The complex imaging environment of hyperspectral imagery and the mixed pixel problem have led to the current applications of soil condition estimation mostly using data-driven methods. However, the estimation process based on data-driven methods cannot be explained by radiative transfer theory. Therefore, in this paper, a semi-empirical soil multi-factor radiative transfer (SESMRT) model combining a soil radiative transfer model and data-driven model is proposed for SOM estimation based on hyperspectral imagery. The radiative transfer model in the SESMRT model fully considers the soil components, such as SOM, soil moisture, soil iron oxides, and particle size distribution, and achieves high-precision simulation of soil spectra by resolving the mechanism of the influence of soil components on spectra. The SOM spectra calculated by the radiative transfer model are then used to estimate the SOM concentration based on a data-driven model. The SOM spectra eliminate the interference of other factors in the spectrum and significantly enhance the correlation between the spectrum and SOM. Thus, the SOM concentration can be estimated with a high degree of accuracy (R² = 0.660 ± 0.034, RMSE = 3.923 ± 0.236, RPIQ = 2.003 ± 0.118 for GF5 and R² = 0.685 ± 0.030, RMSE = 3.543 ± 0.311, RPIQ = 2.309 ± 0.207 for HyMap). Finally, the comparison of the estimated results for 2017 and 2019 shows that the SOM concentration of the cultivated soils in the study area increased, while the opposite is observed for the soils around the mining areas.
Prediction of fat content in salmon fillets based on hyperspectral imaging and residual attention convolution neural network
2023, LWT
The fat content of salmon is an essential indicator to evaluate its quality, and is related to its commercial value. An attention residual convolutional neural network (CNN), namely RACNN, which added channel attention and residual modules to the traditional CNN structure, was proposed to combine near-infrared hyperspectral imaging (NIR-HSI) (900-1700 nm) to predict fat contents in salmon fillets. Partial least squares regression (PLSR), least squares support vector machine (LSSVM) and CNN models were compared with the proposed RACNN models. Various spectrum preprocessing methods were used at the time. And Three feature extraction algorithms (successive projections algorithm (SPA), uninformative variable elimination (UVE), and competitive adaptive reweighted sampling (CARS)) were used to extract the effective wavelengths from the preprocessed NIR spectra. The experimental results showed that, the optimal PLSR and LSSVM models obtained coefficient of determination of prediction ( $R_{p}^{2}$ ) values of 0.8781 and 0.8952, respectively. The RACNN model based on full spectrum gained the highest prediction accuracies with $R_{p}^{2}$ = 0.9033, root mean square error of prediction (RMSEP) = 1.5143, and residual predictive deviation (RPD) = 3.2707. Therefore, the proposed RACNN model enables the rapid, non-destructive and accurate prediction of the fat content in salmon fillets without additional processing on raw spectra.
Enhancing Soil Mapping with Hyperspectral Subsurface Images generated from soil lab Vis-SWIR spectra tested in southern Brazil
2023, Geoderma Regional
The Remote Sensing (RS) products involving the visible, near-infrared and short-wave infrared (Vis–NIR-SWIR, simply Vis-SWIR) wavelengths are applicable for spatial prediction of soil properties, usually in the process called Digital Soil Mapping (DSM). The multispectral optical orbital sensors lack spectral resolution, while the Proximal Soil Sensing (PSS) offers many application possibilities. The limited spectral resolution and soil cover are the main challenging aspects for DSM. The objective of the study was to combine the soil lab Vis-SWIR spectra within DSM technique to create new hyperspectral images to use as covariates in a novel method of soil mapping. The advantages of the method were demonstrated throughout the spatial prediction of the Total Carbon (TC) content on soils of the upper part of Itatiaia National Park (INP), (Rio de Janeiro and Minas Gerais States, southeastern Brazil). The covariates applied were multispectral orbital bands from Sentinel 2, relief with its terrain derivations, geology and geomorphology. Using uppermost soil horizon lab Vis-SWIR spectra information from 72 points in the upper part of INP, 130 hyperspectral images were generated, in other words, Hyperspectral Subsurface Image. The validation methods were 8-fold cross-validation (CV) and External Validation (EV) running alongside within Random Forest algorithm, where randomness was locked. The DSM reaches an average CV coefficient of determination (R²) of 0.39 and Root Mean Square Error (RMSE) of 4.6, while creating a new model with a set of hyperspectral images gave a CV R² of 0.60 and RMSE of 4.06. The EV also presented an advantage for the second approach. The results support the method as the first strong integration between PSS and DSM, named by the authors as Hyperspectral Soil Mapping (HSM), which proved to be more efficient for mapping TC due to the pure soil lab Vis-SWIR spectra signal (free of atmospheric influence and vegetation cover). The method can be applied in dense or sparsely vegetated areas, for agricultural or conservation purposes, for various soil properties, with each and every wavelength.
Influence of arable land location incorporating its roughness on blue-sky albedo variation
2023, Journal of Quantitative Spectroscopy and Radiative Transfer
This paper quantifies the albedo variation in global bare, arable land, as one of the components of the Earth's land surface, that occurs over relatively short periods of the year but over a large total area. The study focused on the impact of the location of the land's dominant soil units, incorporating their roughness in their diurnal albedo (α) variation on the shortest (SDOY) and longest (LDOY) days of the year, and on their mean diurnal albedo (α_d) values throughout the year under clear-sky conditions. Specialized software was used to generate the aforementioned soil albedo values on the basis of soil reflectance spectra stored in a soil database. It was assumed that the air-dried soil surfaces being tested would present extremely different roughness states: very high, corresponding to those shaped by a plow (Pd), and very low, corresponding to a smoothing harrow (Hs). It was found that the farther the surfaces were from the equator, the higher the minimum α values would be on the SDOY, and the lower they would be on the LDOY. It was determined that the relative difference (AVα_d) between the highest and lowest values of α_d for the studied surfaces located within the latitude range of ±15° would be close to 0%. For latitudes within ±20°–±30°, the AVα_d values would be 6 and 5% for surfaces shaped by Pd and Hs, respectively. For those at about 65°, AVα_d values would even reach 63% and 54%, respectively. The α values of soil surfaces generated from many thousands of reflectance spectra collected in soil databases around the world, and taking into account their roughness, may be used in the future to model Earth's climate changes.

View all citing articles on Scopus

View full text

High resolution topsoil mapping using hyperspectral image and field data in multivariate regression modeling procedures

Abstract

Introduction

Section snippets

Geographical survey, geology and terrain

Calibration procedure by PLSR

Conclusion

Acknowledgment

Remote Sens. Environ.

Remote Sens. Environ.

Seeded region growing

IEEE Trans. Pattern Anal. Mach. Intell.

Relating organic matter and clay content to the multispectral radiance of soils

Soil Sci.

Reflectance properties of soils

Adv. Agron.

Near-infrared analysis as a rapid method to simultaneously evaluate several soil properties

Soil Sci. Soc. Am. J.

Mapping of several soil properties using DAIS-7915 hyperspectral scanner data — a case study over clayley soils in Israel

Int. J. Remote Sens.

Introduction to multivariate calibration in analytical chemistry

Analyst

From agronomic research to farm management guidelines: a primer on the economics of information and precision technology

Prec. Agric.

The HyMap airborne hyperspectral sensor: the system, calibration and performance

The potential of near-infrared reflectance spectroscopy to analyse soil chemical and physical characteristics

J. Agric. Sci.

Simultaneous determination of moisture, organic carbon, and total nitrogen by near infrared reflectance spectrophotometry

Soil Sci. Soc. Am. J.

Stability of NIR spectroscopy equations

NIR News

Sand dynamics along the Belgian coast based on airborne hyperspectral data and lidar data

EARSeL Proc.