Evaluation of Mid-Infrared and X-ray Fluorescence Data Fusion Approaches for Prediction of Soil Properties at the Field Scale

Abstract

Previous studies investigating multi-sensor fusion for the collection of soil information have shown variable improvements, and the underlying prediction mechanisms are not sufficiently understood for spectrally-active and -inactive properties. Our objective was to study prediction mechanisms and benefits of model fusion by measuring mid-infrared (MIR) and X-ray fluorescence (XRF) spectra, texture, total and labile organic carbon (OC) and nitrogen (N) content, pH, and cation exchange capacity (CEC) for n = 117 soils from an arable field in Germany. Partial least squares regression models underwent a three-fold training/testing procedure using MIR spectra or elemental concentrations derived from XRF spectra. Additionally, two sequential hybrid and two high-level fusion approaches were tested. For the studied field, MIR was superior for organic properties (ratio of prediction to interquartile distance of validation (RPIQV) for total OC = 7.7 and N = 5.0)), while XRF was superior for inorganic properties (RPIQV for clay = 3.4, silt = 3.0, and sand = 1.8). Even the optimal fusion approach brought little to no accuracy improvement for these properties. The high XRF accuracy for clay and silt is explained by the large number of elements with variable importance in the projection scores >1 (Fe ≈ Ni > Si ≈ Al ≈ Mg > Mn ≈ K ≈ Pb (clay only) ≈ Cr) with strong spearman correlations (±0.57 < rs < ±0.90) with clay and silt. For spectrally-inactive properties relying on indirect prediction mechanisms, the relative improvements from the optimal fusion approach compared to the best single spectrometer were marginal for pH (3.2% increase in RPIQV versus MIR alone) but more pronounced for labile OC (9.3% versus MIR) and CEC (12% versus XRF). Dominance of a suboptimal spectrometer in a fusion approach worsened performance compared to the best single spectrometer. Granger-Ramanathan averaging, which weights predictions according to accuracy in training, is therefore recommended as a robust approach to capturing the potential benefits of multiple sensors.

Optimization of sample preparation and data evaluation techniques for X‐ray fluorescence prediction of soil texture, pH, and cation exchange capacity of loess soils

“…For this, empirical corrections for soil moisture have proven helpful (Stockmann, Jang et al., 2016), but effects of poor detection of light elements (Na, Mg, Al, Si, P, and S)—which were highly important predictors in the current study—must be considered. The combination of laboratory XRF with visible/near‐ and/or mid‐infrared handheld spectrometers (Greenberg et al., 2023; Javadi & Mouazen, 2021) is a promising means to improve model accuracy by providing complimentary information about spectrally active organic molecules and soil minerals. Therefore, combining handheld infrared and portable XRF spectrometers could be a solution to achieving sufficient accuracy with field measurement.…”

“…(2023). Silicon (Si) contents, which cannot be measured using ICP‐OES due to loss during acid evaporation, were obtained via wavelength dispersive XRF on using a Malvern Panalytical Axios advanced spectrometer (Rh X‐ray tube) and the software SuperQ (version 4) at the University of Göttingen (Greenberg et al., 2023). For this, measurements were made on glass disks containing 0.42 g soil mixed with 4.2 g of A12 flux (66% di‐lithium tetraborate and 34% lithium metaborate).…”

“…The reference values for total elemental contents were obtained by acid digestion with HNO 3 , HCl, HClO 4 , and HF followed by ICP-OES, as described in Greenberg et al (2023). Silicon (Si) contents, which cannot be measured using ICP-OES due to loss during acid evaporation, were obtained via wavelength dispersive XRF on using a Malvern Panalytical Axios advanced spectrometer (Rh X-ray tube) and the software SuperQ (version 4) at the University of Göttingen (Greenberg et al, 2023). For this, measurements were made on glass disks containing 0.42 g soil mixed with 4.2 g of A12 flux (66% di-lithium tetraborate and 34% lithium metaborate).…”

Rapid Estimation of Soil Pb Concentration Based on Spectral Feature Screening and Multi-Strategy Spectral Fusion

“…Spectral fusion can be categorized into three stages: data-level fusion, feature-level fusion, and decision-level fusion [ 34 ]. Previous studies have shown that estimation models constructed using feature-level fusion, represented by OPA, and decision-level fusion, represented by GRA, outperform models based on data-level fusion [ 32 , 68 , 69 , 70 ]. OPA can effectively utilize the different properties and complementary information of XRF and vis-NIR spectra, thereby improving the prediction accuracy of soil heavy metal estimation models [ 33 ].…”

Carbon Farming: Bridging Technology Development with Policy Goals

“…Enhanced results have been achieved with data fusion of visible spectrum data, RGB digital camera data, and sentinel 2 bands with the use of machine learning [44]. Other fusion attempts were made between MIR and XRF and the resulting model results did not surpass the individual models [45]. On the other hand, the combination of Vis-NIR and the XRF model produced superior results in SOC estimation compared to the individual models [46].…”