Skip to main content

Assessment of soil carbon security using emerging techniques in hyperspectral imaging, X-ray florescence and pedometrics

Final Report Summary - CSECURE (Assessment of soil carbon security using emerging techniques in hyperspectral imaging, X-ray florescence and pedometrics)

To ensure the long-term sustainability of agricultural systems, soils must be actively managed to simultaneously deliver agricultural productivity and maintain soil quality. Ultimately, the objective for sustainable management of soils is dual purpose, with agronomic and environmental drivers. Sustainable management of soils that is based on soil organic matter management needs to advance to the point where best management practices deliver optimum agronomic crop yield. This also requires that soil organic carbon (SOC) sequestration per cropping cycle to be quantifiable in terms of its quantity and stability.

The fundamental nature of this research was to understand the role of small-scale spatial organisation of SOC within topsoil with respect to the long-term security of the abiotic carbon store. The research objective was to develop a new method for the assessment of SOC in intact soils was developed using a unique combination of state-of-the-art technologies laboratory based hyperspectral imaging and X-ray fluorescence (XRF) core scanning. The ground-breaking generation of high quality, high resolution spatial data at the soil pedon scale allows for the analysis of SOC dynamics to be studied in much greater detail. Quantitative maps of the spatial distribution of SOC throughout the soil pedon allow for a superior understanding of the agricultural management required to efficiently sequester SOC. This new ‘tool’ for the assessment of SOC at the profile scale is timely and compatible with the emerging discipline of digital morphometrics. There is a knowledge gap about SOC dynamics at the soil pedon scale that needs to be addressed in order for better integration of scientific knowledge across scales to deliver effective SOC management.

Specifically, laboratory based hyperspectral imaging captures information in the visible near-infrared (vis-NIR) region on a pixel-by-pixel basis of the acquired image. Chemometric techniques are employed to model spectral data with reference macro measurements of the chemical property of interest, in this case SOC. The predictive model is then used to predict SOC in each pixel of a hyperspectral image to generate a high resolution SOC concentration map. Here, the spectral data acquired in the hyperspectral soil images was modelled with SOC at the macro scale (i.e. standard agronomic soil sampling; macro samples of 10 cm depth interval, as well as a higher sampling resolution of 1 cm depth interval). To increase the accuracy of SOC prediction in the hyperspectral images, micro-scale XRF measurements, acquired at a similar spatial resolution as hyperspectral images, were introduced as reference measurements. Using pedotransfer functions (PTFs), predictive functions of ‘difficult-to-measure’ high resolution (
The potential impact of this work is the generation of high quality high-resolution spatial data at the soil pedon scale that allows multiple soil factors to be examined simultaneously (i.e. SOC, soil geochemistry) in much greater detail to assess the link between SOC and soil management. The findings of the proposed research will have direct impact in ensuring sustainable use of the soil resource and in tackling climate change. The potential to define agricultural management practices to achieve quantifiable SOC sequestration could pave the way for government SOC sequestration programs to combat current levels of atmospheric CO2. Evidence of mechanisms to achieve SOC stability may kick-start a voluntary “C-farming” culture. The positive side-effect is improved overall soil quality will ensure the long-term security of soils.

In addition to high-resolution spectral imaging and XRF (three-dimensional techniques), this study also examined a combinatorial approach to using conventional vis-NIR, mid-infrared (MIR) and XRF point spectroscopy (two-dimensional techniques). Many soil science laboratories are now equipped with technology platforms in portable vis-NIR and XRF spectrometers. These technologies have complementary capabilities, where XRF is known to accurately measure the soil’s inorganic element concentration, and vis-NIR has the ability to estimate the soil’s organic component and mineralogy suites. Data mining techniques were used to estimate soil properties from the vis-NIR spectra, and in a novel way from the XRF spectra. The prediction outcomes were combined into a single prediction outcome, using formal methods, called model averaging procedures. Combining model outcomes derived from spectra using model averaging techniques improved or maintained the prediction status of vis-NIR, MIR or XRF models. The accuracy of the prediction of a suite of soil geochemistry was much improved using this approach (the total number of well predicted elements increased from 15 to 25). Model averaging also improved the estimation of a range of soil properties of agronomic importance and was found to be suitable for soil pH, SOC, soil texture (sand and clay), CEC and total elements K, Mg, Co, Cr and Mn. Most notable is the large number of trace elements (As, Cd, Co, Cu, Hg, Mn, Ni and Zn) predicted to good or reasonable accuracy. When XRF is used in a conventional way to determine elemental concentrations it was demonstrated to be highly reliable for elemental concentrations present in high concentrations, but predictions of elemental content derived from XRF spectra was more effective for elements present in low concentrations. It is concluded that the synergistic use of portable vis-NIR and XRF spectral methods are well placed to replace traditional wet chemistry as a tool to permit large scale routine soil monitoring. This work is relevant to policy that aims to protect the soil resource by monitoring soil quality.