Soil organic matter has an important role in the modulation of the atmospheric CO2. For the calculation of carbon budgets, its dense and accurate monitoring is required. Analyses directly in the field would potentially gear up the number of samples analysed substantially. The overall goal of ProbeField is a best practice protocol for in-field proximal soil sensing.
By | J. Cobos, S. Madenoğlu, M. Knadel, J. A. Cayuela, B. Stenberg
Proximal soil sensors can be used for quick and simple in-field soil sensing. These devices are used as hand-held sensors or carried by on-ground vehicles. Proximal indicates sensors operating in proximity or direct contact to the soil. A few meters distance can be regarded as a maximum. Proximal sensors come in a large variety and include mechanical, electrochemical, electrical and electromagnetic, optical and radiometric sensors. Each of these groups have their advantages and limitations and by far the optical group is the largest. Among these techniques, visible and near infrared (~350-2500nm) spectroscopy (vis-NIRS) has many commercial uses. Analyses related to agriculture food and feed are well established. Since the mid-90th the interest in soil analyses with vis-NIRS has increased rapidly. Reasons for this are many. The technique is relatively simple to use and very quick. A spectrum can be acquired within seconds as well as estimates of constituents. The possibility to use fibre optics allows a very flexible sample presentation and robust instruments can be built for on- and at-line analyses. Vis-NIR soil spectroscopy is identified as one of the proximal sensor techniques with most information about organic matter and clay minerals. In addition to these, vis-NIR is suggested as a predictor for a number of soil components and properties.
Spectroscopic measurements in the vis-NIR range have a very good potential for in-field measurements due to the simplicity, robustness, and flexibility. There are also several commercial and semi commercial instruments available for point and on-the-go-analyses. Despite this, the vast majority of research and applications has focused on lab analyses on dried and sieved samples leaving several knowledge gaps for the in-field applications. A thorough effort to manage the combined challenges of measurements directly in the field and evaluation of the so gathered data remains.
Quick and simple soil analyses directly in the field through proximal sensing has the potential to substantially gear up the number of samples analysed. With focus on visible and near infrared spectroscopy (Vis-NIRS) ProbeField will work to make this happen. The Vis-NIR technique has many advantages required for field analyses of soil properties. There are, however, drawbacks to be overcome. In contrast to spectroscopy in the lab on prepared samples, variable soil moisture and structure in the field will hamper reliability of analyses. ProbeField will test and suggest physical and mathematical procedure to manage these problems. A wide range of soil properties will be analysed, and 3D mapping will be performed to estimate for example carbon stocks.
Quantitative soil attributes cannot be directly read from vis-NIR spectra. Instead, spectra are related to a reference method, typically a standard laboratory analysis, and described in a prediction model that can be used to estimate the corresponding properties of new, unknown, samples.
Over the last decade several soil spectral libraries (SSL’s) have been built at the regional or larger scale for research and the calibration of general prediction models. The rationale is to provide generic and robust models over large areas characterized by large diversity and in thus benefit from this efficient analytical technique.
In ProbeField our aim is to use existing SSL’s and apply them on spectra collected in the field. Since these libraries are predominantly based on laboratory spectra with sieved and dried soil several challenges have to be overcome. One obvious challenge is to apply such lab based spectral libraries on field spectra. Due to moisture and other effects, lab based SSL calibrations are not directly applicable to field spectra. There are, however, examples of spectral transformations that reduce the effects of water and structure and there are also examples of sample presentation procedures in the field that perform better than others. By more thoroughly investigating procedures to make field spectra as similar to lab spectra as possible through novel sample presentation techniques in combination with spectral transformation, we hypothesise will bring field and lab spectra closer to one another. We also hypothesise that sophisticated mathematical procedures will further eliminate the difference between lab and field spectra and enable reliable in-field estimates based on lab based SSL’s.
Merging current spectral libraries to create spectral libraries with bigger geographical and soil type coverage is desirable but in addition to spectral bias between SSL’s differences in the reference methods used must also be considered. ProbeField will cooperate with the GLOSOLAN-spec, IEEE -SA, WorldSoil and or JRC-LUCAS initiatives, which are closely related to this work.
Go to the project website, click here, and read more on the overall
Contact for more information:
Coordinator of ProbeField: Dr. Bo Stenberg (Bo.Stenberg@slu.se)
Project Communication representative: Dr. Maria Knadel (firstname.lastname@example.org).
For more information go to: https://ejpsoil.eu/soil-research/probefield
Joaquín Cobos, Sevinç Mdenoglou, Maria Knadel, José A. Cayuela, Bo Stenberg.