SoilSpec4GG is a USDA-funded Food and Agriculture Cyberinformatics Tools Coordinated Innovation Network. This project will bring together soil scientists, spectroscopists, informaticians, data scientists and software engineers to overcome some of the current bottlenecks preventing wider and more efficient use of soil spectroscopy. A series of working groups will be formed to address topics including calibration transfer, model choice, outreach & demonstration, and use of spectroscopy to inform global carbon cycle modeling.

Conventional procedures to determine soil characteristics are time and labor intensive, typically involving multiple processing, extraction and analysis steps. Soil spectroscopy offers a safe and rapid alternative (Figure 1). 

Figure 1. Schematic of soil spectroscopy

Soil spectroscopy is the measurement of light absorption when light in the visible, near infrared or mid infrared (Vis–NIR–MIR) regions of the electromagnetic spectrum is applied to a soil surface. The proportion of the incident radiation reflected by soil is sensed through Vis–NIR–MIR reflectance spectroscopy. These characteristic spectra (Figure 2) can then be used to estimate numerous soil attributes including:  minerals, organic compounds and water. 

 

Over the last three decades, soil spectroscopy has proven to be a cost- and time-effective, reproducible, nondestructive, and environmental-friendly technique that can be a viable alternative to conventional soil analysis methods. The technique has mainly been used in the laboratory; however, its application in-situ, from air, and from spaceborne sensors is also increasing. Nonetheless there remain challenges, Vis–NIR–MIR spectra include weak, wide, and overlapping absorption bands, which makes identifying specific soil parameters challenging.

Developing a robust connection between spectral patterns and soil properties requires data, a lot of it, to train various machine learning algorithms.  This presents a significant barrier to uptake of this technology. We believe that both the data and these trained models for soil spectroscopy data analysis need to be free, transparent and easy to access to users around the world. Open source data and algorithms will serve transparency, reproducibility, and inform better management decisions. Bringing together these open datasets and algorithms is a  main aim of the Soil Spectroscopy for the Global Good (SoilSpec4GG) project. SoilSpec4GG will enable soil scientists and stakeholders to apply Vis–NIR–MIR reflectance spectroscopy as a mature technique for routine soil analysis.