Landcare Research - Manaaki Whenua

Landcare-Research -Manaaki Whenua

A global spectral library for soils

Visible–near infrared (vis-NIR) soil spectroscopy is a relatively new technology for rapid prediction of some soil properties. The reflectance of light in the visible and infra-red range of the electromagnetic spectrum relates to the bonding and stretching vibrations of molecules in the soil – primarily molecules containing carbon, nitrogen, oxygen, and hydrogen atoms. As a result, reflectance spectra recorded from the soil surface (Fig. 1) can be used to predict properties, such as soil organic carbon content.

A reference set of spectra, with associated laboratory measurements, calibrates a prediction model for soil properties of interest such as carbon or clay contents. The model is then applied to predict soil properties solely from collected soil spectra. Some attributes, such as iron oxides can be predicted directly by reflectance at specific known wavelengths, whereas others, such as soil carbon, require multivariate statistical analysis of the spectral dataset.

Our soil spectroscopy research group at Landcare Research has been collaborating with international collaborators since 2008 to develop a global spectral library of soil Vis-NIR spectra. This collaboration was initiated by Raphael Viscarra Rossel, CSIRO, Canberra, who provided method guidelines and measurement protocols for consistent measurement of soil spectra by laboratories. Contributors were asked to provide a minimum set of analytical data, geographic location, and metadata with the collected spectra. The global database has spectra from 92 countries, representing seven continents, including spectra from soils in the World Soil Information (ISRIC) collection.

This global initiative is addressing the need for more soil data to improve our understanding of soil processes at scales ranging from regional to global. The global spectroscopic database was found to accurately estimate soil organic and inorganic carbon, extractable Fe, and fairly accurately estimate cation exchange capacity (CEC), clay, and silt content and soil pH. The global soil spectral library is the largest of its kind and provides a unique snapshot of global soil diversity.

Traditional methods of soil analysis are very time consuming and costly, limiting the number of samples that can be analysed. In contrast, the collection of Vis-NIR spectra is rapid (Fig. 2) so that many more estimates of soil property values can be derived for the same cost, to drive down the degree of associated uncertainty.

Acknowledgement: Funding provided by The New Zealand Government to support the objectives of the Livestock Research Group of the Global Research Alliance on Agricultural Greenhouse Gases.