Prediction of Soil Fertility Properties from a Globally Distributed Soil Mid‐Infrared Spectral Library

Abstract

All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Permission for printing and for reprinting the material contained herein has been obtained by the publisher. Prediction of Soil Fertility Properties from a Globally Distributed Soil Mid-Infrared Spectral Library Nutrient Management & Soil & Plant Analysis S oil chemical and physical information is needed to give advice on land management. Th is is especially true in developing countries, where soil diagnostic surveillance systems have been proposed to overcome data shortages (Shepherd and Walsh, 2007). Mid-infrared (MIR) diff use refl ectance spectroscopy is a reliable and fast soil analytical tool (Janik et al., 1998) that could form a basis for diagnostic surveillance systems. Soil properties are predicted either by direct absorption of the light associated with functional groups (properties such as organic C, total N, or clay composition; Van der Marel and Beutelspacher, 1976) or by correlation to such properties and the mineral composition of the soil (properties such as cation exchange capacity [CEC] and soil texture). New samples can be predicted only if they fall within the property range of the calibration set (Naes et al., 2002). In many situations, a rapid and approximate estimate of soil chemical and physical properties is adequate, and resources for an elaborate analysis may not be available. A global calibration may meet this purpose. Some studies have tested soil infrared spectroscopy on diverse data sets at the regional scale. Reeves and Smith (2009), working with a North American library of 720 samples, came to the conclusion that neither MIR nor near-infrared (NIR) spectra yielded suitable calibrations even for organic C. Th ey attributed the poor performance to the extreme sample diversity in parent material, land

Mid-Infrared and Total X-Ray Fluorescence Spectroscopy Complementarity for Assessment of Soil Properties

“…The bottoms of the Al wells are roughened to reduce specular reflectance. Soil samples were loaded into four replicate wells, each scanned 32 times, and the four spectra averaged to account for within-sample variability and differences in particle size and packing density, as described by Terhoeven-Urselmans et al (2010).…”

Pearl Millet and Sorghum Yield Response to Fertilizer in the Sahel of Burkina Faso

“…The air-dried samples were sent to the World Agroforestry Center Soil-Plant Spectral Diagnostic Laboratory in Nairobi, Kenya https://www.worldagroforestry.org/sd/landhealth/soil-plant-spectral-diagnostics-laboratory/so ps. The analysis was with mid-infrared spectral analysis, complemented by wet chemistry analysis of about 10% of the samples for calibration of the spectral analysis (Shepherd and Walsh, 2007;Terhoeven-Urselmans et al, 2010;Towett et al, 2015). Organic C and N were determined with a Thermal Scientific Flash 2000.…”

Landscape-scale variability of soil health indicators: effects of cultivation on soil organic carbon in the Usambara Mountains of Tanzania

“…Soil samples that have both MIR spectra and associated wet chemistry data were used to develop the MIR prediction models for the various soil properties (Vågen et al 2016). Processing of the MIR spectra followed the procedures outlined in Terhoeven- Urselmans et al (2010), with first derivatives computed using a Savitsky-Golay polynomial smoothing filter implemented in the locpoly function of the KernSmooth R package (Wand 2015) prior to fitting a random forest (RF) prediction model to the samples from the study area. Random forest modeling is an ensemble modeling approach, where many weak learners (decision trees) are combined or bagged to predict an outcome, SOC in this case (Breiman 2001).…”