Mapping soil clay contents in Dutch marine districts using gamma-ray spectrometry

Abstract

Conventional soil sampling methods to obtain high-resolution soil data are labour intensive and costly. Recently, gamma ray spectrometry has emerged as a promising technique to overcome these obstacles. The objective of our study was to investigate the prediction of soil clay contents using gamma-ray spectrometry in three marine clay districts in the Netherlands: the southwestern marine district (SMD), the IJsselmeerpolder district (IJPD) and the northern marine district (NMD). The performance of linear regression models was investigated at field (<1 km 2 ), regional (1-1000 km 2 ) and district (>1000 km 2 ) scales and for all the Dutch marine districts together. For this study, a database was available with 1371 gamma-ray spectra measured on arable fields in marine clay districts during the period 2005-2008 and these were all linked to laboratory analyses of clay contents. At the field scale, linear regression models based on 40 K, 232 Th, or a combination of these revealed much smaller root mean squared error (RMSE) values (2-3%) compared with a model based on the field mean (8-10%). At the district scale, the regression models for the SMD and IJPD, which have comparable sediments, performed better than for the NMD. This indicates that the prediction of clay contents in late Holocene marine sediments may be made with gamma-ray spectrometry provided that the origin of the parent material results in a unique fingerprint. Because of the heterogeneous parent material of all marine districts taken together, our study shows that no unique and precise fingerprint exists, and the RMSE of 6% between clay contents and gamma-ray spectra is not much different from the RMSE of 7% when using the overall mean as a predictor.

Adaptation of regional digital soil mapping for precision agriculture

“…High-resolution gamma-ray sensing from vehicles on the ground, e.g. in northern Europe, has shown that the thorium isotope ( 232 Th) is highly correlated to soil clay content (Van der Klooster et al 2011;Piikki et al 2013). Detailed research carried out previously within the Skara study area indicates that 232 Th derived from gamma-ray sensing is superior to many other soil sensor data in prediction modelling of soil clay concentration (Piikki et al 2013).…”

“…In digital soil mapping, the relationship between a number of soil properties and gamma-ray data has been successfully used for predicting e.g. weathering, soil parent material, soil texture and soil cadmium concentration (Cook et al 1996;Wilford et al 1997;Wong and Harper 1999;Pracilio et al 2006;Rawlins et al 2007; Van der Klooster et al 2011;Söderström and Eriksson 2013). In addition, vehicle-mounted equipment for gamma-ray measurements has recently been developed and can be used for meticulous gamma-ray scanning (Van Fig.…”

Relationships between gamma-ray attenuation and soils in SW England

“…The subject has a considerable historical pedigree (Talibudeen, 1964;Hyvönen et al, 2005) in terms of early studies. Ground-based studies at the field scale, using soil-sampling to control interpretation, have defined relationships between the measured radioelements and soil properties such as clay content (Van der Klooster et al, 2011), soil pH (Dickson and Scott, 1967), SOC (Wong and Harper, 1999), plant available potassium (Wong and Harper, 1999;Dierke and Werban, 2013) and CEC (Rodrigues et al, 2015). More extensive airborne data sets may also be used in soil property assessments (Cook et al, 1996) although control information is inevitably more sparse.…”

A selection of sensing techniques for mapping soil hydraulic properties

“…• sand, silt and clay content: near and mid infrared spectrometry (Lagacherie et al, 2008;Viscarra Rossel et al, 2009), γ-ray spectroscopy (Pracilio et al, 2006;Viscarra Rossel et al, 2007;Van der Klooster et al, 2011), electromagnetic induction methods (Triantafilis and Lesch, 2005;Cockx et al, 2009) • organic matter content: near and mid infrared spectrometry (Shonk et al, 1991;Ben-Dor et al, 2008) • water content: near and mid infrared spectrometry (Ben-Dor et al, 2008) Validation results of predictions of median grain size of sand fraction (M50) and bulk density by proximal sensing techniques were not found. This overview only focuses on scientifically published studies that included a sound validation of results.…”

“…The RMSE in predicted clay content could be reduced from about 11 dag kg −1 when an areal mean was used as predictor to about 6 dag kg −1 when linear regression models and γ-ray spectrometry were used in spatial prediction. Van der Klooster et al (2011) predicted clay content in the Dutch marine clay area from γ-ray data using linear regression models with the following general structures:…”