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.

No abstract

No abstract