In 2009, within the framework of the European Land Use/Cover Area frame statistical Survey (LUCAS), a soil sampling campaign was implemented in 25 countries. About 22,000 composite topsoil samples were collected following a standardized sampling methodology and analysed in one laboratory. In this study, we present the first map of topsoil organic carbon (OC) content estimates for part of the European Union based on that comprehensive sampling programme. A digital soil mapping (DSM) by regression kriging (RK) approach was followed, and the covariates selected by the model were: land cover, elevation and slope, accumulated annual average temperature and the precipitation over potential evapotranspiration ratio, lithology, net primary productivity, and sand content. The results show high OC contents in northern latitudes and low contents in southern European countries which corroborates current expert knowledge. The overall model-fitting performance (R2) is 0.52 and the root mean squared error of the RK predictions equals 77 g C kg−1. Kriging of the regression residuals create hot-spots of OC content predictions on the map which are not believed to be realistic. It was concluded that different DSM techniques should be tested on the OC measurements data from the LUCAS database to try and improve the predictions and that validation against national datasets should be performed.

DE BROGNIEZ Delphine;  BALLABIO Cristiano;  VAN WESEMAEL Bas;  JONES R.J.A.;  STEVENS Antoine;  MONTANARELLA Luca; 

2014-09-17

Springer

JRC90442

978-3-319-04084-4,   

http://link.springer.com/chapter/10.1007%2F978-3-319-04084-4_39,    https://publications.jrc.ec.europa.eu/repository/handle/JRC90442,   

10.1007/978-3-319-04084-4_39,