Statistiques spatio-temporelles sur les propriétés agronomiques des sols agricoles en France issues de la Base de Données d'Analyses de Terre (BDAT)


In France, farmers commission about 250,000 soil-testing analyses per year to assist them managing soil fertility. The number and diversity of origin of the samples make these analyses an interesting and original information source regarding cultivated topsoil variability. Moreover, these analyses relate to several parameters strongly influenced by human activity (macronutrient contents, pH...), for which existing cartographic information is not very relevant. Compiling the results of these analyses into a database makes it possible to re-use these data within both a national and temporal framework. A database compilation relating to data collected over the period 1990-2014 has been recently achieved. So far, commercial soil-testing laboratories approved by the Ministry of Agriculture have provided analytical results from more than 3,600,000 samples. After the initial quality control stage, analytical results from more than 1,900,000 samples were available in the database. The anonymity of the landholders seeking soil analyses is perfectly preserved, as the only identifying information stored is the location of the nearest administrative city to the sample site. We present in this dataset a set of statistical parameters of the spatial distributions for several agronomic soil properties. These statistical parameters are calculated for 4 different nested spatial entities (administrative areas: e.g. regions, departments, counties and agricultural areas) and for 5 time periods (1990-1994, 1995-1999, 2000-2004, 2005-2009, 2010-2014). Two kinds of agronomic soil properties are available: the first one correspond to the quantitative variables like the organic carbon content, and the second one corresponds to the qualitative variables like the texture class. For each spatial unit and temporal period, we calculated the following statistics sets: the first set is calculated for the quantitative variables and corresponds to the number of samples, the mean, the standard deviation and, the 2-,4-,10-quantiles; the second set is calculated for the qualitative variables and corresponds to the number of samples, the value of the dominant class, the number of samples of the dominant class, the second dominant class, the number of samples of the second dominant class.

Metadata Access
Creator Saby, Nicolas P.A. ORCID logo; Lemercier, Blandine; Arrouays, Dominique; Walter, Christian; Gouny, Laetitia; Swidersky, Chloé; Toutain, Benoît; Bispo, Antonio
Publisher Recherche Data Gouv
Contributor Saby, Nicolas; Paroissien, Jean-Baptiste; Chenu, Jean-Philippe; Millet, Florent; Infosol
Publication Year 2019
Rights info:eu-repo/semantics/restrictedAccess
OpenAccess false
Contact Saby, Nicolas (INRA - Institut National de la Recherche Agronomique)
Resource Type Dataset
Format text/tab-separated-values; text/plain
Size 3385953; 3201842; 3382897; 1385304; 2582972; 3130439; 122776; 119329; 121564; 41537; 108022; 114883; 1051086; 997285; 1036682; 408435; 759755; 972193; 28896; 28833; 28685; 9678; 26212; 28165; 9223
Version 3.1
Discipline Geosciences; Earth and Environmental Science; Environmental Research; Natural Sciences; Soil Sciences