ABSTRACT
New digital soil maps at a global scale with fine spatial resolution have been appearing. Further improvement of the quality of such maps is largely connected to the development of mapping methodology as well as a complete using of legacy data. We provided the results of a comparison of the SoilGRIDs built by ISRIC with the digital soil map of a part of Russia built using spatial disaggregation of the State Soil Map of Russia at 1:1 million scale. Overall, producer’s and user’s accuracies of soil classes prediction were calculated for SoilGRIDs (2 March 2017 version) using the disaggregated national soil map as a reference. Small producer’s accuracies were pointed out for Histosols (2%) and Fluvisols (30%) soil classes. Large areas (thousands of square kilometers) of Cambisols and Podzols instead of Albeluvisols and Phaeozems in SoilGRIDs were also recognized.
