From 1 - 1 / 1
  • Categories  

    The organic carbon content (in mass-%) map for the 0-10-cm soil layer of the United Republic of Tanzania was generated by means of digital soil mapping in a regression-kriging framework (‘Simple kriging with varying local means’) implemented in the Open Source Software R. Over 3,000 soil point observations were used to generate the map. Data sources were NAFORMA, Tanzania National Soil Survey, African Soil Profiles Database Version 1.1, and AfSIS. In addition a suite of environmental GIS data layers were used such as a land cover map, SOTER soil class map, maps of topographic attributes derived from the SRTM-DEM, maps of surface reflectance and vegetation indices derived from satellite imagery. The point observations were correlated to the environmental data layers using a linear regression model. This model was used to predict the carbon content at the nodes of a regular grid with 250 meter cell size. The regression residuals were kriged to the prediction grid nodes and added to the regression prediction to obtain the final prediction of carbon content. Map projection is UTM Zone 36S.The root mean square error, as determined through 10-fold cross-validation, is 0.89%. The map was produced by ISRIC - World Soil Information in a collaborative effort with the National Soil Survey, Ministry of Natural Resources and Tourism, Tanzania Forest Services, Sokoine University, and AfSIS.