Soil is also a crucial factor to soil fertility and the environment because of its vast potential of
sequestration in soil and the biota.
The amount of organic stored in various soil pools at a specific time is the balance between the rate
of input of biomass and that of its mineralization in each of the organic pools. Change in the size of
the soil pool, therefore, can significantly affect atmospheric CO2 concentration.
This study an attempt was made to estimate soil organic and inorganic carbon stock through
Random forest based digital soil mapping technique.
The SOC and SIC densities up to 100 cm depth or paralithic contact were estimated for 1198 soil samples
located across India using a stratified random sampling that integrated land use, soil, topography and
agro-ecological regions.
Using Random forests (RF) based spatial prediction procedure with climatic, land cover, rock type,
soil type, multi-year NDVI, irrigation status as independent input variables, models for predicting
carbon density at 250 m spatial resolution were developed.
About 898 soil profile observations were used for modelling with RF algorithm and remaining 300
were used for validation. Geospatial modelling approach would help in monitoring soil carbon by
detecting changes in SOC and SIC stocks as a function of change in climate, land use and can be
linked to global models to understand the global carbon dynamics in a better way.