Global predictions of primary soil salinization under changing climate in the 21st century

This dataset quantifies the annual average of naturally-occurring soil salinity to the depth of one meter from the soil surface in drylands of the world (aridity index ≤ 0.65) for the period between 1904 and 2100. This is realised by developing 4-D predictive models of the electrical conductivity of water-saturated soil paste extract (ECe). These machine learning-based predictive models predict ECe as target variable at different times and locations, enabling us to quantify different aspects of variability in primary soil salinization by the end of 21th century.