The Use of major soil databases to reveal relationships between soil forming factors and global soil distribution

Relationships between various soil forming environmental factors and global soil distribution were examined using three large world soil databases, namely the ISRIC WISE Global database, the US National Soil Characterisation database and the NSW (Australia) SALIS database. Data on key soil properties and soil types under varying conditions of climate, parent material and topography were derived from the databases. There appears to be little or no previous attempts to use major soil databases in this way to reveal global soil distribution patterns. Distinct trends in the behaviour of soil properties with varying environmental conditions are observed. The relationships are, however, not statistically strong, with generally high standard errors and low correlation (R2) values. The observed relationships are mostly in accord with general theories of soil science and soil formation, but there are some anomalies and subtle complexities that are not readily explained and deserve further consideration. The nature of the relationships vary, with some such as pH v climate (under constant parent material and topography) suggesting a linear relationship while similar plots with base content reveal an exponential relationship. The gradients of regression lines and intersections with y-axes vary in systematic ways under different conditions. Not surprisingly, most soil properties and resulting soil types are influenced by the combined effect of the three environmental factors considered, with individual factors rarely controlling the behaviour of any of the selected soil properties. Many properties are, however, more strongly influenced by some factors than others. These results provide a potentially useful basis for the prediction and modelling of precise soil properties and soil types under different combinations of climate, parent material and topography around the globe. This approach may be particularly useful for areas where no reliable soil maps exist. The predictive potential of this approach is demonstrated with reference to soil conditions expected under a sub-tropical, near level site with a highly siliceous parent material. Further analysis of soil distribution relationships and the development of the predictive model using world soil databases are proposed.