Identification of high-performing soil groups in grazing lands using a multivariate analysis method

Abstract

Understanding and quantifying the complex relationships between soil properties and vegetation health is important for sustainable land management and optimising agricultural productivity. This study tested a spatial data-driven framework to identify the soil groups associated with pasture health using publicly available gridded soil attribute layers from Soil and Landscape Grid of Australia (SLGA) over two adjacent southeast Australian river catchments. Principal component analysis (PCA) followed by isocluster unsupervised classification was applied to seventeen SLGA soil attribute layers to identify dominant soil patterns, which showed good spatial agreement with the Enhanced Vegetation Index (EVI) data derived from Landsat-8 imagery. The soil class demonstrating the highest EVI values (HVR class) and the lowest EVI values (LVR class) were determined. A comparison of these classes with soil types defined in the New South Wales Soil Landscape maps confirmed that the HVR class is predominated by agriculturally productive, basalt-derived 'Ant Hill' soils. The cation exchange capacity (CEC) ranging from ∼10 to 20 meq/100 g, clay content ranging from 20 % to 30 %, pHc (pH in calcium chloride solution) between 5 and 6, pHw (pH in water) between 5.5 and 6.5 and SOC between 2.5 % to 4 % were associated with higher EVI values in the HVR soil class within the study area. This study demonstrated an effective framework for identifying key soil attributes that affect agricultural productivity and pinpointing critical locations for grazing land productivity.