Predictive mapping of soil copper for site-specific micronutrient management using GIS-based sequential Gaussian simulation

Given that soil properties including micronutrient contents vary in space and time, geospatial mapping is necessary for robust site-specific management planning. Kriging interpolation is often used in GIS environments for mapping, but it has inherent limitation of having a smooth effects. In this study, sequential Gaussian simulation (SGS) was used to map the spatial distribution of Cu concentration in and model the spatial uncertainties for an arable dryland in central Botswana. The field was divided into 30 parallel lines in an NE–SW orientation. Soil samples were collected at intervals of 25 m on each line; and 35 sampling points were obtained. A total of 1050 soils sampled at a depth of about 20 cm were air dried and analyzed using an Olympus Delta Sigma^® portable X-ray fluorescence analyser. The average concentration of Cu (146 mg kg⁻¹) in the Maibele Airstrip of Botswana is higher than the global average copper concentration (30 mg kg⁻¹). Low copper contents were found dotted around the northern and southern edges of the study area, while high content zone is found in the interior. In comparison to kriging interpolation, SGS technique shows better performance. Kriging generally overestimated where lower values are probable and underestimated where higher values are probable. A large portion of the area has Cu content above the critical threshold of 125 mg kg⁻¹. Since total and not plant available Cu were the measured parameter, we recommend that further study in this location should focus on confirming the plant available Cu in the high-risk areas.