Andrew Sila

Mapping Soil Properties of Africa at 250 m Resolution: Random Forests Significantly Improve Current Predictions

80% of arable land in Africa has low soil fertility and suffers from physical soil problems. Additionally, significant amounts of nutrients are lost every year due to unsustainable soil management practices. This is partially the result of insufficient use of soil management knowledge. To help bridge the soil information gap in Africa, the Africa Soil Information Service (AfSIS) project was established in 2008. Over the period 2008–2014, the AfSIS project compiled two point data sets: the Africa Soil Profiles (legacy) database and the AfSIS Sentinel Site database. These data sets contain over 28 thousand sampling locations and represent the most comprehensive soil sample data sets of the African continent to date. Utilizing these point data sets in combination with a large number of covariates, we have generated a series of spatial predictions of soil properties relevant to the agricultural management—organic carbon, pH, sand, silt and clay fractions, bulk density, cation-exchange capacity, total nitrogen, exchangeable acidity, Al content and exchangeable bases (Ca, K, Mg, Na). We specifically investigate differences between two predictive approaches: random forests and linear regression. Results of 5-fold cross-validation demonstrate that the random forests algorithm consistently outperforms the linear regression algorithm, with average decreases of 15–75% in Root Mean Squared Error (RMSE) across soil properties and depths. Fitting and running random forests models takes an order of magnitude more time and the modelling success is sensitive to artifacts in the input data, but as long as quality-controlled point data are provided, an increase in soil mapping accuracy can be expected. Results also indicate that globally predicted soil classes (USDA Soil Taxonomy, especially Alfisols and Mollisols) help improve continental scale soil property mapping, and are among the most important predictors. This indicates a promising potential for transferring pedological knowledge from data rich countries to countries with limited soil data.

Pre-treatment of soil X-ray powder diffraction data for cluster analysis

X-ray powder diffraction (XRPD) is widely applied for the qualitative and quantitative analysis of soil mineralogy. In recent years, high-throughput XRPD has resulted in soil XRPD datasets containing thousands of samples. The efforts required for conventional approaches of soil XRPD data analysis are currently restrictive for such large data sets, resulting in a need for computational methods that can aid in defining soil property – soil mineralogy relationships. Cluster analysis of soil XRPD data represents a rapid method for grouping data into discrete classes based on mineralogical similarities, and thus allows for sets of mineralogically distinct soils to be defined and investigated in greater detail. Effective cluster analysis requires minimisation of sample-independent variation and maximisation of sample-dependent variation, which entails pre-treatment of XRPD data in order to correct for common aberrations associated with data collection. A 24 factorial design was used to investigate the most effective data pre-treatment protocol for the cluster analysis of XRPD data from 12 African soils, each analysed once by five different personnel. Sample-independent effects of displacement error, noise and signal intensity variation were pre-treated using peak alignment, binning and scaling, respectively. The sample-dependent effect of strongly diffracting minerals overwhelming the signal of weakly diffracting minerals was pre-treated using a square-root transformation. Without pre-treatment, the 60 XRPD measurements failed to provide informative clusters. Pre-treatment via peak alignment, square-root transformation, and scaling each resulted in significantly improved partitioning of the groups (p < 0.05). Data pre-treatment via binning reduced the computational demands of cluster analysis, but did not significantly affect the partitioning (p > 0.1). Applying all four pre-treatments proved to be the most suitable protocol for both non-hierarchical and hierarchical cluster analysis. Deducing such a protocol is considered a prerequisite to the wider application of cluster analysis in exploring soil property – soil mineralogy relationships in larger datasets.

Estimation of stability-related soil properties using mid-infrared spectroscopy for non-agricultural land use in Lake Victoria basin, Kenya

ABSTRACT Methods for rapid and accurate soil tests are needed for stability-related soil properties, which are valuable for non-agricultural land use. The study tested the applicability of mid-infrared (MIR) spectroscopy for rapid estimation of selected stability-related soil properties. Two sample sets, representing different soils from across the Lake Victoria basin of Kenya, were used for the study. A model calibration set (n = 135) was obtained following a conditioned Latin hypercube sampling, and a validation set (n = 120) was obtained from independent sites using a spatially stratified random sampling strategy. Air-dried and ground (<0.5 mm) soils were scanned using a high-throughput screening accessory for diffuse reflectance attached to a Fourier transform infrared spectrometer. Soil properties for interval-depth datasets were calibrated to smoothed first derivative MIR spectra using partial least-square regression (PLS) and the results were evaluated using hold-out full cross-validation. MIR spectra resulted in good prediction models with coefficient of determination (R2 ) ≥ 0.60 for air-dried moisture content (mc), liquid limit (LL), plastic limit (PL), plasticity index (PI), linear shrinkage (LS), coefficient of linear extensibility (COLE), volumetric shrinkage (VS), cation exchange capacity (CEC) and total sand content (tSa). Further independent validation gave R2 ≥ 0.72 and a ratio of prediction deviation (RPD) 3.8–2.0 for mc, LL, PI, LS, COLE, VS and CEC. MIR provides rapid estimation of several soil properties that provide stability indices valuable for non-agricultural land use in the Lake Victoria basin of Kenya. Further studies should test MIR PLS for interval/separate-depth calibrations and MIR screening of soil properties based on limitation classes commonly applied in civil works.