Developing scoring functions for soil quality to assess land suitability for irrigated wheat in Southern Algeria

Efficient land-use management could be optimized through appropriate advice regarding land suitability and achieving a reasonable yield while preserving soil and environmental conditions. The main objective of this study was to determine and define those locations capable of wheat cultivation with irrigation in the municipality of N’goussa (Southern Algeria). Expert opinion (EO) method was used for selecting minimum soil data set (MDS_-EO), where six influencing soil properties (sand, pH, salinity, organic matter, gypsum and lime) and other three key parameters (slope, land use/land cover, and proximity to roads and electricity network) were considered important for irrigated wheat cultivation in the study area. Indicators were weighted according to principal component analysis (PCA) loadings and summed. PCA revealed that the selected nine indicators are appropriate for accounting 73.63% of the total variance. Two maps of land suitability were created here (expressed as a soil quality index within a range from 0 to 1) from multivariate analysis and already known standard scoring functions (SSFs) by using two data sets (pure and scored). The distribution of soil quality index (IQI) showed that above 70% of the land was moderately (MS) to highly suitable (HS) for irrigated wheat. The estimation accuracy of soil quality classes has been evaluated using random forests (RF), having as a dependent variable the final class of land suitability used in each map. It was determined that estimation accuracy could reach 60% (pure data set) and 65% (scored data set) at 0.5 cut-off value for soil quality classes. Final prioritizing according to ranks of sensitivity index (SI) and efficiency ratio (ER) showed that IQI obtained from scored data set is the most suitable approach in soil quality assessment of arid areas. Therefore, it could be applied in southern Algeria for land management, integrated planning and environmental assessments.