GIS-fuzzy logic technique in modeling soil depth classes: Using parts of the Clay Belt and Hornepayne region in Ontario, Canada as a case study

• Soil depth classes have been successfully predicted on the landscape.
• A 10 m LiDAR-derived DEM was useful in generating terrain attributes.
• We generated a digital soil depth classes' map with overall accuracy of 94%.
• GIS-fuzzy logic technique has proven useful in predicting soil depth classes.

Globally, there is a growing desire to predict and map the spatial distribution of soil depth on the landscape due to its numerous roles in areas such as agriculture, forestry, hydrology and ecological land classification. The aim of this study was to develop a technique to model and map soil depth classes based on GIS-fuzzy logic modeling approach and to examine the mapping accuracy of this technique with the accuracy of soil depth classes' derived using randomForest modeling approach. This was carried out for a portion of the Clay Belt and Hornepayne region in Ontario, Canada as a case study. The GIS-fuzzy logic approach was performed based on the soil-environment model (case-based reasoning and rule-based reasoning) using a 10 m LiDAR-derived Digital Elevation Model (DEM) and derivatives such as curvature, slope, aspect, slope position classification, smooth multi-path wetness index and surface roughness in combination with mode of deposition and spatial observations of soil depth information. A digital soil depth map with classes of deep (> 120 cm), moderately deep (> 60 and ≤ 120 cm), shallow (> 15 and ≤ 60 cm) and very shallow (≤ 15 cm) was predicted across 430,076 ha. An overall mapping accuracy of 94% in validation was obtained from the GIS-fuzzy logic modeling approach relative to 88% derived from the randomForest modeling technique. The GIS-fuzzy logic modeling approach could be readily implemented to predict and map the spatial distribution of soil depth classes on various landscapes.