Recognition of potential outliers in soil datasets from the perspective of geographical context for improving farm-level soil mapping accuracies

•Soil sample outlier recognition method from geographical perspective is proposed.•It can fully excavate the local spatial structure information of soil sample data.•Spatial search distance is automatically determined by the spatial distance among soil samples.•Spatial autocorrelation and heterogeneity are considered to calculate outlier scores.•Threshold to potential outliers in soil datasets was determined by a frequency histogram.