Modelling experts’ judgments on soil compaction to derive decision rules for soil protection—A case study from Switzerland

Soil compaction has been recognized as a serious environmental problem, and therefore its prevention is incorporated in the soil protection legislations of many countries. Implementation of these legislations needs decision rules and thresholds to assess the state and risk of soil compaction. The following contribution presents the results of a Delphi survey among Swiss soil experts eliciting their personal experience in judging soil compaction in order to derive their implicit decision rules and thresholds. The experts were asked in the first turn to mention the three most important physical soil parameters they use to judge the state of and the susceptibility to compaction. In the second and third turn, they had to group 21 data records with real measurements of soil physical parameters into five different classes of severity for the state of and susceptibility to compaction. We analysed the experts’ answers with descriptive statistics and a cluster analysis. Further, we developed a fuzzy membership model to classify the soil data after the experts’ groupings. The model was validated with a Jack–Knife estimation for each data record. The Delphi survey revealed a high convergence of the experts’ answers. Although, in the first turn, the experts stated qualitative parameters like water regime and soil structure to be very important for judging soil compaction, in the second and third turn they mentioned only quantitative parameters to be crucial for their groupings of the data records. The decisive parameters to judge the state of compaction were coarse pore content followed by bulk density and precompression stress. Conversely, for judging the susceptibility to compaction precompression stress was stated as crucial parameter, followed by bulk density, coarse pore content and clay content. We observed that the experts allowed for different soil parameters simultaneously. We particularly assume that they took the clay content into account while judging the parameter of bulk density. The implicit thresholds of the experts were derived from the mean values and the ranges in the different classes of the state of and susceptibility to compaction. The critical limits correspond to indications in the literature. The experts defined soils with less than 7% coarse pore content and more than 1.7 packing density (a composite parameter of bulk density and clay content) as compacted. Soils with less than 65 kPa precompression stress and less than 1.7 packing density were considered very susceptible to compaction. While the statistical cluster analysis could not model the experts’ groupings in a satisfactory way, the fuzzy membership model was successful in 18 cases (out of 21) for both, the state of and susceptibility to compaction. The Jack–Knife estimation was successful in 11 cases for the state of compaction and in 12 cases for the susceptibility to compaction. The number of available samples was too small for a better Jack–Knife prediction. Although this model is still a prototype, it allows more for uncertainties of the experts’ decisions than thresholds from descriptive statistics, and may therefore be a more traceable tool to support the execution of the soil protection legislations.