GIS-based sinkhole susceptibility mapping using the best worst method

Sinkholes are among karst forms and their formation is continuous and their identification is essential in several fields of life, such as water resources management, environmental hazards management, and tourism. This study aimed to identify the sinkholes and the sinkhole susceptibility in the Bistoon-Parav karst region, Iran. Ten sinkhole causative factors, precipitation, temperature, evaporation, lithology, soil type, slope, latitude, fault, stream and vegetation were involved in the sinkhole susceptibility model applying the best worst method, and we also determined the importance of the factors. The final sinkhole susceptibility map was produced by the weighted summing up the factors based on the variable importance. Lithology was the most important factor with 31.52% in the formation of sinkholes. The validation step was executed with a sinkhole database based on visual interpretation of high-resolution imagery. Finally, the receiver operating characteristic (ROC), completeness, correctness and quality index were applied to validate the performance of the sinkhole susceptibility map model. According to the validation parameters, the value of the ROC, completeness, correctness and quality was 81.90%, 100%, 59.41% and 59.41%, respectively. Thus, it can be said that the produced model shows acceptable performances for sinkhole susceptibility mapping. Also, this model showed that almost 7.4% of the region has the potential to become a sinkhole in the future.