A new approach in comparison and evaluation of the overall accuracy of six soil-water retention models using statistical benchmarks and fuzzy method

Abstract In this research, six samples of valid and widely soil-water retention curve (SWRC) estimation models, including van Genuchten, Brooks and Corey, Fredlund and Xing, Durner, Kosugi, and Seki were studied. To realize this approach, in the first step, the accuracy of each model was calculated using ten statistical benchmarks, and then the numbers obtained from the fitting accuracy were standardized based on each benchmark by the standard fuzzy method so that all of them had a similar scale. Finally, with the sum of the fuzzy standardized values, an index for each model was obtained as a new index called the Best-Fit Model (BFM) index, which was the basis for comparing and evaluating the overall accuracy of the models in each soil texture. Accordingly, if the BMF index is more extensive and closer to 10, the model is more fitted and gets a higher rating. The superiority of this method to other similar studies is that here, as in the multi-criteria evaluation methods, it can simultaneously assess in terms of different statistical benchmarks and ranking the models according to the diversity in the reaction of each to various benchmarks provided. The results showed that Brooks and Corey model with the lowest and the Durner model with the highest BMF and rank among other models in most soil textures are considered as the weakest and as the most suitable model from the overall accuracy viewpoint of fitting based on the approach used throughout this study, respectively.