Ferdowsi University of MashhadWater and Soil2008-475726320120822Using Fractal Dimension of Particle Size in Estimating Saturated Hydraulic ConductivityUsing Fractal Dimension of Particle Size in Estimating Saturated Hydraulic Conductivity3615010.22067/jsw.v0i0.14929FAV. YazdaniWater Engineering Department, Faculty of Agriculture,
Ferdowsi University of MashhadB. GhahremanWater Engineering Department, Faculty of Agriculture,
Ferdowsi University of Mashhad0000-0002-8201-5060K. DavariWater Engineering Department, Faculty of Agriculture,
Ferdowsi University of MashhadM.E. FazeliDepartment of Water Engineering, Islamic Azad University, Ferdows BranchJournal Article20120829One of the important aspects of soil is, knowing the relationships between spatial features of soil and quantity in statistical model. The goal of this research is to estimate saturated hydraulic conductivity by regression and Co-Active Neuro-Fuzzy Inference System (ANFIS) with using the parameters of Bulk density, real density, porosity, Fractal dimension of particle size, and clay percent, silt percent, sand percent. So experiment related to saturated hydraulic conductivity calculation and soil physical properties calculation of soil in 54 points which were specified 5 by 5 meters. Also, amount of bulk density based on paraffin Hunk, Fractal dimension of particle size by wet sieve method, the percentage of sand, clay and silt by Hydrometry and saturated hydraulic conductivity above water table by double rings method was measured. The best regression model for Pedo transfer function (PTF) was chosen according to minimized the goal function based on statistical parameters R2, RMSE and MAE. Parameters sand and silt percent, bulk density, real density, Fractal dimension of particle size, and porosity were chosen as input. In PTF amount of R2, RMSE, NRMSE and MAE, are 0.65, 0.017, 0.96 and 0.012 respectively. ANFIS with four layers input includes bulk density, real density, porosity and Fractal dimension of particle size and an output layer with the best performance. In this research, the amount of R2 in the presented ANFIS model in training and test is 0.88 and 0.86 respectively, and RMSE values will be 0.012 and 0.02 respectively. Noticing to sensitivity analysis result, PTF has the least sensitivity than changes in porosity and Fractal dimension of particle size, on the other hand, it has the most sensitivity than changes in the values of bulk density, silt and sand percent. ANFIS model is like PTF is more sensitivity than changes in values of bulk density. In addition, the outcome shows more effect on ANFIS than PTF. Evaluation of models show that estimation in clay soil is not acceptable, in contract contrast its model for estimate saturated hydraulic conductivity in soil texture (lom sandy, lom and silt lom) is suitable.One of the important aspects of soil is, knowing the relationships between spatial features of soil and quantity in statistical model. The goal of this research is to estimate saturated hydraulic conductivity by regression and Co-Active Neuro-Fuzzy Inference System (ANFIS) with using the parameters of Bulk density, real density, porosity, Fractal dimension of particle size, and clay percent, silt percent, sand percent. So experiment related to saturated hydraulic conductivity calculation and soil physical properties calculation of soil in 54 points which were specified 5 by 5 meters. Also, amount of bulk density based on paraffin Hunk, Fractal dimension of particle size by wet sieve method, the percentage of sand, clay and silt by Hydrometry and saturated hydraulic conductivity above water table by double rings method was measured. The best regression model for Pedo transfer function (PTF) was chosen according to minimized the goal function based on statistical parameters R2, RMSE and MAE. Parameters sand and silt percent, bulk density, real density, Fractal dimension of particle size, and porosity were chosen as input. In PTF amount of R2, RMSE, NRMSE and MAE, are 0.65, 0.017, 0.96 and 0.012 respectively. ANFIS with four layers input includes bulk density, real density, porosity and Fractal dimension of particle size and an output layer with the best performance. In this research, the amount of R2 in the presented ANFIS model in training and test is 0.88 and 0.86 respectively, and RMSE values will be 0.012 and 0.02 respectively. Noticing to sensitivity analysis result, PTF has the least sensitivity than changes in porosity and Fractal dimension of particle size, on the other hand, it has the most sensitivity than changes in the values of bulk density, silt and sand percent. ANFIS model is like PTF is more sensitivity than changes in values of bulk density. In addition, the outcome shows more effect on ANFIS than PTF. Evaluation of models show that estimation in clay soil is not acceptable, in contract contrast its model for estimate saturated hydraulic conductivity in soil texture (lom sandy, lom and silt lom) is suitable.https://jsw.um.ac.ir/article_36150_cd6c368f43d8056ad25b92d741abdfb4.pdf