The Comparing of Infiltration Models through Parameters Uncertainty Analysis into Two Types of Soil Texture

Background and objectives: One of the main important strategies can be achieved to preserve the water resources is the optimal use of water in the agricultural sector. Water infiltration as a key component of water resource plays a significant role in this challenging problem. Water permeability in soil strongly depends on environmental factors, climatic conditions, latitude and soil characteristics, and has high spatial variability. The various simulation models are used to predict the amount of water infiltration in the soil. there have been a lot of research on the estimation of water infiltration models in the soil, which has been studied in the public only to study the models of infiltration and influence only one of the effective factors on infiltration. The proper prediction of the infiltration and the uncertainty assessment of these models through GLUE (Generalized Likelihood Uncertainty Estimation) algorithm, considering the region conditions and the combined effect of several factors affecting the infiltration process, is main aim of this study.

Materials and methods: Their measurement is conducted in one of the farms of the county in Roshtkhar. The required measurements through double cylindrical method were carried out in three iterations and in two different soil textures, sandy loam and clay loam. In this research, the uncertainty assessment was performed by four models of Kostiakov, SCS, Philip and Horton infiltration. Uncertainty prediction of these four models through GLUE (Generalized Likelihood Uncertainty Estimation) algorithm was used in the MATLAB programming environment, with 100,000 iterations. GLUE mapped all sources of uncertainty into parameter uncertainty. Moreover GLUE conceptual simplicity and it flexibility has led to this method being considered as one of the most applied methods in the uncertainty evaluation in other sciences in last decade studies. Here one percent of best simulations were selected to define the 95percent prediction uncertainty. Hence posterior distribution of each model parameters is plotted and assessed.
Results: In order to quantify the results of uncertainty, four indicators "the percentage of data placement measured in the desired confidence ranges, the bandwidth of the simulated data in the desired confidence range, the degree of asymmetry of the simulated data in the desired confidence range", which is briefly depicted in letters" P, d, s, and T ", respectively. The results showed that the parameters of the Kostiakov model in both soil texture with a value of P greater, d less, s between zero and 0.5 And T between zero and one show more certainty, so that the values of these indicators in sandy loam soils equal to 100, 0.378, 0.055 and 0.388 respectively, and in clay loam soils In 100, 0.519, 0.147 and 0.558 respectively.

Conclusion: According to the four P, d, s, and T indices, it was found that the paramteres of Kostiakov model has more certainty than other model parameters, which can be considered as an appropriate model.