JWSS - Journal of Water and Soil Science

Soft computing techniques have been extensively studied and applied in the last three decades for scientific research and engineering computing. The purpose of this study was to investigate the abilities of multilayer perceptron neural network (MLP) and neuro-fuzzy (NF) techniques to estimate the soil-water retention curve (SWRC) from Khozestan sugarcane Agro-Industries data. Sensitivity analysis was used for determining the model inputs and appropriate data subset. Also, in this paper, the van Genuchten and Fredlund and xing models were used to predict SWRC. Measured soil variables included particle size distribution, organic matter, bulk density, calcium carbonate, sodium adsorption ratio, electrical conductivity, acidity, mean weight diameter, plastic and liquid limit, resistance of soil penetration, water saturation percentage and water content for matric potentials -33, -100, -500 and -1500 kPa. The results of this study in terms of various statistical indices indicated that both MLP and NF provide good predictions but the neural network provides better predictions than neuro-fuzzy model. For example, using MLP and NF models values of NMSE at prediction θs, θr, α, n and m in Fredlund and Xing equation corresponded to (0.059, 0.065), (0.154, 0.162), (0.109, 0.117), (0.129, 0.135) and (0.129, 0.145), respectively. Furthermore, α and n parameters at the first depth, and θr and α parameters at the second depth in Fredlund and Xing equation were estimated with higher accuracy compared with equivalent parameters in van Genuchten equation

Modeling of flow and transport processes in variably saturated porous media requires detailed knowledge of the soil hydraulic properties. The hydraulic properties to be determined by the inverse problem solution are the unsaturated hydraulic conductivity K(h) and the water retention curve θ(h). The inverse modeling approach assumes that both θ(h) and K(h) as well as transport parameters can be determined simultaneously from transient flow data by numerical inversion of the governing flow and transport equations. In order to find answers to the questions of uniqueness, identifiability and stability of different experimental setups, a new numerical experiment of redistribution was carried out. To study the shape of the objective function near its minimum, response surfaces for the estimated parameters were generated. The sensitivity of model outputs with respect to changes in input parameters was also computed and analyzed. Results of the redistribution experiment suggest that the non-uniqueness increases when the model output variables are not sensitive enough to the optimized parameters. As expected, the estimated values of parameters were sensitive to the magnitude of error in the measured data. In this experiment, the parameter estimation based on the pressure head observations provides unique solution. Due to preferential flow in the sample, tensiometric observations may provide poor results for inverse problem solution. Taking into account information about saturated hydraulic conductivity, Ks improved the likelihood of uniqueness and reduced the errors in parameter estimation of the shape parameters (α, n). It was found that the sensitivity analysis could be a useful tool to design the optimal time and location distribution of experimental observations.

Precipitation is an important element of the hydrologic cycle and lack of this data is one of the most serious problems facing research on hydrological and climatic analysis. On the other hand, using satellite images has been proposed by many researchers as one of practical strategies to estimate precipitation. The present paper aims to evaluate the accuracy of satellite precipitation data, provided by PERSIANN and TRMM-3B42 V7 in Gorganrood basin, Iran. To achieve this aim, two sets of daily precipitation ground-based data, 2003 to 2004 and 2006 to 2007, from six stations of Gorganrood basin, named; “Tamer”, “Ramian”, “Bahalkeh-ye Dashli”, “Gorgan Dam”, “Ghaffar Haji” and “Fazel Abad” have been used in this paper. The evaluation indices have been calculated and analyzed in different time scales, including daily, monthly and seasonal. The results indicated that the two above mentioned satellite models are not accurate in daily scale. However, they showed reasonable accuracy in monthly and seasonal scales. The highest correlations between satellites and recorded data in daily and monthly scales, for TRMM-3B42 V7 in “Gorgan Dam” and “Bahlke Dashlei” stations, are 0.397 and 0.404, respectively. The comparison of measured and satellite data of winter showed better agreement for PERSIANN model. However, TRMM-3B42 V7 shows better correlation in other seasons. The results also indicated that while TRMM-3B42 data displays higher correlation with measured data, PERSIANN provids better results in predicting the number of rainy days. 

Selection of optimum interpolation technique to estimate water quality parameters in unmeasured points plays an important role in managing the quality and quantity of water resources. The aim of this study is to evaluate the accuracy of interpolation methods using GIS and artificial neural network (ANNs) model. To this end, a series of qualitative parameters of samples from water taken from Dehgolan aquifer located in Kurdistan, Iran including CL, EC and PH were evaluated by any of the models. In this study, qualitative data from 56 observation wells with good dispersion in the whole plain was used. The data of 46 observation wells were used for calibration and the data of other 10 wells were used for verification of models. The results showed ANNs, IDW, and Kriging excellence and accuracy over other models in estimation of quality parameters CL, PH and EC. However the ANNs model is more accurate than other models. In case of lack of time and the need for acceptable accuracy and less risk in the estimation of qualitative parameters, the use of ANNs model is superior to other statistical models used.