Considering that the ground water used for supplementary irrigation has variable chemical characteristics that may affect soil properties, a typical Argiudoll (A1) and an aquic Argiudoll(Ap) were treated with artificial irrigation waters with different sodium adsorption ratios (RAS): 6.3, 11.0, 16.5, 21.5 and 24.3. The following tests were carried out: carbon (Walckey Black); pH (1:2.5) (by pHmeter); cation exchange capacity (CIC); Ca2+ , Mg2+ and H+ (volumetric method) and Na+ and K+ (spectroscope of atomic emissions). The objectives of this work were to find prediction models such as Na exchange percentage (PSI) vs. RAS, pH vs. RAS and pH vs. adsorbed Na; to analyze if there is a limit value for adsorbed Na; to determine the evolution of organic matter, and to compare PSI using the obtained prediction. No Na adsorption limit value was found, but a carbon loss was encountered in relation to initial values. The following prediction models were determined: PSI vs. RAS: quadratic for Ap (p &#8804; 0.0261; r²aj = 0.839) and linear for A1 (p < 10-4; r²aj = 0.828); pH vs. Na: quadratic (p < 10-4; r²aj = 0.991) for Ap and A1 (p < 10-4; r²aj = 0.995), and pH vs. RAS: also quadratic for Ap (p < 10-4; r²aj = 0.979) and A1 (p &#8804; 0.0011; r²aj = 0.911). PSI errors were lower with the prediction equations than with Richards&#8217; equation. The models found allow diagnosis of the sodicity and pH that the studied Argiudolls acquire when subjected to the irrigation water with different levels of RAS. | Considerando que el agua subterránea usada para riego suplementario tiene características químicas variables que pueden afectar las propiedades del suelo, se trataron un Argiudol típico (A1) y un Argiudol ácuico (Ap), con aguas de riego artificiales de diferentes relaciones de adsorción de sodio (RAS): 6,3; 11,0; 16,5; 21,5 y 24,3. Se analizó: carbono (Walckey Black); pH (1:2,5) (a pHmetro); capacidad de intercambio catiónico (CIC), Ca2+, Mg2+ y H+ (volumetría) y Na+ y K+ (espectroscopía de emisión átómica). Los objetivos del trabajo fueron: encontrar modelos de predicción de porcentaje de sodio intercambiable (PSI) vs. RAS, pH vs. RAS y pH vs. Na adsorbido; analizar si existe un valor límite de Na adsorbido; determinar la evolución de la materia orgánica y comparar PSI usando las ecuaciones predictivas obtenidas. No se encontró un valor límite de adsorción de Na+ y sí pérdida de carbono con relación al valor inicial. Se determinaron los siguientes modelos de predicción: PSI vs. RAS: cuadrático para Ap (p &#8804; 0,0261; r²aj = 0,839) y lineal para A1 (p < 10-4; r²aj = 0,828), cuadráticos para pH vs. Na (p < 10-4; r²aj = 0,991) para Ap y para A1 (p < 10-4 r²aj = 0,995) y también cuadráticos para pH vs. RAS (p < 10-4; r²aj = 0,979) para Ap y para A1 (p &#8804; 0,0011; r²aj = 0,911). Usando las ecuaciones de predicción se obtuvieron errores de PSI menores que con la ecuación de Richards. Los modelos encontrados permiten diagnosticar el grado de sodificación y el pH que adquieren los argiudoles estudiados sometidos a aguas de riego de diferentes RAS.

Jinci Spring in Shanxi, north China, is a major local water source. It dried up in April 1994 due to groundwater overexploitation. The groundwater system is complex, involving many nonlinear and uncertain factors. Artificial neural network (ANN) models are statistical techniques to study parameter nonlinear relationships of groundwater systems. However, ANN models offer little explanatory insight into the mechanisms of prediction models. Sensitivity analysis can overcome this shortcoming. In this study, a back-propagation neural network model was built based on the relationship between groundwater level and its sensitivity factors in Jinci Spring Basin; these sensitivity factors included precipitation, river seepage, mining drainage, groundwater withdrawals and lateral discharge to the associated Quaternary aquifer. All the sensitivity factors were analyzed with Garson’s algorithm based on the connection weights of the neural network model. The concept of “sensitivity range” was proposed to describe the value range of the input variables to which the output variables are most sensitive. The sensitivity ranges were analyzed by a local sensitivity approach. The results showed that coal mining drainage is the most sensitive anthropogenic factor, having a large effect on groundwater level of the Jinci Spring Basin.