Abstract
Many attempts have been created to determine wetting pattern under trickle irrigation using sophisticated mathematical and numerical models, required detailed information concerning soil physical properties and too complicated for routine use. For this reason, an alternative methodology is proposed, which combines artificial neural networks (ANNs), laboratory and field experiments. The model input parameters were saturated hydraulic conductivity, application rate, volume of water applied and average change of moisture content. The model outputs were surface wetted radius and vertical advance of wetting front. A total of 280 and 100 vectors were used to train the ANNs model for surface wetted radius and vertical advance of wetting front estimations, respectively. To test the ANNs model, a total of 132 and 76 vectors were selected in case of surface wetted radius and vertical advance of wetting front estimations, respectively. Results of the test show that the surface wetted radius and vertical advance of wetting front can be predicted with a determination coefficient (r2 ) of 0.80 and 0.81 for the surface wetted radius and vertical advance of wetting front, respectively. Additionally, the ANNs approach was found to produce equally or more accurate descriptions of wetting pattern as compared to several analytical and empirical models which suggested for point source trickle irrigation design.