Modeling sprinkler irrigation infiltration based on a fuzzy-logic approach

Abstract

In this study, the irrigation water infiltration rate (IR) is defined by input variables in linguistic terms using a fuzzy-logic approach. A fuzzy-logic model was developed using data collected from published data. The model was trained with three fuzzy membership functions: triangular (‘trimf’), trapezoid (trapmf), and pi (‘pimf’). The fuzzy system considered the number of irrigation events, applied water depth, polyacrylamide application rate, water application time, water electrical conductivity, soil surface slope, and soil texture components as input variables. The inputs were classified in terms of low, medium, and high levels. The output variable (i.e., IR) was rated in terms of five levels: very low, low, medium, high, and very high. Using statistical analysis, the values of IR resulting from the developed fuzzy-logic model were compared with the observations from the experiments. The results confirm that the agreement between the observations and predictive results was acceptable, except for fuzzy 'trimf'. The coefficient of determination provided the greatest value when using the 'trapmf' and 'pimf', with the value estimated for the 'pimf' slightly higher than that of 'trapmf'. Based on the results that were obtained, irrigation managers can use the fuzzy-logic approach to modify their field practices during the growing season to improve on-farm water management.