Assessment Chemical Properties of Soil in Intercropping Using ANN and ANFIS Models

1 University of Zabol, Departments of Agronomy, Faculty of Agriculture, Zabol, 98615-538 Iran
2 University of Zabol, Deptartments of Civil Engineering, Zabol, 98615-538 Iran
3 Departments of irrigation, Faculty of Water and Soil Sciences, Zabol, 98615-538 Iran


Dahmardeh, M., B. Keshtegar and J. Piri, 2017. Assessment chemical properties of soil in intercropping using ANN and ANFIS models. Bulg. J. Agric. Sci., 23 (2): 265–273

Intercropping as an example of sustainable agricultural systems follows objectives, such as the ecological balance, more exploitation of resources and increase in soil fertility. Evaluation of soil nutrients in intercropping is a basic criterion for selecting the type of cultivation and increasing the productivity of the soil. Based on experimental study, evaluation of soil needs time and cost too much. However the soil parameters can be rapidly estimated using predicted meteorology and can be appropriately assessed. Linear estimation methods are less accurate than non-linear methods, but non-linear methods for modeling of soil elements are diffi cult due to highly computing times. The artifi cial intelligence is a powerful tool for fast modeling, more accurately. In this paper, artifi cial intelligence method such as Artifi cial Neural Networks (ANN) and Adaptive Neuro-Fuzzy Inference System (ANFIS) has been used to estimate soil chemical properties (carbon and nitrogen). Four input parameters including the soil temperature, type of intercropping (different ratios of Roselle – green gram), type of tillage (no-tillage, minimum tillage, conventional tillage) and sodium have been used in ANN and ANFIS prediction. ANFIS and ANN modeling have been compared using several statistics (root mean square error (RMSE) and Mean Absolute Error (MAE). The results indicate that the artifi cial intelligences can be successful applied for estimating soil carbon and nitrogen. ANFIS is found to be more accurate than the ANN. A sensitivity analysis has conducted based on ANFIS estimate. It is shown that increased percent of green gram in intercropping can reduce the percentage of carbon to nitrogen (C/N).

Key words: Intercropping; ANN; ANFIS; soil nutrients; tillage
Abbreviations: ANN – Artifi cial Neural Networks, ANFIS – Adaptive Neuro – Fuzzy Inference System, MAE – Mean Absolute Error, RMSE – Root Mean Square Error

See the article as a PDF.