Comparison between the performance of artificial neural network and adaptive neuro-fuzzy inference system in modelling crop evapotranspiration of a maize crop in soil amended with biochar and inorganic fertilizer

• Published in: 2023 International Conference on Science, Engineering and Business for Sustainable Development Goals (SEB-SDG) Vol. 1; pp. 1 - 8
• Main Authors: Faloye, Oluwaseun Temitope, Ajayi, Ayodele Ebenezer, Babalola, Toju, Adabembe, Bolaji, Adeyeri, O. E., Ogunrinde, Akinwale Tope, Okunola, Abiodun, Fashina, Abayomi
• Format: Conference Proceeding
• Language: English
• Published: IEEE 05-04-2023
• Subjects: Adaptation models; Artificial intelligence; Artificial neural networks; Biological system modeling; Crop water use; Crops; Estimation; maize; modelling; Predictive models; Soil
• DOI: 10.1109/SEB-SDG57117.2023.10124578

The success of irrigation management is highly dependent on the accurate estimation of crop evapotranspiration (ETc), necessitating the need to accurately and precisely determine the performance of artificial intelligence (AI) in predicting maize crop under different soil conditions. The crop evapotranspiration is affected by the soil conditions and the meteorological parameters. Therefore, the amount of soil conditioners (biochar and inorganic fertilizer) in the field study under different water application levels were used as model input. In addition, the meteorological data were included as part of the model input. AI models (ANN and ANFIS) were used for the prediction. The crop evapotranspiration were determined in the field using water balance method. Also, the performance evaluation of the considered models were carried out by using metrics like Root Mean Square Error (RMSE), Normalized Root Mean Square Error (NRMSE), Mean Absolute Error (MAE) and the coefficient of determination (\mathbf{r}^{2}) . Result of the analysis showed that during training, the coefficient of determination (\mathbf{r}^{2}) was 0.998, 0.95, and 0.96 for ANFIS, ANN-LOGSIG and ANN-TANSIG, respectively. Similarly, the testing result showed a very high accuracy and precision in terms of the \mathbf{r}^{2} , RMSE and the MAE values. During validation, \mathbf{r}^{2} values were 1, 0.996 and 0.999 for ANFIS, ANN-LOGSIG and ANN-TANSIG, respectively. The prediction of all data showed that \mathbf{r}^{2} were 0.988, 0.984, and 0.997 for the ANN-LOGSIG, ANN-TANSIG and ANFIS models. Therefore, the incorporation of AI model into the estimation of crop water use may be important for proper and adequate water budgeting in Agriculture