Application of artificial neural networks for prediction of output energy and GHG emissions in potato production in Iran

This study was carried out in Esfahan province in Iran in order to model output energy and greenhouse gas (GHG) emissions of potato production on the basis of input energies using artificial neural networks (ANNs). Data were collected from 260 farms in Fereydonshahr city with face to face questionnaire method. Accordingly, several ANNs were developed and the prediction accuracy of them was evaluated using the quality parameters. The results illustrated that the average total input and output energy of potato production were 83,723 and 83,059 MJ ha−1, respectively. Electricity, chemical fertilizers and seed were the most influential factors in energy consumption with amount of 30.5, 28 and 12 GJ ha−1. Energy use efficiency and energy productivity were 1.03 and 0.29 kg MJ−1, respectively. Total GHG emission was calculated as 116.4 kg CO2 per ton of potato produced. The ANN model with 12-8-2 structure was the best one for predicting the potato output energy and total GHG emission. The coefficient of determination (R2) of the best topology was 0.98 and 0.99 for potato output energy and total GHG emission, respectively.