Modeling and optimization of CO2 emissions for tangerine production using artificial neural networks and data envelopment analysis

摘要

The aims of this research were to scrutinize CO2 emissions based on different size levels of orchards, modeling and optimization of CO2 emissions using artificial neural network (ANN) and data envelopment analysis (DEA) for tangerine production in Guilan province, Iran. The results revealed that the total CO2 emissions and yield were calculated about 622 kgCO2eq. ha-1 and 49 ton ha-1, respectively. Also, the large orchards had the highest emissions and yield among all of the groups. The results of ANN modeling indicated that the best topology was 8-4-1 for prediction of tangerine yield based on emission inputs. Also, the R2, RMSE and MAPE (%) of the best structure was computed as 0.964, 0.111 and 0.312, respectively. Based on DEA approach, the mean of technical, pure technical and scale efficiency was found to be 0.802, 0.890 and 0.894, respectively. In optimal units, the total CO2 emissions were calculated as 483.83 kgCO2eq. ha-1. The results indicated that optimization of CO2 emissions by the DEA can reduce the total emissions about 138 kgCO2eq. ha-1 and the reduction of electricity consumption had the highest positive effect in CO2 emissions reductions.