Sensitivity Assessment of Nitrous oxide Greenhouse Gas Emissions in Agricultural Lands of Khuzestan Province with Linear and Non-linear Models

The aim of this study is to estimate emission of nitrous oxide gas in rice, wheat and sugarcane fields of Khuzestan using four models: DAYCENT, DNDC, YLRM and IPCC_EF. For this purpose, nitrous oxide gas precipitation was first measured. Then, using models was estimated nitrous oxide gas expansion rate. To evaluate and compare accuracy of models, statistical characteristics were used, coefficient of determination, maximum error, root mean squares error, modeling efficiency and remaining coefficient of residual mass. Release of nitrous oxide in rice cultivation in four models was estimated to be between 0.17 and 0.171. Rate of nitrous oxide emission from wheat cultivation was between 0.5-0.049 and from Shushtar station sugarcane cultivation was between -0.0371 and from Abadan station sugarcane cultivation was between 0.03-0.85. In linear regression model of rice cultivation (1.17), in IPCC_EF model, wheat cultivation (0.5) and sugarcane (3) obtained the highest amount of nitrous oxide gas per ton per hectare per year. According to results of statistical indicators for four models DAYCENT, DNDC, YLRM and IPCC_EF to estimate nitrous oxide gas were determined, respectively, the coefficient of determination (0.86, 0.94, 0.99 and 0.82), root mean squares error (0.03, 0.01, 0.85 and 0.26) and modeling efficiency (0.55, 0.94, -4.87 and -30.63).Compared to observed values, DAYCENT model for corn, DNDC model for rice, linear regression model for sugarcane cultivation of Abadan station showed good performance. Based on results of coefficient of determination, YLRM and DNDC models received the highest accuracy based on modeling efficiency of DNDC model.