Optimization of energy consumption and analysis of the emission of greenhouse gases in the production of rice (Case study: Mazandaran province of Iran)

The present study has been conducted to investigate the energy consumption pattern, the economic analysis of production and assessing the greenhouse gas emissions (GHG) resulting from the production of rice in the Mazandaran province. The input and output energies were calculated and energy indices were determined. The eciency of the rice farms and the optimization of energy consumption were estimated by using data envelopment analysis (DEA) method and the amount of greenhouse gas emissions was determined after improving the consumption of inputs. Results showed, the total energies of consumed inputs and outputs for the production of paddy were calculated to be 91061.5 and 100967.95 MJ ha -1 , respectively. The energy input of diesel fuel, machinery and chemical fertilizers were determined to be 41.27%, 19.63% and 19.58% of the total energy inputs, representing the most highly consumed energy inputs in the production of paddy.

return to scale model. With the improvement of the consumed inputs, the total energy saving was determined to be 19.8%, while the consumptive fuel energy had the biggest share among the energy inputs out of the total energy saving at 43.41%. The quantities of greenhouse gas emissions under the real and improved states were reported to be 1,847.26 kg CO 2 eq. ha -1 and 1,483.52 kg CO 2 eq. ha -1 , respectively [36].
In investigating the consumptive energy of rice production in the Guilan province, energy use e ciency and energy productivity were expressed to be 1.53 and 0.09 kg MJ -1 , respectively. The input and output energies of rice production were determined to be 39,333 and 60,341 MJ ha -1 where the fuel energy, with a share of 46 percent, and chemical fertilizers, with a share of 36 percent, represented the highest portions in energy consumption. Moreover, the shares of indirect and non-renewable energies were 51 percent and 89 percent respectively, indicating the big share of fossil energy in the production of rice. The tted econometrics model, assisted by the Cobb-Douglas production function, signi ed that the impact of fuel input energy and machinery on yield of rice is signi cant. The analysis of the sensitivity of the energy of the inputs in this study showed that the highest marginal physical production of inputs pertained to the fuel energy at 0.92 followed by machinery at 0.23. The economic parameters of rice production including the total cost of production, gross return, net return and the bene t to cost ratio were determined to be 2116,842 $.ha -1 , 2071,738 $.ha -1 , 1277.914 $.ha -1 and 1.604, respectively. The overall conclusion of their research showed that more extensive rice elds (larger than one hectare) demonstrated a better management in energy consumption and performed more successfully [41].
In investigating the e ciency of the consumptive energy, the emission of greenhouse gases and carbon e ciency of rice elds in the Sari County, the energy use e ciency and the net energy gain were reported to be 1.83 and 27,932 MJ.ha -1 . The tted econometrics model, assisted by the Cobb-Douglas production function, showed that the impact of the machinery's input energy and human labor on the yield of the crop is signi cant. The largest elasticity of production (regression coe cient) for the machinery's input energy was obtained to be 0.64, showing that with a 1% increase in the energy consumption of this input, the value of the yield of rice increases by 0.64%. The elasticity of production of human labor input energy was determined to be -0. 37. In other words, with a 1% increase in the energy consumption of this input, the value of the yield of rice decreases by 0.3%. The analysis of the sensitivity of the energy of seed input, human labor, machinery, diesel fuel, fertilizers and chemicals were reported to be 0.058, -0.992, 0.078, 0.004, 0.027 and 0.089, respectively. In other words, with the rise of the consumption of 1 MJ of these inputs, provided that the energy consumption of other inputs remains unchanged, the rice yield changes by 0.058, -0.992, 0.078, 0.004, 0.027 and 0.089 kg ha -1 [13].
In a study, the input and output energies for the production of three varieties of local, high yield and hybrid rice using two traditional and mechanized cultivation methods in the Mazandaran province were investigated. The energy use e ciency for the production of rice using two traditional and mechanized cultivation methods were expressed to be 1.72 and 1.63, respectively [1].
Chauhan et al. [5] investigated the improvement of the energy consumption of paddy production in the alluvial regions of India's West Bengal state by taking into account the e ciency of different units and by using the input-oriented BCC and CCR models of the Data Envelopment Analysis method. In this study, input parameters including the consumptive energy of the inputs of human labor, machinery, diesel fuel, seed, organic fertilizers, chemical fertilizers and the yield of the crop as the output were investigated. They reported that 11.6% of the total consumptive energy in rice production is savable without leading to any decline in the rice yield in case the farmers employ the suggestions presented in the research. They also suggested that better use of consumptive energy in the tillage stage and primary modi cations in the machinery can improve the e ciency of energy in the region. They expressed that better use of the tiller and machinery power improves the energy use e ciency, followed by an improvement in the energy productivity in rice production.
Technical e ciency, pure technical e ciency and scale e ciency of rice producers in India were investigated in terms of the quantity of energy consumption in a non-parametric analysis using the Data Envelopment Analysis method as well as BCC and CCR models, and the e ciency of farmers was determined in terms of energy consumption for the production of rice in ve different sizes and four different climatic regions. Technical e ciency, pure technical e ciency and scale e ciency for the region 2, which was higher than other regions, were reported to be 0.88, 0.91 and 0.96 respectively. Moreover, out of the total 363 farmers studied in the region, 26 people were e cient in the BCC model and 13 people were e cient in the CCR model [38].
The present study has been carried out with the aim of investigating the energy pattern of consumptive inputs and outputs, optimizing the consumptive energy using the data envelopment analysis method, and to conduct economic assessment and analysis on the emission of greenhouse gases resulting from the production of paddy in the Mazandaran province.

Agricultural situation of Mazandaran province
The present research is a eld survey carried out in the rice production farms of Mazandaran province of Iran in the year 2018. In the Figure 1, the geographical position of Mazandaran province is illustrated. This province is situated between 35° 47' to 36° 35' northern latitude and 50° 34' to 54° 10' eastern longitude from the prime meridian.
74 percent of lands of Mazandaran are foothill and mountainous while 26 percent are plain. One of the most important features of the province is its high cultivation coe cient, i.e., 1.4, against the national average of 0.7, which means annually 610,000 hectares of the provincial lands are used for agricultural crops and orchards. In the agricultural year 2017, Mazandaran boasted 213,157 hectares of paddy lands in which 1,261,847 tons of paddy (taking into account the harvesting of ratoon and renewed seedling) were produced, accounting for more than 40% of rice production nationally.
Representing a 37.5% share of the harvest area of the lands under cultivation of rice, Mazandaran province ranks rst nationally, followed by Guilan Province, which comes second in the countrywide ranking by harvesting 31.2 percent of the nation's rice elds. These two provinces together allocate 68.7 percent of the country's rice cultivation area to themselves [48]. The collection of data required in the research was made possible through lling out questionnaires and face-to-face interviews with rice farmers in the Mazandaran province. Other information required were obtained through literature and library research as well as agricultural experts.
In the present research, the simple random sampling technique was used. The population of the research comprised the entire rice-producing farmers in Mazandaran province. The Cochran formula (equation 1) was used to determine the sample size and the number of questionnaires [18,17,8].

Equation 1
Were, n is the sample size, N is the population size, S is the population's standard deviation, d is the favorable probable accuracy and t is equivalent to 1.96.
Accordingly, the required sample size was determined to be 100 rice elds. The rice production input energies included the energy of labor, machinery and tools, fuel, irrigation water, electricity required to pump water, chemical fertilizers (nitrogen, phosphate, potassium and micronutrients), chemical pesticides (herbicides, insecticides and fungicides) and seeds. The output energy included the energy of paddy and straws. To calculate the input and output energies, the quantity of each consumptive input or output was multiplied by its energy equivalent (energy coe cient). The equivalent values of the consumptive inputs and outputs are presented in the Table 1. Table 1 2

.3. Calculating the energy indices
In order to determine the relationship between input and output energies, following indices are de ned and employed. These indices are shown in the equations 2 to 9 [18,30,33,49].

Equation 2
Equation 3 Equation 4 Equation 5 Equation 6 Equation 7 Equation 8 Equation 9 The energy cost is obtained by converting the total input energy into barrel of oil equivalent (BOE) and calculating the price value of crude oil [33]. Every barrel of crude oil is approximately 159 liters, and the energy released from the complete burning there of is 5.8 BTU (British Thermal Unit), or an equivalent of 6.117 gigajoules (GJ).
Researchers consider the energy use e ciency as a criterion for the advancement of technology; therefore, energy use e ciency is the most important index in assessing the energy of agricultural systems. The entire data collected were entered in the excel and the required computations were carried out.

Data Envelopment Analysis
Farrell [11] determined relative e ciency in a production unit like engineering topics with a single input and output. Later on, Charnes et al. [4] improved Farrell method and presented a pattern that could be determined relative e ciency with multiple inputs and outputs. This method was known as DEA. DEA is a relatively new data oriented approach for evaluating the performance of a set of peer entities called decision making units (DMU) which convert multiple inputs into multiple outputs. A great variety of applications of DEA for use in evaluating the performance of different kinds of entities have been seen in the recent years [7].
In DEA, all collected observations are used for measuring e ciency. DEA is making and solving n numbers of models to evaluate n numbers of productivity. In this method, a virtual unit, with the highest relative e ciency, is de ned by combining all understudy units. So, ine ciency of units can be identi ed by comparing them with the virtual unit [43].
In fact, the data envelopment analysis method does not calculate the e ciency of one unit; rather, it calculates the e ciency of one unit against the e ciency of other units and assesses it with other units. The advantages of using the data envelopment analysis method include its comprehensibility, being easy-touse, realistic assessment, the simultaneous assessment of the entire set of factors in uencing the model, being needless of weights and pre-determined production function (non-parametric), portraying the best performance situation instead of the favorable situation, the ability to enter multiple inputs and multiple outputs, not requiring the same measurement units for the inputs and outputs and the direct comparison of units with one level or a combination of corresponding collections [19]. E ciency by DEA is de ned in three different forms: Technical E ciency (TE), Pure Technical E ciency (PTE) and Scale E ciency (SE) [9]. TE represents the potency of a DMU to produce maximum output given the set of inputs and technology (output-oriented) or to obtain minimizing inputs while maintaining the same level of outputs (input-oriented) [11]. Technical e ciency is a measure evaluating DMU performance relative to that of other DMUs in consideration; it is also called global e ciency.
Pure technical e ciency separates technical and scale e ciencies in other words pure technical e ciency is technical e ciency that has the effect of scale e ciency removed. The advantage of this model is that it compares scale ine cient farms only to e cient farms of a similar size. The main advantage of the VRS model is that scale ine cient farms are only compared to e cient farms of a similar size [2]. The scale e ciency will also be obtained through dividing technical e ciency by pure technical e ciency [34]. In this study, the most comprehensive models of Data Envelopment Analysis, namely the return to constant scale model (CCR) and return to variable scale model (BCC) have been used to calculate e ciency. Each of these models include two input-oriented and output-oriented study directions. This means an ine cient DMU can turn into an e cient unit as a result of the reduction of the levels of input while the output is constant (input-oriented), or conversely, turn into an e cient unit by keeping the level of inputs constant and increasing the values of output (outputoriented) [37]. Choosing between an input-oriented study and an output-oriented study depends on the distinct characteristics of decision-making units being studied. In agricultural studies, input-oriented methods appear to be more appropriate as the number of outputs is limited, while several inputs are used for the production of agricultural crops and there is greater control over the consumption of inputs [31].
The variable return to scale model is used to calculate the pure technical e ciency and constant return to scale is used to calculate the technical e ciency [6].
The technical e ciency of an index is aimed at determining the e ciency of units based on the CCR model. The values of technical e ciency may range between zero and one. Technical e ciency is de ned in the Equation 10 [34].

Equation 10
In the Equation 10, u 1 , u 2 , …, u r are the weights given to outputs for j th unit (r=1,2,…,s); y 1 , y 2 , …, y r are the amount of outputs for j th unit; v 1 , v 2 , …, v i are the weights given to inputs for j th unit (i= 1,2,…,m) and x 1 ,x 2 ,…,x i are the amount of inputs for j th unit, and j (j=1,2,…,n) is the number of the decision-making unit.
In order to solve the Equation 10, linear programming has been used, which is presented in the Equation 11 [16].

Equation 11
Where θ is the technical e ciency, u is the weight of outputs, y represents the outputs, v is the weight of inputs, x represents the inputs, n is the number of decision-making units, s is the number of outputs and m is the number of inputs.
Moreover, to calculate the pure technical e ciency, the linear programming model has been used which is given in the Equation 12 [34].

Equation 12
Where z and u 0 are scalar and free in sign. Eventually, the scale e ciency was calculated through dividing the technical e ciency by pure technical e ciency [9]. The scale e ciency is indication of the impact of the size of decision-making units on the productivity of a system. It shows that some ine cient parts point to the inappropriate size of the decision-making units. If each decision making unit moves toward the best size, it is possible to improve the overall productivity (technical e ciency) to some extent as the level of technology (input) [38]. If a DMU is fully e cient in both CRR and BCC scores, it will operate at the most productive scale size. If a DMU has the full BCC score, but a low CCR score, then it is locally e cient but not globally e cient due to its scale size.
Thus, it is reasonable to characterize the scale e ciency of a DMU by the ratio of the two scores [44].
The relationship between scale e ciency, technical e ciency and pure technical e ciency is illustrated in the Equation 13.

Equation 13
In the present study, the e ciency of decision-making unit in paddy elds has been analyzed using the Dea-Solver software and the farms were assessed in terms of the consumption of energy. Subsequently, e cient and ine cient units in the consumption of energy were determined, and eventually the quantity of the consumption of inputs.

Greenhouse gas emissions
Agricultural production necessitates employing a multitude of input materials (fertilizers, biocides, seeds, etc.) and energy carriers (natural gas, diesel fuel, etc.). Production, formulation, storage, distribution of agricultural inputs and their applications with agricultural machinery lead to combustion of fossil fuel, and use of energy from alternative sources which emit CO 2 and other greenhouse gases(GHGs) into the atmosphere [26]. Carbon emission coe cients of agricultural inputs were applied to quantify the GHG emissions of rice production. GHG emission coe cients are depicted in Table 2. GHG emissions were worked out by multiplying the input application rate (diesel fuel, chemical fertilizers, machinery, pesticides, electricity and natural gas) by its corresponding emission coe cient. Table 2 Results And Discussion The average consumption of inputs and outputs and their total energy equivalent of paddy production in Mazandaran province are presented in Table 3. The average yield of paddy and rice straw equals to 5253.5 kg ha -1 and 1899.3 kg ha -1 respectively. Table 3 The total input and output energies were determined to be 91061.50 and 100967.95 MJ ha -1 . The contribution of consumed inputs to the total input energy is shown in Figure 2. Given the results obtained in this study, diesel fuel was considered to be the most widely used energy input with a share of 41.27%. Despite the limited application of agricultural machinery during growing stage, compared with the extensive application of machinery in the tillage stages, planting and harvesting, the diesel fuel came to be known as the most widely used energy input in the production of rice. After diesel fuel, the largest shares out of the total input energy in the production of paddy belonged to the energy inputs of machinery and chemical fertilizers, standing at 19.63% and 19.58%, respectively.
One of the reasons for the largeness of the share of the consumed energy of machinery input was the using of different tools for the tillage operation. After the inputs of fossil fuel, machinery and chemical fertilizers, the irrigation energy, with a 16.31% share out of the total input energy, was determined to be an energy-consuming input in the production of paddy in Mazandaran Province. Chemical fertilizers, human labor, seed and animal manure had the smallest share out of the total input energy with values of 1.19%, 1.09%, 0.9% and 0.02%, respectively.

Figure 2
The energy indices, i.e., energy use e ciency, energy productivity, speci c energy and net energy gain are presented in Table 4.
According to the results obtained, energy use e ciency was estimated to be 1.11. Energy productivity was calculated to be 0.058 kg MJ -1 . In fact, in order to produce one kilogram of paddy, 17.24 MJ of energy were consumed. Based on the studies, energy use e ciency for the production of rice using two methods of traditional and mechanized cultivation in the Mazandaran province stood at 1.72 and 1.63 respectively. In the same study conducted in Guilan province of Iran, the energy e ciency of rice production was obtained 1.53 [1,41]. Table 4 The positivity of the net energy demonstrated that the production of rice in studied area was justi able in terms of energy balance.
In Figure 3, shows the percentages of direct energy and indirect energy as well as renewable and non-renewable energies. The results indicated that with a consumption value of 74368.90 MJ ha -1 (81.67%) out of the total energy used, non-renewable energies retained a higher share than the renewable energy with a value of 16692.61 MJ ha -1 (18.33%). Results also showed the direct and indirect energy quantities involved in the production of paddy were calculated to be 53451.86 and 37609.64 MJ ha -1 , indicating the 58.70% share of direct energies and 41.30% share of indirect energies out of the total input energy involved in the paddy production in the province. The reason why the share of direct energy is higher than that of indirect energy is the high percentage of the consumption of irrigation and fuel inputs. According to studies carried out in the Guilan province, the shares of renewable and non-renewable energies from total consumed energy were found to be 11.22% and 88.78%, respectively. Moreover, 50.7% and 49.3% of the input energy represented the shares of indirect and direct energies respectively [41]. The results of technical e ciency, pure technical e ciency and scale e ciency of one hundred rice elds studied in the Mazandaran province using the inputoriented data envelopment analysis are given in Table 5 Therefore, the other 51 farms only had scale ine ciency. This showed the combination of inputs in these units was accurate; however, what has caused them to have e ciency score less than 1, was the lack of performance in the optimal scale. These items require a long-term planning to make the units e cient through moving toward optimal scale.
The results shown in Table 5 demonstrated that out 63 units of 66 technically ine cient farms, have increasing return to scale and 3 units have decreasing return to scale. In units with increasing return to scale, the production unit needed to be enlarged so that e ciency improves while in units with decreasing return to scale the production unit needed to be downsized. In other words, given the de nition of technical e ciency, the reduction of production scale for the above mentioned units with decreasing return to scale, leads to a growth of e ciency as a result of which, the e ciency of these units will increase. In the farms where there is a decreasing return to scale, adding for example 1% to the quantity of the inputs will result in the increasing of the crop to an extent less than 1%. Therefore, the managers of these farms should make efforts to increase their output to input proportion. Table 5   Table 6 shows the average technical e ciency, pure technical e ciency and scale e ciency of the paddy elds studied. The results indicated that the average values of these indices were 0.970, 0.996 and 0.974, respectively. Moreover, the standard deviations of these indices are shown in Table 6. The average technical e ciency of one hundred paddy elds investigated was determined to be 97%. In other words, they can reach the threshold of e ciency with 97 percent of production inputs and save three percent of inputs with an increase in their e ciency. As it can be seen from table 6, the technical e ciency score and scale e ciency range between 0.4812 and 1 and their standard deviation were greater than the pure technical e ciency score, indicating that farmers were not fully aware of the production techniques or did not employ the techniques accurately. Table 6 In Figure 4, the technical e ciency, pure technical e ciency and scale e ciency of one hundred rice producers studied in the Mazandaran province have been categorized in different groups. It can be seen that in the constant return to scale, 34 units retained absolute e ciency while more than 82% of the units had a technical e ciency score of more than 95%.

Figure 4
This gure illustrates that after removing the scale ine ciency in the variable return to scale method, 85 paddy elds turned e cient while 98% of the studied farms had a pure technical e ciency of more than 95%. 34 farms retained scale e ciency and 86% of the farms had a scale e ciency of more than 95%. Table 7  In the data envelopment analysis method, for each of the ine cient units, one unit or a combination of both or several e cient units are introduced as the pattern. The rest of the units could achieve the optimal level considering the weights (coe cients) given pertaining to each pattern unit. The unit No. 4 found to be the most ine cient unit with an e ciency of 48.12%. Since the pattern units of this unit are the farms No. 1, 3, 6, 13 and 46, it should consume, in accordance with the weight coe cients given in the Table 11, 0.039, 0.185, 0.0001, 0.173 and 0.055 of the production factors used in those units in order to become e cient. Therefore, considering the Table 8, unit 4 should lower the quantity of its seed energy, fuel, machinery, chemical fertilizers, animal manure, pesticide, irrigation and labor force by 422.36, 11,546.05, 3,828.32, 8,727.26, 0.08, 416.45, 5,235.06 and 847.77 MJ ha -1 to become an e cient unit. After optimization, the total input energy consumption of unit 4 declines from 58,280.11 to 27,256.77 MJ ha -1 and a total of 53.23% of energy would be saved. The fact that unit 4 should achieve maximum e ciency with these values of input in order to become e cient means that this unit should increase its rice production capacity, and for this purpose plant breeding and crop improvement are appropriate. Table 7   Table 8 The other production units whose e ciencies are less than 100 percent can identify their pattern unit. The units with an analysis like the one given above, can identify the analysis of inputs in order to achieve a certain level of output.
The optimal value of energy consumption and the quantity of energy saving for each of the inputs of rice production are presented in Table 9. These values have been calculated in terms of the variable return to scale model. The results showed the average quantity of energy consumption under optimal circumstances equals to 86115.829 MJ ha -1 , in compared to the status quo which is 4945.675 MJ ha -1 lower. This articulates that by taking into account the recommendations of this study and without reducing the current yield level of rice, it is possible to save 5.43% of consumed energy.
The highest quantity of saving was observed in the fuel consumption input energy at 2,796.09 MJ ha -1 followed by the energy of irrigation inputs and chemical fertilizers at 917.23 and 914.47 MJ ha -1 . Considering the detrimental environmental impacts of the fuel consumption and chemical fertilizers, the optimization and consumption management of these inputs can result in greater sustainability of rice production in the studied areas in addition to reducing the consumption of energy. Table 9 In Figure 5, shows the percentage of energy saving for each of the consumptive inputs of rice production out of the total savable energy. As it can be seen frm the gure, the fuel energy, irrigation and chemical fertilizers indicated the highest percentages of energy saving of %56.54, %18.55 and %18.49, respectively.
This demonstrated that in ine cient farms, these inputs were not employed appropriately and it is imperative that a suitable consumption pattern for these inputs is promoted in the areas. Singh et al. [47] investigated the consumption of energy in wheat production in different areas of India. They reported that with the optimal use of energy in the areas, it is possible to increase the performance of wheat between 4.2% and 22.3%.

Figure 5
The data envelopment analysis technique leads to the optimization of energy indices in the production of rice crop. The results of the optimization of the energy indices in the production of rice are presented in Table 10. As it can be seen, energy e ciency was calculated to be 1.172 under optimal circumstances, which shows a 5.38% improvement compared to the status quo (1.09). Moreover, the productivity of rice seed under the current and optimal circumstances was found to be 0.059 and 0.061 kg MJ -1 , respectively, indicating that with the optimization of energy consumption, the quantity of produced crop increases by 3.28 percent per each unit of consumed energy. Evidently, with the optimization of input energy, the share of indirect and non-renewable energies out of the total input energy increases. Moreover, the shares of direct energy and renewable energies have been on a declining course in a reverse manner and this is due to the large share of irrigation energy in improving energy. Table 10 The detail of the pure technical e ciency (PTE), the current use of energy, and the optimal energy required from different energy sources for the ine cient farmers are given in table 11. Using such data, it is possible to recommend that the decision making could reduce the quantity of input energies by taking into account the operational methods in which the output quantity (yield) remains constant. Therefore, the publication of these results contributes to the improvement of the productivity of energy for the production of rice in the studied area. In the last column of Table 15, the energy saving target ratios (ESTR) for 15 farmers affected by management ine ciency are presented. As can be seen, for the ine cient farmers, ESTR ranged between 0.2 percent (farmers No. 38) and 14.19 percent (farmer No. 84). The average energy saving target ratio stood at 5.44%, indicating that among the ine cient farmers, the farmers No. 38 was the best and the farmer No. 84 was the most ine cient. Table 12 shows the results of the analysis of paddy elds in the input-oriented variable return to scale model for the determination of excess inputs and the slacks of the yield of farms. In this table, it was determined for 56 ine cient combined units to what extent they should reduce the use of excess inputs so that they become e cient. For example, the farm No. 9, with a pure technical e ciency of 0.991, should rst reduce its entire inputs to 99.1 percent and then decrease its consumptive seed input to 230.49 units and its chemical pesticides to 72.16 units to be placed on the BCC e cient threshold. Table 12 shows 71.43% of paddy elds boast excessive consumption in the use of seed input, as is the case with 73.21% of farms in the consumption of fuel input, 37.50% in the use of machinery input, 32.14% in the use of chemical fertilizers, 19.64% in the use of animal manure, 28.57% in the use of pesticides input, 60.71% in the use of irrigation input and 28.57% in the use of human labor input. Table 11   Table 12 3

.2 Economic analysis
The production costs included the wages of the human labor, land preparation, agricultural inputs, the costs of transplanting, cultivation and harvesting, conversion and processing. In the present study, the conventional average rental fee for each hectare of paddy eld amounted to 823.53$ and the conventional interest rate was considered to be 15%. The results of this research showed that the gross production value and the total rice production cost in the Mazandaran province amounted to 3394.776 $ ha -1 and 2116.862 $ ha -1 , and the bene t to cost ratio was estimated to be 1.604 (Table 13). The xed costs comprised 37.5% and the variable costs comprised %62.5 of the total production costs. The net return resulting from the production of rice equaled to 1277.914 $ ha -1 , signifying the economic justi ability of the production of rice in the Mazandaran province. Table 13 In Table 13, the economic productivity of rice production was found to be 1.024 kg $ -1 . The energy intensiveness index showed that on average, per each dollar of rice production costs, 43.017 MJ of energy from different inputs would be supplied. The energy intensiveness value index expresses that on average, for the supplying of 26.824 MJ of energy from different energy inputs to be used, one dollar of revenue would be made. The energy intensity cost index showed that for each kilogram of produced rice, the equivalent value of consumptive energy would be 0.481$ and the energy cost ratio showed that for each unit of rice production cost, the equivalent value of the energy supplied from different inputs to be used in the paddy eld equaled 0.492 units. In the research by Pishgar et al. [41] the gross income, gross pro t and net pro t were obtained to be 4,095.6 $ ha -1 , 1,641.98 $ ha -1 and 939.71$ ha -1 , respectively. In their research, Pishgar et al. [41] reported the xed and changing costs of rice production in the Guilan province to be 702.27 $ ha-1 and 2,453.62 $ ha-1, respectively. They also reported the total cost of rice production in the Guilan province to be 3,155.89 $ ha-1 and 0.9 $ kg-1. The bene t to cost ratio equaled 1.3 and the economic productivity of rice production was expressed to be 1.12 $ kg -1 .

The results of greenhouse gas emissions of rice production
The quantities of GHG emissions under real and optimal circumstances are presented in Table 14. The The GHG emissions under current conditions obtained 4,251.33 kg CO 2 eq. ha -1 , while decreased to 4,146.21 kg CO 2 eq. ha -1 under optimal cultivation system. In other words, GHG emissions would be dropped by 2.7% if agricultural inputs used for rice production in the studies area were employed e ciently. The largest values of the mitigate of greenhouse gas emissions caused by the electricity, diesel fuel and machinery were obtained to be 35.76, 31.01 and 27.06 kg CO 2 eq. ha -1 .
The use of chemical fertilizers (particularly nitrogen fertilizer) more than what is required for the plant leads to the pollution of water and soil at the same time as emitting a great load of greenhouse gas emissions. Moreover, the emissions related to the consumption of diesel fuel have to do with the use of outdated tractors in different agricultural operations, the incompatibility of power capacities in the machinery and tractors and the extensive consumption of energy resulting from intense tillage operations.
The use of appropriate tillage operations such as minimum tillage method and the use of combined machinery would result in the reduction of fuel consumption. Moreover, using new electromotors with a high working safety, together with the farmers' awareness of the real water requirements throughout the different stages of rice production would result in the reduction of greenhouse gas emissions pertaining to the consumptive electricity. emissions in producing one hectare of chickpea and wheat were obtained to be 250 and 768 kg CO 2 eq. ha -1 , respectively.
Moreover, the proportion of the yield of crops to the total greenhouse gas emissions in the chickpea and wheat crops were estimated to be 12.6 and 3.68 kg (kg CO 2 eq.) -1 , which was indicative of the higher yield of chickpea as compared to wheat [23].
In a study performed to calculate the greenhouse gas emissions of wheat production in Isfahan province, electricity and nitrogen fertilizer boasted the largest consumptive shares among other inputs with 2,400.7 (74%) and 371.2 (14%) kg CO 2 eq. ha -1 , respectively [28].
As it can be seen in Table 14, the Data Envelopment Analysis method was capable of reducing the total greenhouse gas emissions by 105.118 kg CO 2 eq. ha -1 . In a study on the reduction of greenhouse gas emissions resulting from strawberry production in the Guilan province, it was shown that the Data Envelopment Analysis method is capable of reducing 5,774 kg CO 2 eq. ha -1 of cultivation of the crop. Moreover, natural gas, with a 65% share, had the highest potential for the reduction of greenhouse gas emissions [25].
In a study on the amount of the emission of greenhouse gases resulting from cultivating Sorghum in the Sistan region, electricity retained the highest quantity of greenhouse gas emissions with 2,981.27 kg CO 2 eq. ha -1 out of a total of 3,746.7 kg CO 2 eq. ha -1 of greenhouse gases. After that, animal manure and diesel fuel retained the highest quantity of greenhouse gas emissions at 277.72 and 258.83 kg CO 2 eq. ha -1 . Table 14 Figure 6 illustrates the share of inputs out of the reduction of greenhouse gas emissions after the improvement of consumption. The electricity input, diesel fuel and machinery had the largest share in the reduction of greenhouse gas emissions resulting from the production of rice at 34.02%, 29.5% and 25.75%, respectively.

Figure 6
In another study the optimization of GHG emissions associated with plant and ratoon farms of sugarcane in the Khuzestan province has been studied using the Data Envelopment Analysis method. Based on the results obtained, the total greenhouse gas emissions in one hectare of plant and ratoon farms amounted to 5,825.25 and 4,310.7 kg CO 2 eq. where electricity, diesel fuel and nitrogen fertilizer in the plant farms retained the largest shares out of the total greenhouse gas emissions at 74.37%, 17.22% and 4.1%, respectively, while in the ratoon farms, electricity and diesel fuel had the largest shares at 79.29% and 11.7%. The emission of greenhouse gases in the plant and ratoon farms under optimal circumstances dropped by 633.12 and 110.01 kg CO 2 eq.

Conclusion
In the present study, the total input and output energies in the rice production process in the Mazandaran province were determined to be 91,061.50 and 100,967.95 MJ ha -1 , respectively. The input of diesel fuel, boasting a 41.27% share out of the total input energy, is considered to be the most highly consumed energy input. The reason for that is the extensive use of rice machinery and tools in different stages of land preparation, transplanting, cultivation and harvesting of rice due to the special characteristics of paddy production. The energy use e ciency and energy productivity of rice in the Mazandaran province were calculated to be 1.11 and 0.058 kg MJ -1 . The positivity of the net energy demonstrates that the production of rice in the Mazandaran province is justi able in terms of energy balance. Non-renewable and renewable energies respectively accounted for shares of 81.67% and 18.33% out of the total input energy in the production of paddy while direct and indirect energies accounted for shares of 58.70% and 41.30%. The reason why the share of direct energy is larger than that of indirect energy is the high percentage of the consumption of irrigation inputs and fuel.
The average technical e ciency, pure technical e ciency and scale e ciency of one hundred rice producers were obtained %97, %99.6 and %97.4, respectively. With the improvement of the consumption of inputs, the average energy consumption quantity will be 86,115.829 MJ ha -1 which is 4,945.675 MJ ha -1 less than the status quo. Therefore, considering the recommendations of this study and without reducing the current outputs level, it is possible to save %5.43 of energy. The highest rate of saving was observed in the fuel input energy valued at 2,796.09 MJ ha -1 , followed by the energy of irrigation inputs and chemical fertilizers with 917.23 and 914.47 MJ ha -1 , respectively. Given the detrimental environmental impacts of using fuel and chemical fertilizers, optimization and consumption management of these inputs can lead to increased sustainability in rice production in the area studied as well as reducing the consumption of energy.
Under optimal conditions, the energy e ciency will be 1.172, which shows a 5.38% increase compared to the current 1.09 value. Moreover, the quantity of energy productivity of rice under the current and optimal circumstances were obtained to be 0.059 and 0.061 kg MJ -1 , respectively. As a result, by optimizing the consumption of energy, the quantity of produced crops increases by 3.28 % for each unit of consumed energy. Evidently, by optimizing the input energy, the share of indirect and non-renewable energies in proportion to the total input energy has increased, and also the share of direct energy and renewable energies in proportion to the reverse state has experienced a decline, which is due to the large share of irrigation energy in improving energy.
The bene t to cost ratio stood at 1.604 and the net return resulting from the rice production equaled to 1277.914 $ ha -1 , signifying the economic justi cation of the production of rice in the Mazandaran province. On average, per each dollar of rice production cost, 43.017 MJ of energy is consumed from different production inputs in paddy elds. On average, for the entry of 26.824 MJ of energy of different production inputs into the paddy elds of Mazandaran province, one dollar of revenues would be made. Estimates on the energy intensity cost index revealed that the equivalent value of consumed energy for the production of each kilogram of crop is 0.481 $. The energy ratio cost index showed that for each unit of rice production cost, the equivalent cost of the energy procured from different production units for being used in the rice eld equals 0.492.
With the optimization of the consumption of inputs, the greenhouse gas emissions drop by 2.7 percent. The highest reductions in the greenhouse gas emissions for the electricity input, diesel fuel and machinery were found to be the equivalent of 35.76, 31.01 and 27.06 kg CO 2 eq. ha -1 . The inputs of electricity, diesel fuel and machinery, with values of 34.02%, 29.5% and 25.75% represented the highest shares in the reduction of greenhouse gas emissions.
The optimization of the fuel consumptive energy, agricultural machinery, chemical fertilizers, most notably nitrogen and water for irrigation are among the important pathways to the better management of energy in producing rice in the paddy elds of the province studied. Replacing machinery and new tools with outworn machinery will result in the reduction of energy consumed by the machinery and diesel fuel and result in the lessening of greenhouse gas emissions.
In order to reduce the fuel consumption and machinery energy, striking a balance between the machinery and tractor and lessening the extremely high energy resulting from the tillage operation is necessary.
Using suitable tillage methods such as conservation tillage, dry working and using combined machinery result in the reduction of fuel consumed in the province. The ine ciency of irrigation systems, using outworn electromotors and the farmers' unawareness about the plant's genuinely required water are among the reasons for the growth of consumptive energy in the region studied. Therefore, the revision of outworn and dilapidated systems in the irrigation sector and using modern irrigation methods are recommended.
The management of the use of chemical fertilizers, applied research to determine the extent of the need of plants for nutrients during the different stages of growth and determining the appropriate quantity of chemical fertilizers required by the soil using the soil test play a substantial role in reducing the greenhouse gas emissions in the region studied as well as reducing the consumptive energy of this input. In order to reduce the emission of greenhouse gases, efforts should be made to optimize the consumption of chemical fertilizers and replace the organic fertilizers with chemical fertilizers. Reducing the use of chemical pesticides and using of bio-inhibitors instead of chemical pesticides is effective in optimizing the consumption of energy and the reduction of greenhouse gases.           Figure 1 Geographical location of the studied rigion (Mazandaran province in north of Iran). Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors. The share of total mean energy inputs for rice production in Mazandaran province, Iran Figure 3 The share of total energy input in the form of direct, indirect, renewable and non-renewable source of rice production in Mazandran province, Iran. Contribution to the total saving energy of rice producer in Mazandaran province, Iran Figure 6 The share of each input emission reduction of for GHG rice production.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.