تجزیه و تحلیل حساسیت از مجموع شدت انتشار CO2 برآورد شده با استفاده از جدول داده ستانده

This study addresses the problem of uncertainties of total CO2 emission intensities estimated using an input–output (I-O) table for a life-cycle inventory analysis. The validity of the total CO2 emission intensities has been questioned mainly because the amounts of commodities are measured in monetary units, and because various commodities are produced in a single sector of an existing I-O table. In this study, first, a sensitivity analysis of total CO2 emission intensities estimated using the Japanese I-O table was performed to identify the elements that significantly influence the total CO2 emission intensity. If an influential element identified by sensitivity analysis varies widely, the total CO2 emission intensity is greatly influenced. Secondly, how much total CO2 emission intensities vary associated with the variation of those influential elements was evaluated. It was concluded that the evaluation of uncertainties using a stochastic approach as well as the improvement of accuracy by disaggregating the original I-O sectors, focusing on influential elements, are important.

Life-cycle assessment (LCA) is an increasingly important tool to evaluate the environmental impacts associated with products, services and technologies. In performing life-cycle inventory (LCI) analysis, which is a basic component of LCA, process analysis has been used as the most common one among LCA practitioners. On the other hand, an input-output (I-O) table has been applied because it theoretically enables one to estimate both the direct and indirect emissions induced by the production of goods and services. However, the applicability of total emission intensities estimated using an I-O table in performing LCI studies has been questioned because of uncertainties associated with the inherent characteristics. For example, regarding Japanese existing I-O tables, a range of goods and services that exist in society are classified into about 400 sectors. Therefore, the calculated total emission intensity corresponds to an average good virtually produced in a single I-O sector. In addition, all goods and services are measured not in physical units (e.g. kg, kcal) but in monetary units (‘yen’ in Japanese I-O tables). Taking these characteristics of I-O tables into consideration, it is reasonable to suppose that total emission intensities estimated using an I-O table are uncertain with some degree of variation. However, the degree of variation has not been evaluated so far. Therefore, some people remain skeptical about the application of the total emission intensities estimated using an I-O table in performing a LCI for a specific product. The question discussed in the present paper is about uncertainties arising from the inherent characteristics of an existing I-O table. It is the aim of this study to evaluate uncertainties of total emission intensities estimated using an I-O table and to examine the sources of these uncertainties. The present paper is organized as follows: first, the sources causing uncertainties of input coefficients and direct emission intensities are shown these contribute to the variation of total emission intensities. Secondly, by performing a sensitivity analysis, those elements among all the input coefficients and all the direct emission intensities that significantly influence the total CO2 emission intensity for each good or service are identified. When an influential element identified by sensitivity analysis varies widely, the total emission intensity is greatly influenced. Thirdly, the variations of some influential elements are quantitatively estimated, and then it is roughly calculated how much total emission intensities vary associated with the variations of these influential elements. Finally, conclusions are drawn in terms of the handling of uncertainties in the total emission intensities estimated using an I-O table.

Input coefficients and direct emission intensities, which suffer from various uncertainties, are the basic elements for estimating total emission intensities using an I-O table. In this study, first, the contributions of variations of all the elements to the total CO2 emission intensity for each commodity were quantitatively evaluated using the concept of rate sensitivity. Although the uncertainty of total CO2 emission intensities has partially been performed [11], this study attempted a systematic explanation in order to obtain an entire understanding on the uncertainty of total CO2 emission intensities. The present paper shows, as an example, the results of a sensitivity analysis of total CO2 emission intensities for industrial products. First, the variation of direct CO2 emissions from electricity and iron/steel production is strongly linked with uncertainties of total CO2 emission intensities. Secondly, the variation of input coefficients relevant to iron/steel production and processing/assembly are influential with respect to total CO2 emission intensities. In this study, although a sensitivity analysis was performed based on the assumption that all elements are independent of each other, the results would change if the dependence among elements was considered. In addition, it is debatable whether the rate sensitivity is appropriate or not as an index of uncertainty when elements that vary widely are dealt with.8 The development of a quantitative index considering these two questions is a task ahead. Secondly, we evaluated to what extent the total CO2 emission intensities vary associated with the variation of some influential elements that were identified by a sensitivity analysis. For example, the input coefficients relevant to iron/steel production, which is one of the influential elements, can vary widely due to the ‘multiple goods to one sector’ and ‘multiple technologies to one good’ uncertainties. This can cause large variations of total CO2 emission intensities for industrial products. The existence of uncertainty does not deny the effectiveness of an I-O table. The control of uncertainties makes the best use of the features that an I-O table covers all transactions of all commodities, and that a commodity described in an I-O table is not ‘specific’ but ‘average'. Therefore, it is important to evaluate reliabilities by sensitivity/uncertainty analysis, which can contribute to practical and effective use of an I-O table within a LCI. In order to obtain reliable results using an I-O table, in conclusion, there are two approaches: the evaluation of uncertainty by a stochastic approach and the improvement of accuracy by disaggregating original I-O sectors in more detail. The balance of both approach is as important. Although not discussed in the present paper, I-O based sensitivity analysis is available for generating a preliminary LCI before a process-type inventory analysis [13]. By employing this technique, an appropriate selection of a system boundary can be obtained not by subjective judgment but based on quantitative and objective information. In addition, ranking flows in the product system studied in order of importance by this technique can effectively collect inventory data. In performing a LCI, as a result, time and expense can be reduced without sacrificing the reliability of the LCI results.