The overlap technique is often used when a new method is introduced but data are not available to apply the new method to the early years in the time series, for example when implementing a higher tier methodology. If the new method cannot be used for all years, it may be possible to develop a time series based on the relationship (or overlap) observed between the two methods during the years when both can be used. Essentially, the time series is constructed by assuming that there is a consistent relationship between the results of the previously used and new method. The emission or removal estimates for those years when the new method cannot be used directly are developed by proportionally adjusting the previously developed estimates, based on the relationship observed during the period of overlap. In this case, the emissions or removals associated with the new method are estimated according to Equation 5.1:2

Equation 5.1

Recalculated emission or removal estimate computed using the overlap method

7-ÏÏ • Z A

{(n - m +1) i=m x, y

y0 = the recalculated emission or removal estimate computed using the overlap method x0 = the estimate developed using the previously used method y and x are the estimates prepared using the new and previously used methods during the period of overlap, as denoted by years m through n

A relationship between the previously used and new methods can be evaluated by comparing the overlap between only one set of annual estimates, but it is preferable to compare multiple years. This is because comparing only one year may lead to bias and it is not possible to evaluate trends.

Figure 5.1 shows a hypothetical example of a consistent overlap between two methods for the years in which both can be applied. In Figure 5.2 there is no consistent overlap between methods and it is not good practice to use the overlap technique in such a case.

Other relationships between the old and new estimates may also be observed through an assessment of overlap. For example, a constant difference may be observed. In this case, the emissions or removals associated with the new method are estimated by adjusting the previous estimate by the constant amount equal to the average difference in the years of overlap.

Figure 5.1 Consistent overlap

Figure 5.1 Consistent overlap

2 Overlap Equation 5.1 is preferred to the equation described in Good Practice Guidance for National Greenhouse Gas Inventories (GPG2000, IPCC, 2000):

V i = m / i = m because the latter gives more weight to overlapping years with the highest emissions. However in practical cases the results will often be very similar and continued use of the previous equation is consistent with good practice where its use gives satisfactory results.

Figure 5.2 Inconsistent overlap

Figure 5.2 Inconsistent overlap Surrogate data

The surrogate method relates emissions or removals to underlying activity or other indicative data. Changes in these data are used to simulate the trend in emissions or removals. The estimate should be related to the statistical data source that best explains the time variations of the category. For example, mobile source emissions may be related to trends in vehicle distances travelled, emissions from domestic wastewater may be related to population, and industrial emissions may be related to production levels in the relevant industry. See Chapter 2, Approaches to Data Collection.

In its simplest form, the estimate will be related to a single type of data as shown in Equation 5.2:

Equation 5.2

Emission/removals trend estimates using surrogate parameters


y = the emission/removal estimate in years 0 and t s = the surrogate statistical parameter in years 0 and t

Although the relationship between emissions/removals and surrogate can be developed on the basis of data for a single year, the use of multiple years might provide a better estimate.

Box 5.2 provides an example of the use of surrogate data for estimating methane emissions from underground coal mining in the United States. In some cases, more accurate relationships may be developed by relating emissions to more than one statistical parameter. Regression analysis may be useful in selecting the appropriate surrogate data parameters. Using surrogate methods to estimate otherwise unavailable data can improve the accuracy of estimates developed by the interpolation and trend extrapolation approaches discussed below.

Box 5.2

Case study of surrogate data -Methane emissions from underground coal mining in the United States

On a quarterly basis, the U.S. Mine Safety and Health Administration (MSHA) measures methane emissions levels at underground mines with detectable levels of methane in their ventilation air. USEPA uses these measurements as a basis for calculating national emissions from underground coal mining. These data were not available for the years 1991-1992, however, because of restructuring within the Department of Labor. To estimate emissions for these years, USEPA used total underground coal production as a surrogate data set. The graph below shows the relationship between underground coal production and measured emissions, which are closely but not perfectly correlated. Differences reflect the fact that individual mines vary greatly in their emission rates, and as production levels at mines change over time, the weighted average emission rate also changes. USEPA applied Equation 5.2 to estimate emissions for 1991 and 1992 using Tier 3 emissions data and coal production for 1990. These data points are crossed by the dashed line in the graph. Note that this procedure is very similar to an overlap with the Tier 1 method because coal production is the recommended activity data for Tier 1. Comparison of implied emission factors from estimates using surrogate data with Tier 1 default factors would be a useful QA/QC check.

Surrogate Data for Coal Mining in the United States

450,000 400,000 350,000 300,000 250,000 200,000 150,000 100,000 50,000

450,000 400,000 350,000 300,000 250,000 200,000 150,000 100,000 50,000

o Id

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Year

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Year

Iu o Id

-Coal Production

■Measured Emissions Interpolation

In some cases it may be possible to apply a method intermittently throughout the time series. For example, necessary detailed statistics may only be collected every few years, or it may be impractical to conduct detailed surveys on an annual basis. In this case, estimates for the intermediate years in the time series can be developed by interpolating between the detailed estimates. If information on the general trends or underlying parameters is available, then the surrogate method is preferable.

Figure 5.3 shows an example of linear interpolation. In this example, data for 1994 and 1995 are not available. Emissions were estimated by assuming a constant annual growth in emissions from 1993-1996. This technique is appropriate in this example because the overall trend appears stable, and it is unlikely that actual emissions for 1994 and 1995 are substantially different from the values predicted through interpolation. For categories that have volatile emission trends (i.e., they fluctuate significantly from year to year), interpolation will not be according to good practice and surrogate data will be a better option. It is good practice to compare interpolated estimates with surrogate data as a QA/QC check.

Figure 5.3

Linear interpolation










« 1







■ Interpolation

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000


Trend extrapolation

When detailed estimates have not been prepared for the base year or the most recent year in the inventory, it may be necessary to extrapolate from the closest detailed estimates. Trend extrapolation is conceptually similar to interpolation, but less is known about the actual trend. Extrapolation can be conducted either forward (to estimate more recent emissions or removals) or backward (to estimate a base year). Trend extrapolation simply assumes that the observed trend in emissions/removals during the period when detailed estimates are available remains constant over the period of extrapolation. Given this assumption, it is clear that trend extrapolation should not be used if the change in trend is not constant over time. In this situation it will be more appropriate to consider using extrapolations based on surrogate data. Extrapolation should also not be used over long periods of time without detailed checks at intervals to confirm the continued validity of the trend. In the case of periodic data, however, extrapolations will be preliminary and the data point will be recalculated at a later stage.

Box 5.3 in this Section shows an example in which activity data for forests are available only at periodic intervals, and data for the most recent years are not yet available. Data for recent years can be extrapolated on the basis of a consistent trend, or on the basis of appropriate data. It should be noted, however, that the uncertainty of the extrapolated estimates increases in proportion to the length of time over which the extrapolation is made. Once the latest set of periodic data becomes available, it will be necessary to recalculate the part of the time series that had been estimated using trend extrapolation.

The example in Box 5.3 assumes a linear extrapolation, which is likely to be appropriate for the forest land category. Non-linear extrapolations are possible, and may be more appropriate given an observed trend, (e.g., exponential growth in the use of ODS Substitutes). Countries using non-linear extrapolation should provide clear documentation for the choice and explain why it is more appropriate than linear extrapolation.

Box 5.3

Case study on periodic data, using extrapolation

Consider a case where a national forest inventory is conducted every 5 years. Estimates of several types of required data (e.g., tree growth) will therefore only be obtained at certain intervals. On the assumption that growth is on average reasonably stable between years, inventory estimates for the years after the last available data should be made using extrapolations of past estimates (i.e., tree growth trends). As shown in the figure below, a biomass estimate for 2005 for a plot is obtained in this way, although the latest measurement was made in 2000. The trend between 1995 and 2000 is simply extrapolated linearly. In practice, a log scale might be used to accommodate exponential behaviour but this is not considered for this simple example. Also, extrapolation can be improved using surrogate data or more sophisticated modelling taking into account parameters influencing the parameter we want to extrapolate.

Linear Extrapolation in AFOLU

■Actual (Periodic) Data) - - • - - Original Extrapolation

65 60

50 45 40

■Actual (Periodic) Data) - - • - - Original Extrapolation

65 60

50 45 40



1995 Year





1995 Year



Unlike periodically available data, when data are not available for the first years in the time series (e.g., base year and pre base year data on for example waste disposal and land use) there is no possibility of filling in gaps with future surveys. Trend extrapolation back in time is possible but should be done in combination with other splicing techniques such as surrogate data and overlap. Some countries that have undergone significant administrative and economic transitions since 1990 do not have consistent activity data sets for the entire time series, particularly if national data sets covered different geographic areas in previous years. To extrapolate backwards in these cases, it is necessary to analyze the relationship between different activity data sets for different periods, possibly using multiple surrogate data sets. OTHER TECHNIQUES

In some cases, it may be necessary to develop a customised approach to best estimate the emissions over time. For example, the standard alternatives may not be valid when technical conditions are changing throughout the time series (e.g., due to the introduction of mitigation technology). In this case, it will be necessary to carefully consider the trends in all factors known to influence emissions or removals over the period. Where customised approaches are used, it is good practice to document them thoroughly, and in particular to give special consideration to how the resultant emissions estimates compare to those that would be developed using the more standard alternatives. SELECTING THE MOST APPROPRIATE TECHNIQUE

The choice of splicing technique involves expert judgement, and depends on an expert assessment of the volatility of emissions trend, the availability of data for two overlapping methods, the adequacy and availability of surrogate data sets, and the number of years of missing data. Table 5.1 summarises the requirements for each technique and suggests situations in which they may or may not be appropriate. Countries should use Table 5.1 as a guide rather than a prescription.

Table 5.1 Summary of splicing techniques





Data necessary to apply both the previously used and the new method must be available for at least one year, preferably more.

• Most reliable when the overlap between two or more sets of annual estimates can be assessed.

• If the trends observed using the previously used and new methods are inconsistent, this approach is not good practice.

Surrogate Data

Emission factors, activity data or other estimation parameters used in the new method are strongly correlated with other well-known and more readily available indicative data.

• Multiple indicative data sets (singly or in combination) should be tested in order to determine the most strongly correlated.

• Should not be done for long periods.


Data needed for recalculation using the new method are available for intermittent years during the time series.

• Estimates can be linearly interpolated for the periods when the new method cannot be applied.

• The method is not applicable in the case of large annual fluctuations.

Trend Extrapolation

Data for the new method are not collected annually and are not available at the beginning or the end of the time series.

• Most reliable if the trend over time is constant.

• Should not be used if the trend is changing (in this case, the surrogate method may be more appropriate).

• Should not be done for long periods.

Other Techniques

The standard alternatives are not valid when technical conditions are changing throughout the time series (e.g., due to the introduction of mitigation technology).

• Document customised approaches thoroughly.

• Compare results with standard techniques.

Was this article helpful?

0 0
Guide to Alternative Fuels

Guide to Alternative Fuels

Your Alternative Fuel Solution for Saving Money, Reducing Oil Dependency, and Helping the Planet. Ethanol is an alternative to gasoline. The use of ethanol has been demonstrated to reduce greenhouse emissions slightly as compared to gasoline. Through this ebook, you are going to learn what you will need to know why choosing an alternative fuel may benefit you and your future.

Get My Free Ebook

Post a comment