Review of New Zealand's 2007 Net Position Report

• Executive Summary
• 1 Introduction
• 2 General Review
• 3 Sector Reviews
• 4 Uncertainties
• 5 Conclusions and Recommendations
• 6 References

3 Sector Reviews

3.1 Energy

3.1.1 Modelling approach

Projections of emissions from the energy sector are derived from the Ministry for Economic Development’s SADEM model. As a partial equilibrium model, SADEM describes energy supply by fuel and energy demand by sector, together with a more complex electric power simulation tool. Hence, it is a collection of modules, interacting with other modules. On the supply side, while oil and coal supply are assumed to be perfectly elastic, the natural gas market operates in equilibrium. On the demand side, five sectors have been defined: residential, transport, heavy industry, other (light) industry and commercial demand. All sectoral energy demands (except transport) are fed into the electric power and renewables module as well as fuel prices. This section will focus on energy emissions from the residential, commercial, heavy industry (excluding specific emissions from industrial processes) and other (light) industry sectors, as well as providing general comment on the SADEM model and the modelling of electricity generation. Transport and industrial process sectors are covered in more detail in sections 3.2 and 3.3 respectively.

Our conversations with the Ministry of Economic Development indicated that emissions from electricity generation over the first commitment period is forecast by the Electricity Commission’s Generation Expansion Model (GEM), a shorter-term model focused on the New Zealand electricity sector.³ GEM is a more detailed sector-specific electricity model used for grid planning analysis. Beyond the Kyoto period, the SADEM model is used.

Since the review of 2005 net position report, the Ministry of Economic Development has focused on implementing recommendations around the commercial and industrial sectors. Improvements include the disaggregation of the ‘other industrial and commercial’ model into two separate models. Within the ‘heavy industry’ sector, there are industry-specific models for metals (steel and aluminium), petrochemicals (urea and methanol), refining, forestry and dairy. These industry sub-models have been significantly redeveloped since 2005, based on the findings of a 2006 report on heavy industry energy demand commissioned by the Ministry for Economic Development (Covec et al, 2006). This report provided a basis for more detailed industry-specific (bottom-up) modelling and informed the development of a separate model for the dairy sector.

The industrial and commercial sectors together have a much greater influence on emissions projections than the residential sector. The Ministry for Economic Development’s work on the industrial and commercial models represents a significant improvement to the modelling of energy emissions in this year’s net position. The residential sector has had less attention, due to its relatively low impact on overall emissions projections.

In projecting emissions from the residential, commercial and industrial sectors, the SADEM model is used. A two-step approach is employed, first estimating overall demand and then attributing this demand out to individual fuels. Yearly energy saving potentials are also calculated off line for the residential and commercial sectors, and then subtracted from projections. The 2007 projections also include policies on which Cabinet has made a substantive decision. One of these is the solar water-heating programme, which was not included in the 2006 net position report.

In addition to these domestic emissions, an international policy instrument that may effect New Zealand’s net emissions is “Projects to Reduce Emissions” (PREs). PREs provide an incentive, in the form of Kyoto Protocol emission units, for projects that reduce emissions below business as usual levels during the first commitment period. As such, they also contribute to achieving the outcomes of the National Energy Efficiency and Conservation Strategy (NEECS). As the vast majority of all projects funded are in the electricity sector, PREs are applied to this sector. They are modelled as an effective discount on the capital costs for new generation projects at the margin of viability. The 2007 net position report states that tradable emissions credits have been allocated for a number of projects already.

3.1.2 Reasonableness of approach taken

The major input assumptions to SADEM, including GDP, exchange rates and fossil fuel prices appear reasonable. Macroeconomic forecasts have been taken from Treasury projections, and fossil fuel prices are consistent with those used in New Zealand Energy Strategy. Since 2005, particular care has been taken in addressing the uncertainty in future fossil fuel prices, and several studies have been carried out on prices of coal, oil and gas, as discussed in the next section.

From a modelling perspective, the SADEM model gives a reasonable depiction of the specifics of the New Zealand energy sector and hence is an appropriate tool for projecting energy related greenhouse gas emissions. In order to represent electricity sector in greater detail, the GEM is used in the 2007 net position analysis. GEM was originally developed to support the derivation of Grid Planning Assumptions (GPAs), a regulatory requirement on the Commission in undertaking its oversight of Transpower’s proposed grid investments. It is a capacity-planning model with a time span of 20 to 40 years. In particular, from a given list of potential new generation plants, it determines which ones to build in what year.

Each plant option is characterized by parameters such as location, technology, fuel type, capacity, capital cost, and operating costs. GEM is run under a series of scenarios for given levels of electricity demand and fuel availability as well as being subject to Government energy policies. The model minimises aggregate capital and operating costs, while satisfying a number of technical, physical and economic constraints.

In estimating emissions from the residential sector, firstly overall demand is estimated and then it is distributed to individual fuels. The price of fuel, historical levels of energy demand and GDP, and degree days are used as the explanatory inputs. Even though this type of top-down approach in analysis of energy use is not uncommon, due to saturation effects, historical relationships between energy demand and other factors, such as GDP, may break down. Also, the 2005 review expressed concerns over way in which the influence of degree days on energy demand was estimated, as equal weighting was given to both cooling and heating degree days. In reality, the former would presumably necessitate electric air conditioning while the latter may require other fuels for space heating. A report commissioned by the Ministry of Economic Development in 2005 (Covec, 2005) highlighted the statistical insignificance of degree days on model results, and therefore the Ministry of Economic Development have placed more focus on improving modelling of the commercial and industrial sectors. They report that this issue will be resolved as part of future work to redevelop the residential model.

The emissions projections from the residential sector have been reviewed to include the new Government policies that are in place, for example the solar water heating programme. An approximate saving of 2100 kWh per unit per annum is assumed for the 2007 projections. Hence, during the first commitment period, a total saving of 37.8 GWh is calculated.⁴ While it is not uncommon to calculate such savings off line from an energy model, a technologically explicit model would provide physical interpretations for energy use and energy saving potentials. The 2007 projections also include overall 1.70 PJ of energy savings per year during the first commitment period due to residential energy efficiency improvements⁵, but it is not clear how this figure is derived.

3.1.3 Implementation of previous recommendations

The 2005 review included a list of modelling improvements that the Ministry of Economic Development had committed to deliver. Table 1 summarises this list and the improvements made since then, based on information provided in the 2007 net position report, information on the Ministry of Economic Development’s website and communication with experts in the Ministry of Economic Development.

Table 1: Modelling improvements that MED committed to deliver following the 2005 review*

Improvements, rated A	Implementation
Energy outlook (including assumptions document, model runs, and write-up)	Completed1
NIWA model runs	Completed
Fourth national communication	Completed2
Input to post-Maui analysis	Completed
PRE analysis	Completed
Oil price paper	Completed3
Emerging supply side technologies project	Completed4
Two sections of model documentation requiring updating	In progress
Gas and coal assumptions papers	Completed5
Improvements to non-carbon dioxide emissions from energy use	Completed
Improvements, rated B
Energy outlook workshops	Completed: x 3 stakeholders engagement workshops were held between 2005 and 20066
Integrating demand models and fuel shares models for residential sector	Not implemented
Greenhouse gas paper	Completed7
Additional modelling papers
LNG pricing	Completed8
Alternative liquid fuels	Completed9

* While ‘A’ rated improvements include all the issues mentioned in the 2005 review, ‘B’ rated improvements are selective as many of them are on modelling of electricity generation in SADEM, which is replaced by GEM model for first commitment period in 2007 review.

1 For details, please see http://www.med.govt.nz/energy/eo/

2 For details, please see http://www.mfe.govt.nz/publications/climate/national-communication-2006/

3 For details see http://www.med.govt.nz/templates/Page____10612.aspx

4 For details see http://www.med.govt.nz/templates/MultipageDocumentTOC____20918.aspx

5 Part of energy outlook workshop presentation, for details see http://www.med.govt.nz/upload/27041/gas-coal.pdf

6 For details, please see http://www.med.govt.nz/templates/StandardSummary____15288.aspx

7 For details, please see http://www.med.govt.nz/templates/MultipageDocumentTOC____17799.aspx

8 For details, please see http://www.med.govt.nz/templates/MultipageDocumentTOC____23939.aspx

9 For details, please see http://www.med.govt.nz/templates/MultipageDocumentTOC____26587.aspx

3.1.4 Treatment of uncertainty

The 2007 net position analysis report provides a non-exhaustive list of factors that could affect actual emissions from the energy sector during the first commitment period. The extensive list includes all the major sources of uncertainty.

In addition to these, we suggest that the possible effects of energy efficiency programmes are expanded upon. Energy efficiency results in cost savings, hence allowing consumers to be able to use more energy, the so-called rebound effect. The academic literature identifies three types of rebound effects: i) increases in demand for the same type of energy services; ii) increases in demand for different types of energy services; and iii) emerging of demand for new types of energy services (for example, paying lower power bills in winter due to more efficient boiler and appliances, and hence being able to afford a winter holiday in a ski resort).

In addition to the uncertainty about the actual size of the energy savings which might be realised, there is uncertainty about the composition of these energy savings. SADEM assumes an aggregate energy saving of 1.70 PJ in the residential sector, but the actual composition of these savings⁶ (for example, whether they would be realised in space heating or lighting or another type of end-use category) could depend on a number of factors.

The extensive list of sources of uncertainty in the energy sector projections (including the transport sector) is accompanied by an upper and lower emission scenario, with 187.7 million tonnes carbon dioxide equivalent and 162.8 million tonnes carbon dioxide equivalent reported respectively. The upper and lower scenarios were included as requested in the net position 2007 report to indicate the range of uncertainty associated with emissions projections, rather than to provide a detailed sensitivity analysis. However, these scenarios are defined by different sets of variables rather than changing a single variable as inputs to scenario, hence making it difficult to disentangle the relationships between these selected variables. An alternative approach would be running a large number of sensitivity runs on individual parameters in order to get a better understanding of which changes have the most impacts on the results. It should be noted that some sensitivity analysis of energy sector emissions has already been undertaken in the Ministry of Economic Development’s New Zealand’s Energy Outlook to 2030. These then can be used in a Monte Carlo type of analysis to obtain a distribution of results.

Further recommendations on treating uncertainty are included in section 4.

3.1.5 Proposals for future improvements

A brief review of the GEM model indicates that it is reasonable to use it in projecting emissions from the energy sector. It contains a simple representation of the electricity network by simulating two nodes, the North and South Islands. Hence, transmission losses on the High Voltage Direct Current (HVDC) are modelled using average loss factors, whereas alternating current (AC) losses are accounted for by adjusting load.

Currently, it does not allow for hydro storage, therefore inflows are used or spilled contemporaneously, whereas the previously used SADEM model did employ a stochastic approach on availability of hydro resources. However, this issue is partially addressed by allowing the user to select among 74 annual hydro inflow sequences (based on historical data from 1932), even though one might argue that future inflow pattern might be different from the historical ones. Equally important is improving the representation of transmission and distribution losses in GEM.⁷ Nonetheless, although simple in its approach, our brief review indicates that these improvements should have limited effects on GEM outputs for the first commitment period of the Kyoto Protocol.

On the other hand, the projection of emissions from the residential sector is conducted in a top-down manner. However, bottom-up methods are very useful for policy analysis, particularly for non-price instruments. Hence, we still believe that use of bottom-up models to compare (if not incorporate) the residential energy use is of great importance as they address efficiency and consumer preferences directly.

3.1.6 Conclusions

Overall, the SADEM model provides a reasonable depiction of the specifics of the New Zealand energy sector and is an appropriate tool to project energy-related carbon dioxide emissions over the first commitment period. While energy savings are factored into the model, further work could be done to refine the modelling of energy efficiency improvements and other non-price based climate policies.

Many of the recommendations made on the projection of emissions from the energy sector in the 2005 review have been implemented, providing the analysis with a more solid structure and background. Specifically, improvements to separate the ‘other industrial’ and commercial models within SADEM, and to include demand elasticities into these models, have refined the modelling of these sectors since the 2005 review. The development of a separate sub-model for the dairy industry has also improved projections, as have revisions to existing industrial sub-models. Focus was placed on improving the industrial and commercial sectors since the last net position report, as these sectors have a significantly greater impact on emissions projections than the residential sector. As briefly summarised in the previous section, we still believe modelling of residential sector can be improved further. In particular, use of a more technologically explicit, bottom-up model to address energy efficiency and consumer preferences for different energy services may be appropriate. We note that the issue raised over degree days in the previous review is planned to be addressed in the proposed redevelopment of the residential energy model.

3.2 Transport

3.2.1 Modelling approach

Emissions from the transport sector have grown significantly since 1990, averaging over 3% growth per annum. For the first commitment period of the Kyoto Protocol, emissions from the transport sector are projected to be 80.1 million tonnes carbon dioxide equivalent (41% of total emissions) in the 2007 net position report.

One of the key recommendations from the 2005 review was to adopt a bottom-up modelling approach, which would capture the technology-dependent nature of the transport sector and its energy demand. The 2007 net position report shows that this recommendation has been implemented through the use of the Ministry of Transport’s Vehicle Fleet Emissions Model (VFEM). VFEM models changes to the on-road vehicle fleet, including the number of vehicles, kilometres travelled per vehicle and average fuel economy. In addition to improving the estimation of on-road transport energy demands, the 2007 net position report also states that use of VFEM improved estimates of petrol and diesel use by off-road vehicles and other types of transport.

The 2007 net position report also includes the impacts of new Government policies, including the recently introduced biofuels sales obligation. This policy requires that 3.4% of total diesel and petrol demand will be supplied by biofuels (bio-ethanol and bio-diesel) by 2012. Companies not complying with the Government’s obligation levels will face a financial penalty. Hence, the biofuels sales obligation is expected to reduce carbon dioxide emissions in the transport sector.⁸

It is anticipated that all of the biofuel consumed in the 2008–2012 period will be produced in New Zealand, and most will be derived from the agricultural by-products whey and tallow and from fuel crops such as maize. Second generation biofuels are not anticipated to be used (and therefore do not feature in the model) until the post-2015 period. It is estimated that bioethanol production will have associated production emissions that are 11% of those that would be emitted by the displaced petrol, whilst biodiesel’s greenhouse gas emissions will equate to 54% of those emitted by conventional diesel. We consider that the value of 11% of conventional petrol emissions for bioethanol may be low, even given that bioethanol is expected to be produced from whey where the emissions are already associated with dairy production, or produced from maize with geothermal energy used as a production heat source. The value of 11% quoted in the derivation of the savings is for second generation production of bioethanol produced by acid hydrolysis of wood.⁹ We recommend that the international literature is reviewed to establish a more representative carbon dioxide reduction figure for bioethanol production as it would occur in New Zealand. We do not believe, however, that the change in the estimated saving would have a significant impact on the estimate of the net position over the first commitment period.¹⁰

3.2.2 Reasonableness of approach taken

The VFEM is used to project road transport emissions. The model works in a bottom-up manner and considers a number of technical and economic parameters. The vehicle type, engine size, fuel type and vehicle age come from Land Transport New Zealand Vehicle Registry data. In order to account for the economic drivers of travel demand, population, GDP and/or GDP per capita are used. In particular, as presented in the New Zealand’s Energy Outlook to 2030, the VFEM computes fuel use as follows:

Fuel use = number of vehicles x kilometres per vehicle x average fuel economy

Fuel use demand for road transport is calculated for cars, light commercial vehicles and heavy commercial vehicles based on this formula. All three variables on the right-hand side are projected in the VFEM. The total number of cars and light commercial vehicles is projected by considering the historical growth in vehicle numbers and its relationship with population. For heavy commercial vehicles, a relationship between GDP and vehicle numbers is used. For cars, vehicle ownership is expected to approach a saturation level of 700 cars per 1000 people. The vehicle kilometres travelled data are based on the Ministry of Transport’s estimates of the kilometres travelled by the entire on-road vehicle fleet. The estimates have been determined from traffic count surveys, which are then extrapolated over the entire road network. The future projections are based on the historical relationship of vehicle kilometres travelled with population, GDP per capita, and fuel price. Vehicle fuel economy is input into the VFEM at a detailed level by vehicle type, engine size and fuel type, and also allows for the road conditions in terms of congestion level and urban versus rural running. The VFEM assumes that current ratios of new and used vehicles and their age profile will continue through to 2030.

New Zealand’s energy statistics identify the diesel used by rail and sea, and some of that used by industry in either stationary applications or by vehicles off-road. These amounts, together with the diesel identified as used on-road, are subtracted from total fuel to give an estimate of the remaining off-road diesel use. The future off-road, rail, and non-transport energy demand is then modelled as being driven by population, GDP and fuel price. The off-road projections are then added to the VFEM’s on-road transport projections to yield total land transport energy demand.

While this approach to on- and off-road transport takes into account the characteristics of the vehicle fleet, we have concerns over multiple uses of the same explanatory variables (GDP and population) for different factors in vehicle fuel use for road transport. However, lack of data is an issue and the approach adopted to calculate off-road, rail and non-transport energy demand seems to be a reasonable one to follow.

Air and sea transport energy demands have been projected by applying simple time series methods to the historical demand data. The conversations we have had with the Ministry of Economic Development indicate that previous projections using these methods have proved reasonably accurate and this is expected to remain so through the first commitment period projections. However, further research into modelling aviation demand and reliable econometric drivers is envisaged.

3.2.3 Implementation of previous recommendations

One of the key recommendations from the 2005 review was the integration of the Ministry of Transport’s bottom-up VFEM with SADEM model. The 2007 net position report confirms that this recommendation has been implemented.

3.2.4 Treatment of uncertainty

As listed in the 2007 net position report, the overall performance of the New Zealand economy, the impact of Government policy measures, and fluctuations in fuel commodity prices are the key drivers of uncertainty for transport sector emissions projections.

At a more detailed level, improvements in vehicle technologies (through imports) would also have an effect on New Zealand’s emissions from the transport sector, even though in the short term (between 2008 and 2012) this should not result in a radical change unless there is a significant breakthrough in the efficiency or cost of vehicles.

3.2.5 Proposals for future improvements

The projections of sea and air transport have been simplified to utilize simple time series models to project trends into the future. Even though the use of current trends into the future is problematic in many ways in the long term (mainly due to changes in the characteristics of population and saturation effects), in the short term this may not be an issue. Nonetheless, the growing energy use of air transport may require enhanced modelling to include the drivers of demand.

Also, a split between freight and passenger transport demand would allow a better characterisation of transport demand.

3.2.6 Conclusions

Modelling of transport emission has been improved, as following the recommendations of the 2005, a bottom-up model (the Ministry of Transport’s VFEM) is now used to estimate on-road motor vehicle transport emissions. This model takes account of a number of technical parameters, including vehicle type, engine size, fuel type and vehicle age (from Ministry of Transport statistics) and uses socio-economic drivers (population, GDP and/or GDP per capita) to project future travel demand.

The projections of emissions from sea and air transport are made by applying time series methods to historical demand data. In the future, it is recommended that due to the growing energy use of air transport, modelling which include the drivers of demand is considered. In addition, a split between freight and passenger transport demand would allow a better characterisation of transport demand.

3.3 Industrial processes

3.3.1 Modelling approach

Direct emissions (that is, non-energy related) emissions from industrial processes contribute around 6% of New Zealand’s total greenhouse gas emissions.

The majority of these emissions are from six major industrial processes:

• Production of iron and steel.

• The oxidisation of anodes in aluminium production.

• The production of hydrogen.

• The calcination of limestone of use in cement production.

• The calcination of limestone for lime.

• The production of ammonia and urea.

The industrial processes sector is dominated by emissions from the metal industry. Assumptions on future production (that it remains constant) are based on work done for the Ministry for Economic Development on the energy demand of heavy industry and are consistent with assumptions on energy modelling for heavy industry.

For the industrial processes, the Ministry for Economic Development models carbon dioxide emissions only. Emissions of other greenhouse gases (methane, nitrous oxide and the fluorinated gases)¹¹ are not modelled explicitly, but are allowed for by increasing the modelled emissions of process carbon dioxide emissions from industry by 19%. This is the average ratio within the industrial processes sector of non-carbon dioxide emissions to carbon dioxide emissions over the period 1990 to 2005. Projected carbon dioxide equivalent emissions from the industrial processes are provided in Table 2. The Ministry for the Environment informed us that projections of fluorinated gases are not modelled explicitly, primarily as a result of the source and type of fluorinated gas emissions changing recently and therefore no robust historic dataset being available on which the projections could be based.

3.3.2 Reasonableness of approach taken

The methodologies used for industrial projections are those in use at the time of the 2005 review, at which point they were considered to be appropriate. The scaling of non-carbon dioxide greenhouse gases to carbon dioxide emissions from industry was considered to be an appropriate approach given that the non-carbon dioxide emissions are less than 1% of the inventory. However, the sources of the non-carbon dioxide gases are not the same as for process gases, and if New Zealand wishes to continue to improve the quality of its projections then we suggest that it considers how these emissions might be better modelled on a gas-by-gas basis. In particular, emissions of hydrofluorocarbons have been identified as a key category in the level and trend analysis of the 2005 inventory. It should be possible to model perfluorocarbon emissions on the basis of assumptions in the energy modelling about future aluminium production.

Table 2: Projected CO₂ and CO₂ equivalent emissions from Industrial processes

Year	Industrial processes (kt CO2) (provided by MED)	Industrial processes (kt CO2 – e) (estimated by MfE)
2008	3,684	4,384
2009	3,710	4,415
2010	3,735	4,445
2011	3,759	4,473
2012	3,781	4,499
Total CP1	18,670	22,217

Note: CP1: ‘first commitment period’.

3.3.3 Implementation of previous recommendations

There were no specific recommendations in the 2005 review report on industrial processes.

3.3.4 Proposals for future improvements

• A fuller description of the assumptions, methods and data sources behind the industrial process emission projections should be given.

• An explicit modelling approach should be taken for the non-carbon dioxide industrial process emissions, particularly hydrofluorocarbons, which are a key category in the inventory.

3.3.5 Conclusions

Carbon dioxide equivalent emissions are expected to increase from the industrial process sector between 2008 and 2012. However, there is limited information available to explain why this is the case. Although the predicted increase is small, it is recommended that the Ministry for Economic Development should provide a full description of the assumptions, methods and data sources that have been used to forecast emissions from this sector.

3.4 Waste

3.4.1 Modelling approach

Annex D of the 2007 net position report sets out the approach taken to projecting emissions from landfill sites and domestic and industrial waste water treatment plants. For landfill sites, the methodology used to produce emissions estimates for the historical inventory is also used for projections. This is a tier-2 model consistent with the Intergovernmental Panel on Climate Change (IPCC) recommendations, which uses data specific to New Zealand on waste generation rates, waste composition and rates of landfill gas collection and combustion. Projections of population growth are used to drive the projections of waste generation rates.

The reductions associated with the New Zealand Waste Strategy (in particular, the diversion of biodegradable waste away from landfill), and with the national environmental standard for landfill gas collection, are then subtracted from the projected gross emissions to give a net value.

Projections of wastewater emissions are estimated by using a linear projection of emissions from 1990 to 2003 and extrapolating this trend.

3.4.2 Reasonableness of approach taken

The methodologies used for projections are very similar to those in use at the time of the 2005 review, at which point they were considered to be generally sound.

3.4.3 Implementation of previous recommendations

The 2005 AEA review found that methodologies and input assumptions were generally sound, but had four main recommendations and suggestions:

1. Consideration should be given to whether future wastewater emissions could be modelled directly for future years using the inventory methodology, as while the linear projection has an acceptable r², it does not take account of the potential for the current upwards trend to plateau. A linear extrapolation is still used for wastewater emissions, but there is an ongoing research project to examine the methodology used to estimate wastewater emissions which is due to report soon. This will use population growth as a driver for emissions.
2. The population data set used should be consistent with those used in other sectors. The Ministry for the Environment has confirmed that a single population dataset is now used in all sectors. This same population data set will also be used in the new projections of emissions from wastewater treatment.
3. It was noted that no documentation was available to support the estimate of reductions due to the two waste policies and it was suggested that the ‘most likely’ value assumed might be too optimistic. The draft 2007 net position report records that the ‘most likely’ value of the reduction arising from the Waste Strategy has now been scaled back to 75% of the original values. Further discussion with Ministry for the Environment officials clarified that the original estimate of the savings, which would be achieved through the New Zealand Waste Strategy, had been made by a consultant in 2002, and that the downward revision of the target was made in light of the 2006 review of progress towards Waste Strategy targets. It would improve the transparency of the net position report to include a reference to the original source document containing the estimate of the emissions reduction and to make clear that the downwards revision is due to a review of progress in the Waste Strategy. The reduction in emissions due to increases in landfill gas emissions comes from work by consultants estimating landfill gas recovery rates based on landfill owner intentions. While this is made clear in the draft report, it would again improve transparency if a full reference was included for the report.
4. It was suggested it would be appropriate to build a measure of uncertainty into key input variables in the optimistic/pessimistic scenarios if data is available to allow a sensible estimate to be made. The scenarios in the 2005 report were related to the outcome of the Waste Strategy policy rather than uncertainty in key input variables. This is still the case in the 2007 report: the high and low ranges represent the differences in assumptions about the outcome of this policy and additional causes of uncertainty have not been examined.

The only recommendation that has still to be progressed is the estimate of uncertainty in key variables.

The recommendation on improving wastewater treatment is being progressed, and the recommendation on improving documentation of estimated reductions could be implemented by adding references of the appropriate documents to the net position report.

3.4.4 Treatment of uncertainty

As discussed above, at present the only treatment of uncertainty is in terms of the impact of policy, other uncertainties are not assessed. The treatment of uncertainties is discussed in more detail in section 4.

3.4.5 Conclusions

The methodology used for the waste sector projections is broadly sound and, as recommended, the same population data set has been used across all sectors. Recommended improvements to the wastewater treatment emissions projections are in progress. Reporting of the estimated reductions achieved from policies could be made more transparent and references for reports containing the estimated reductions should be included.

Estimated emission reductions from policies should be kept under review and should be linked to progress towards meeting the targets in the New Zealand Waste Strategy for diversion of biodegradable waste.

3.5 Agriculture

3.5.1 Modelling approach

The projections examined were based on the methodologies used in the National Greenhouse Gas Inventory submitted to the UNFCCC annually, and on econometric and physical models developed by the Ministry of Agriculture and Forestry. The inventory methodology is reported to conform to the Good Practice Guidance methodologies developed by the IPCC and adopted by the UNFCCC.

The emission factors quoted for methane, and annual estimates of nitrogen excretion, are the result of in-country calculations and therefore go beyond default values. The numbers quoted are reasonable when compared with estimates available from other sources; for example, the current and projected United Kingdom estimates for nitrogen excretion by dairy cows for 2010.

The likely increase in the size of the New Zealand dairy herd, in response to increased prices for milk on the world market, has been taken into consideration in the report. The outcome from this assessment was only a small change in emissions due to the increase in dairy cow numbers being counterbalanced by a decline in sheep numbers, but illustrates that new developments likely to impact on national greenhouse gas emissions have been taken into consideration. Moreover, the inherent uncertainty in projections has been explicitly addressed by estimating the 95% confidence intervals of the projections.

3.5.2 Reasonableness of approach taken

Projections were driven by future estimates of:

• Animal numbers by species: dairy cattle, beef cattle, sheep and deer in 2010 using the Ministry of Agriculture and Forestry’s Pastoral Supply Response Model.

• Ruminant methane emissions per animal based on changes in past emissions per animal between 1990 and 2005.

• Nitrogen output per animal based on changes in past nitrogen output per animal between 1990 and 2005.

• Nitrogen fertiliser use based on an econometric model that projects future use from projected animal numbers, fertiliser prices and other variables (output prices, agricultural productivity growth).

Changes in estimation methodologies have been implemented to take into account new approaches and information obtained since the last update in May 2006.

Changes were implemented in two areas:

• The projection of nitrogen fertiliser usage in 2010 was based on an improved methodology developed by the Ministry of Agriculture and Forestry. This resulted in a 6% increase in nitrogen fertiliser use projections.

• Updating of the National Greenhouse Gas Inventory methodology in keeping with UNFCCC guidance for maintaining ‘Good Practice’. Two changes were implemented in the agricultural section of the National Greenhouse Gas Inventory in 2007, in which the 2005 emission levels are reported. These were:

  • The crop residue-burning fraction for 2005 was reduced from 50% to 30% based on updated expert advice. This change had a minor impact on emissions output.
  • Beginning in January 1989, the commencement of the period over which animal emissions are reported was changed from a July-to-June year to a January-to-December year, as three-year averages are used to derive animal populations. This change resulted in an approximate 1% increase in the 1990 emissions levels for agricultural methane and nitrous oxide.

3.5.3 Implementation of previous recommendations

The 2005 AEA review found that methodologies and input assumptions were generally sound, but had four main recommendations and suggestions:

1. We have noted that the assumptions about deforestation and afforestation through conversion to/from grazing land in the post model adjustments to the livestock projections from the Pastoral Supply Response model may not be consistent with the assumptions made in the land use change and forestry projections.
2. The linear extrapolation of enteric implied emission factors and factors for nitrogen excreted per head should be kept under review. It might be useful to examine the trend if single year rather than three-year average data is used. A check on the linear extrapolation method could be provided by estimating the implied emission factors if industry targets for productivity, or the increases in productivity assumed in the Pastoral Supply Response Model are met.
3. In the future it would be useful to understand the drivers for the rapid increase in nitrogenous fertiliser use and judge whether these can be expected to continue and whether other measures to limit fertiliser use may have an impact, before using a linear trend extrapolation to provide an estimate of future use.
4. A more useful optimistic and pessimistic scenario for livestock numbers might be derived by examining the effect of potential changes in the Pastoral Supply Response Model drivers rather than a range for the 2004 livestock numbers.

The implementation of these recommendations was discussed with the Ministry of Agriculture and Forestry which reported that:

1. Information from experts in the forestry sector suggests that land conversion to forestry only marginally impacts on the pastoral sector. Furthermore, the change in the emissions from the pastoral sector is insignificant compared to those arising from deforestation. We have also examined this (see section 2.2) and concur with this view.
2. Non-linear relationships for emission factors have been examined, but no significant improvement in relationship was gained. The effect of using a single year figure in deriving the trend has not been examined, as it is not believed it would have a material effect.
3. It appears that the drivers for the rapid increase in nitrogenous fertiliser use have been incorporated into the Pastoral Supply Response Model, and a revised estimate of nitrogenous fertiliser requirements has been made.
4. Optimistic and pessimistic scenarios for livestock numbers are now modelled with the Pastoral Supply Response Model by making changes to drivers (for example, commodity prices) in the model (see below).

3.5.4 Treatment of uncertainty

The report acknowledges that projections of livestock numbers and performance, and hence future emissions, need to be assessed within the uncertainties of biological systems as affected by climate and changing economic conditions, including changing international commodity prices and the New Zealand dollar exchange rate. The report states that every effort has been made to provide the best projections of future emissions as at June 2007.

As well as the baseline scenario, two further scenarios of projected emissions in 2010 were produced. These represent emission estimates using the 95% prediction intervals for the upper and lower bounds of methane and nitrous oxide emissions and animal numbers. These two scenarios gave an estimate of the values of the upper and lower bounds of future projected emissions at the 95% confidence level.

The report states that to derive livestock forecasts for different scenarios, exogenous price uncertainty was introduced into the Pastoral Supply Response Model through specifying the possible movements in commodity prices for the forecast period. Variation of prices (or the standard deviation of the 95% confidence level) during the last 10-year period for each price series was used. This gave estimates for the upper and lower bounds of the stochastic forecasts that could be considered as lower and higher scenarios due to the movement in prices.

3.5.5 Proposals for future improvements

Projected ruminant methane emissions, are based on changes in past methane output per animal between 1990 and 2005. Individual animal emissions between 1990 and 2005 are based on national data on individual animal performance and diet quality. Emissions per animal 1990–2005 are obtained using the model developed by Clark et al (2003) which takes account of performance and diet. Projected emissions per head are obtaining by fitting linear regressions to the 1990–2005 emissions data and extrapolating to 2010.

No attempts have been made to fit trends to the animal performance data and run the projected performance data through the Clark et al (2003) inventory model because of the large number of relationships that would be need to be estimated if this approach was adopted.

The justification for using a simple approach of projections to 2010 based on modelled output is that the 1990–2005 time series shows such a strong linear trend. We recommend that the trend is reviewed annually and that the approach is reviewed if the trend begins to become less strongly linear.

3.5.6 Conclusions

The agricultural inventory appears well thought out, conforms to UNFCCC good practice and is likely to give accurate results. The approach has been improved in the light of comments made at the last review and gives a robust estimate of greenhouse gas emissions from agriculture for the period to 2010.

3.6 Land use, land use change and forestry (LULUCF)

3.6.1 Modelling approach

Net removals by the land use, land use change and forestry (LULUCF) sector for 2008 to 2012 are projected to be between 16.0 million tonnes carbon dioxide (pessimistic) and 98.3 million tonnes carbon dioxide (optimistic), with the most likely scenario being 58.0 million tonnes carbon dioxide. Estimates of net removals in the optimistic and pessimistic scenarios have changed due to allowances being made for the uncertainty of the Kyoto forest area and changes in future deforestation scenarios.

The projected removals are based on carbon modelling, using three projection scenarios (pessimistic, most likely and optimistic) to incorporate natural variability, measurement uncertainty, gaps in knowledge and alternative future outcomes (due to changes in policies or the economic context, for example). The carbon model simulates the Kyoto forest area, based on a carbon yield table of a typical forest stand and driven by annual planting rates (actual and predicted) between 1990 and 2012. Changes in the soil carbon pool and emissions from deforestation are estimated in separate model components. The net carbon uptake from the model is taken to be the carbon stock change during the first commitment period of the Kyoto Protocol. This approach is based on the methodology currently used to estimate forest carbon stock change for New Zealand’s UNFCCC greenhouse gas inventory. This approach will be replaced by the Land Use and Carbon Analysis System (LUCAS), formerly known as the New Zealand Carbon Accounting System, which is due to be operational from 2010. The LUCAS project will report emissions and removals of greenhouse gases from forests planted since 1990 and land deforested during the first commitment period.

The key assumptions used in the projections (that is, those factors used to define the pessimistic, base and optimistic scenarios) are clearly described and discussed in the appendix to the net position report. They are: future afforestation rates, area of existing Kyoto forest, Kyoto forest growth rates, changes in soil carbon with afforestation, area of ineligible Kyoto planting and future deforestation rates of plantations. Some of these factors are correlated, so the three sets of scenario assumptions were run in separate simulations to produce the combined model results.

Good practice, as described in the IPCC Good Practice Guidance for LULUCF, has been followed throughout, with the use of higher tier approaches commensurate with the contribution of these categories. Consistency with the Good Practice Guidance is embedded in the LUCAS project design (using Tier 3 approaches) and so the outputs from LUCAS will also follow good practice.

3.6.2 Reasonableness of approach taken

The approach taken, in terms of the methodologies and input assumptions, is reasonable and clearly described. There are weaknesses in the data currently available for carbon stock accounting but these will be addressed by the LUCAS when it becomes operational in 2010. The production of robust estimates for the volume of carbon in New Zealand’s carbon pools for reporting under Article 3.3 of the Kyoto Protocol, and hence the ability to claim removal units, is reliant on the full implementation of the LUCAS during the first commitment period. The risks associated with project slippage have been identified in the project charter; however, progress is currently on track for delivery in mid-2010.

3.6.3 Implementation of previous recommendations

In the review of the 2005 net position report there were four issues (summarised in Table 3) where further consideration was recommended in order to minimise uncertainty in future projections.

The economic and policy environment of New Zealand forestry in recent years (described in the 2007 net position report) have induced substantial changes in the sector in the past five years. The current regime of afforestation and deforestation rates is so different from the historical trend (over the past 30 years) that there is considerable uncertainty as to their trend in the future. The first recommendation was to improve knowledge of the reasons and drivers for the downward trend in new forest planting to allow improved forecasting concerning the future possible intentions of forest owners. This has been achieved by a report on the financial returns from forestry (Horgan 2007) and deforestation intentions surveys undertaken in 2005 and 2006 for the Ministry of Agriculture and Forestry. In the 2007 net position report, three afforestation scenarios were presented (planting rates of zero (0), 5000 and 20,000 (five-year average) ha/year from 2007 for the pessimistic, most likely and optimistic scenarios respectively). Based on current afforestation rates these scenarios are expected to cover the range of afforestation levels expected under current market and policy conditions during the first commitment period. In terms of impact on New Zealand’s net position, afforestation is of lesser importance than rates of deforestation. Any afforestation from 2006 onwards will have little impact on net removals during the first commitment period because newly established forests remove little carbon dioxide from the atmosphere during the early years of growth. For the same reason, the potential impact of any new policies encouraging afforestation will mostly be felt beyond the end of the first commitment period in 2012. The extent of existing (1990–2006) Kyoto forests will be verified by the land use mapping workstream in the LUCAS project.

The second recommendation was to improve quantification of the areas of planting of post-1990 forest at national scale into existing shrublands that meet the Kyoto Protocol definition of forest. The confusion in this area has arisen because New Zealand shrublands have never been considered ‘forest’ but some areas could be defined as such under New Zealand’s single minimum values for accounting under Article 3.3 of the Kyoto Protocol: a tree crown cover of 30%, a minimum land area of one hectare, and a tree height of five metres. The projection scenarios in the 2007 net position report use estimates of 8%, 16% and 21% of post-1990 forest planting as ineligible under the Kyoto protocol due to it being planted into existing shrublands that come under the forest definition.

Further quantifiable information is required to assess the proportion of ineligible planting and therefore to reduce the range of uncertainty. Land use mapping for the LUCAS project will produce this information. There is also ongoing regional research (by LandCare Research and AgResearch) in support of the LUCAS mapping, including farm interviews, which may give a more accurate picture of the extent of post-1990 planting in shrubland that meets the Kyoto protocol definition of forest. Any regional variation in the proportion of ineligible planting will need to be taken into account when making the national assessment.

A secondary recommendation related to this was to improve the information on the burning and decay of scrub vegetation. Estimates have been made in the projections and have a small impact. The area of scrub cleared depends on the area of ineligible planting – so there are more emissions from scrub clearance in the optimistic scenario because more of the shrubland area remains eligible for Kyoto forest planting.

The third recommendation was to improve estimation of the area deforested and drivers for this process. Deforestation rates look likely to have a significant effect on New Zealand’s net position in the first commitment period and there have been considerable efforts in this area. Surveys of forest owners’ actual intentions to deforest were made in 2005 and 2006 for the Ministry of Agriculture and Forestry. ENSIS modelled five deforestation scenarios, including three based on the results from the forest owners’ survey, from which the Ministry of Agriculture and Forestry selected the base, pessimistic and optimistic scenarios for the net position report. There was also discussion in the net position report of drivers that might increase deforestation beyond the pessimistic scenario. Decisions on deforestation policy options are anticipated by the end of 2007. These should enable additional refinement of the projection scenarios in the future.

There remains uncertainty about the level of clearance of indigenous forest or shrubland that meets the Kyoto forest definition, although it is thought to be insignificant and not to result in land use change. Deforestation of this type is not currently considered in the projections. The assessment of the level of this sort of deforestation will be possible once the LUCAS is operational.

The final issue was the need to undertake further research on the time pattern of loss of carbon from soils after afforestation. Currently the changes in soil carbon stocks following afforestation are modelled as occurring gradually at a linear rate of change over 20 years, rather than instantaneously. The soil carbon work stream in LUCAS will contribute to the investigation of this issue, through the harmonisation of existing soil data, the identification of knowledge gaps and additional sampling as required. Given that much planting of Kyoto forests has taken place on hill slopes prone to erosion and with high natural variability, this is a complex area of investigation that can only be fully addressed by an ongoing research/monitoring programme. Such a programme should be achieved when the LUCAS soil methodologies transition to business as usual when the system becomes fully operational.

An additional recommendation was made that a single document should be produced for any future projection estimates that would provide detailed descriptions and sources for all calculations. The LULUCF appendix to the net position report largely fulfils this recommendation. The basis of the uncertainty ranges is clearly defined and the direction of possible changes in projections in the future is described in many cases. There has also been progress in refining the projections in areas that were not covered in the original review recommendations.

Table 3: Summary of the status of previous recommendations

Recommendations considered as being of high priority	Implementation
LULUCF: Improve knowledge of the reasons and drivers for the downward trend in new forest planting to allow improved forecasting concerning the future possible intentions of forest owners	Implemented. The net position report uses three afforestation scenarios that under current market conditions and policy settings cover expected afforestation levels out to 2012. Information on afforestation drivers has been gathered but this is considered to be more suited for longer term projections and evaluating policy options.
LULUCF: Improve quantification of the areas of planting of post-1990 forest at national scale into existing shrublands that meets the Kyoto Protocol definition of forest	Awaiting data from the LUCAS project (due to become operational in 2010).
LULUCF: Improve estimation of areas deforested and drivers for this process	Implemented. The LUCAS project will provide additional quantification.
LULUCF: Undertake further research on the time pattern of loss of carbon from soils after afforestation	Awaiting data from the LUCAS project (due to become operational in 2010).
Process: Improve written documentation to fully describe methodology and assumptions made across all sectors	Implemented.
Uncertainties: Improve the transparency of the uncertainty calculations by clearly defining the basis for calculating the uncertainty ranges and using a consistent approach to quantifying uncertainties across sectors in terms of deciding what is trying to be quantified, that is, the uncertainty caused by future unpredictability or value uncertainty	Implemented. Data from the LUCAS model will improve the quantification of uncertainties.
Recommendations considered as being of lower priority
LULUCF: Improve the information/quantification on the burning and decay of shrub vegetation	Partially implemented. Awaiting additional data.

3.6.4 Treatment of uncertainty

The LULUCF sector is the greatest source of uncertainty for New Zealand’s net position. Three scenarios are used to qualitatively assess uncertainty of the different factors used to estimate emissions and removals.

These scenarios have to incorporate considerable uncertainty in future outcomes, measurement uncertainty/knowledge gaps and high natural variability in the factors themselves (a major source of uncertainty). This is further discussed in section 4.6.

The main uncertainties in terms of reduction of net removals are future rates of deforestation and the proportion of ineligible Kyoto afforestation. The main source of uncertainty in terms of increasing net removals is the Kyoto forest growth rate.

Data being collected for the LUCAS project will help to reduce the uncertainties associated with measurements and information gaps, specifically Kyoto forest growth rate, soil carbon change with afforestation and the area of ineligible Kyoto afforestation. The LUCAS data will also validate existing quantitative measures of uncertainty, such as the area of Kyoto forests that have already been planted (estimated 5% uncertainty). However, there are still large uncertainties over the future implementation of policy and changing economic conditions, particularly those concerning the deforestation cap and what the Government will do if the cap is exceeded. The uncertainty over future activities cannot be significantly reduced below the existing estimates at the present time.

3.6.5 Proposals for future improvements

The implementation of the LUCAS will meet the most pressing data requirements, which are the most obvious areas for improvement. Although LUCAS is due to be fully operational in 2010, some data work streams will be completed before that time, and it is expected that this data will be used to reduce uncertainties when it becomes available. The scope for future improvements will be clearer once the LUCAS is fully operational and sampling/monitoring has transitioned to business as usual.

It is recommended that the LULUCF documentation (in the appendix to the net position report) is kept up-to-date with the latest methodological developments. It would also be useful if more information could be included about the LUCAS project, such as the project objectives, a brief description of the main data work streams and progress towards full implementation.

In their 2007 projections report, Ensis gave options for improving removals estimates with the current methodology before LUCAS becomes fully operational. Some of these options may become appropriate if LUCAS is not fully operational by 2010. The choice of option would depend on the areas for which data was incomplete and would be influenced by the short time scale for implementation before the end of the first commitment period. The options are:

• Data from LUCAS should be used to reduce uncertainties in the projections as soon as they become available.

• More information about the LUCAS project should be included in the LULUCF section in the net position report.

• Documentation should be kept up-to-date with the latest methodological developments.

3.6.6 Conclusions

The approach taken to estimating the LULUCF sector’s contribution to New Zealand’s net position is reasonable, given the current data limitations, and is clearly described. Many of the current data issues will be addressed by the implementation of the LUCAS, which should become operational in 2010. There has been significant progress on the implementation of the recommendations of the 2005 net position report review. The amount of effort put into different factors has sensibly been linked to their anticipated impact on New Zealand’s net position, with estimates of the deforestation area receiving most attention. Where implementation of the recommendations is only partially complete, this is due to the requirement for data from the LUCAS project.

---

---

Previous|Next