Annex B: Supplementary information on Projections Modelling methodology
Agriculture
Projections of the animal numbers
Projections of the livestock numbers for dairy cattle, beef cattle, sheep and deer are undertaken with an econometric model, the Pastoral Supply Response Model (PSRM). The PSRM is an annual time-series model that is representative of the biological constraints and investment decisions made by New Zealand farmers. The projections are based predominantly on the final June 2004 results of Statistics New Zealand's Agricultural Production Survey. Product prices are those used in Ministry of Agriculture and Forestry projections prepared for the Treasury's Pre-Election Economic and Fiscal Update.
Post-model adjustments are carried out based on known and estimated factors that may reduce the land area available for livestock as follows:
- The removal of South Island high country leasehold areas from livestock farming from June years 2002 to 2013 has an estimated cumulative loss of 0.5 million stock units (SU).
- The annual area of grazing land converted to forestry rises from the current low of 10,600 hectares in the year to June 2004 to 18,700 hectares in 2010, and to 20,000 hectares in 2020. This compares with the Ministry of Agriculture and Forestry's separately provided scenarios of 10,000 hectares, 20,000 hectares and 30,000 hectares for June 2010. It is assumed that these areas displace sheep.
- Anticipated deforestation of an estimated total of 34,000 ha spread over June years 2005 to 2010 in the central North Island. The land will go mostly into dairy.
- Sheep numbers are adjusted down in 2009 so that total SU of livestock plus the cumulative opportunity loss of SU displaced by forestry are equal to the level as at June 2008. The maximum SU is 97.711 million with 94.153 million from dairy, beef, sheep, deer and goats, and the balance is the cumulative SU displaced by forestry. This ensures that feed demand approximates feed supply over the longer term.
Livestock numbers for June 2020 are extrapolations from projections to 2014. These assume constant levels for dairy, beef and deer, while sheep numbers decrease in line with increasing new plantings of plantation forestry.
Projections of enteric methane emissions per animal
Projections of methane emissions per animal in 2020 are derived from linear trends of the methane emissions per animal 1990-2003, extended out to 2020. The per animal emissions used to derive the linear trends are sourced from the current national methane inventory and are calculated using a model (Clark et al., 2003) (4NC Figure B1) .
4NC Figure B1: Outline of the process used to calculate methane emissions in the national inventory
See the figure at its full size (including text description).
The model determines monthly feed intakes for different age classes of each animal species based on the mean national animal performance data derived from national statistics. For example, in dairy cattle, inputs include: animal liveweight, milk production per animal, milk fat percent per animal and milk protein percent per animal. For each animal species, an empirical relationship has been derived for the amount of enteric methane produced per unit of feed intake. These relationships have been developed in New Zealand for deer, beef and dairy cattle, and sheep using the SF6 (sulphur hexafluoride) tracer technique to assess methane emissions from animals consuming forage based diets. From these estimates of feed intake per animal, and methane produced per unit of intake, an implied annual emission factor has been calculated per animal that takes into account the changes in animal performance over time.
The implied methane emission factors for dairy cattle, beef sheep and deer in 1990, 2010 and 2020 along with the correlation coefficient (r) for the linear trend in per animal emissions 1990-2003 are presented in 4NC Table 21, Chapter 5. These data indicate a strong linear trend in the increase in methane emission per animal over the period 1990 to 2003.
The models discussed above developed by Clark et al. for methane emissions provide all the information needed to estimate nitrogen output per animal.
Projection of nitrogen fertiliser use
Nitrogen fertiliser use has increased nearly six-fold from 1990 to 2003. Two methods were used to assess projections of nitrogen fertiliser to 2010. Further assumptions were made for the change in nitrogen fertiliser use between 2010 and 2020. The first method used projections of nitrogen fertiliser use derived from a linear trend of fertiliser use from 1990 to 2003. The correlation (r) was 0.96. The projected value for 2010 was 433,700 tonnes of nitrogen.
The second method used best fertiliser industry estimates provided through the Fertiliser Manufacturers Research Association which takes into account future exchange rates, agricultural commodity prices, shipping costs and general projected economic circumstances for agriculture. The projected best estimate value for 2010 was 408,500 tonnes.
The mean value between these two estimates of 421,100 tonnes was used for estimating nitrous oxide emissions in 2010.
The approach to assessment of usage from 2010 to 2020 used different assumptions. These involved projections on world growth in consumption using extensions of the projections of future consumption by the International Fertiliser Industry Association of 1.5 percent per annum.
Energy
The greenhouse gas emissions from the energy, transport, and industrial processes sectors are obtained from the Ministry of Economic Development's Supply and Demand Equilibrium Model (SADEM). The Ministry has used SADEM since the early 1990s for both internal policy analysis and to prepare published projections of New Zealand's energy supply, demand, and prices.
SADEM is really a collection of models. There are supply models for electricity and various fuels, and demand models by sector or industry. Most of these models are quite simple. The notable exception is the model that deals with electric power and renewables, which is reasonably complex.
Figure B2 summarises the design of SADEM. Each box represents a separate model. The arrows indicate the connections; note that some arrows flow "under" the Electric Power and Renewables box. If an arrow points upward, it means that prices generated by the lower model feed into the upper model. If the arrow points downward, it means that quantities of energy demanded feed from the upper model back to the lower model. If the arrows point in both directions, it means that prices are flowing up and quantities are flowing down. These bidirectional arrows indicate "equilibrium relationships", where a solution requires the quantity supplied to equal the quantity demanded.
The arrows coming out of oil product supply and coal supply point upwards only. This is because the model assumes oil and coal supply to be perfectly elastic. That is, their prices are fixed, so prices flow upward only. New Zealand is assumed to be a price taker in these markets.
Heavy industries on the other hand - forestry, basic metals (steel and aluminium), and petrochemicals (methanol, urea, and refineries - are assumed to have inelastic demand. That is, their demand is fixed by assumption and does not respond to price. The Ministry currently lacks an understanding of how these industries would respond to changes in energy prices, but has recently initiated a study to address the issue.
SADEM thus equilibrates the four models in the right half of this diagram: gas supply, electric power and renewables, residential demand, and other industrial and commercial demand. The model finds the equilibrium prices and quantities through an iterative process of trying various values until it finds the right one.
The model first finds the year of gas exhaustion, which determines the gas price path, as well as the electric power prices, so as to equilibrate the supply and demand for gas and electric power, given an assumed fleet of generators. Once it has found a set of electric and gas prices that equilibrate supply and demand, it then asks if it is profitable to add more generation. If so, it adds some generation of the most profitable type and recalculates the gas and electric power prices. It continues to add generation until it has either used all the available sites (hydroelectricity, geothermal, and wind potential are limited by the available sites) or until no more generation can be profitably added, which is when the algorithm stops.
The electric power and renewables sector
SADEM models the electric power and renewables sector by simulating generation dispatch for each of the seasons of each year many times - currently it cycles through each year 550 times. For each season, the model draws a water inflow from a random distribution, developed in consultation with the New Zealand's National Institute of Water and Atmospheric Research. At the end of the season, the ending reservoir levels become the starting reservoir levels for the next season. Individual hydroelectricity projects are not represented; rather the model assumes New Zealand has one giant hydroelectricity reservoir.
Since demand varies hour by hour, the hours of the season are grouped into six categories, ranging from highest to lowest, to form a "load duration curve". The model seeks to find the lowest cost way to dispatch generation for each of the six load duration curve segments. For each load duration curve segment, generation is dispatched in order of variable costs until the remaining demand can be served with the water available for storable hydro. 4NC Figure B3 illustrates this process. Given the average dispatch pattern over many seasons, it is possible to project marginal generation costs and prices.
Gas price projection
In the gas supply model, gas prices are based on the year domestic supplies are assumed to be exhausted. To determine this year, SADEM sums up the cumulative amount of gas assumed to be available from each major field each year, plus new discoveries. It does the same for gas demand, calculating a total demand each year and a cumulative total demand going forward. The year in which cumulative demand exceeds cumulative supply is the year of exhaustion.
As shown in Figure B4, the model assumes that after the year of exhaustion, the price of gas is set by the price of imported liquefied natural gas (LNG). The price of LNG is a simple function of the assumed price of oil, since this appears to be how LNG prices are usually set in the real world. Prior to the year of exhaustion the price moves along a smooth curve from whatever price we assume in the year 2007 to the price of LNG in the year of exhaustion.
Demand
4NC Table B1 summarises the demand models used in SADEM. Three major energy demand sectors are modelled; residential, industrial and commercial, and transport. Each sector includes several sub-sector models. Approximately two-thirds of the total energy demand is modelled using a multi-variate approach that includes price response. About a fifth of the total energy is modelled based on forecasts directly from the industries concerned. The remaining portion is modelled using ordinary least squares (OLS) regressions that effectively produce extrapolations of past trends.
4NC Table B1: Demand sectors and modelling techniques
| Major sector | Sub-sector | Model | Net energy (PJ, 2004) | Percentage |
|---|---|---|---|---|
Residential |
Residential |
Multivariate (GDP, Price, HDD, Lagged Demand) |
52 |
10% |
Industrial & Commercial |
Forestry |
MAF forecasts |
23 |
4% |
Metals |
Company forecasts |
40 |
8% |
|
Petrochemicals |
Company forecasts |
52 |
10% |
|
Other Industrial and Commercial |
Multivariate (GDP, Price, HDD, Lagged Demand) |
105 |
20% |
|
Transport |
Petrol (Land) |
Multivariate (GDP, Price, Lagged Demand) |
104 |
20% |
Diesel (Land) |
Multivariate (GDP, Price, Lagged Demand) |
80 |
15% |
|
Aviation |
OLS |
48 |
9% |
|
Sea |
OLS |
22 |
4% |
|
Other |
OLS |
9 |
2% |
|
TOTAL |
533 |
100% |
Greenhouse gas emissions
The one last step after modelling energy supply and demand is to model the emissions of greenhouse gases. Modelling carbon dioxide emissions is fairly simple, since the output of carbon dioxide is a simple function of the amount of each type of fuel burned. Non-carbon dioxide greenhouse gases are more complicated to estimate than carbon dioxide, since the emission factors depend not only on the specific fuel types, but also on the type of use to which the fuel is put.
