Skip to main content.

5 Development of Baseline 1990 Estimates for Forest and Scrub

5.1 Methodology

Indigenous forests are a major carbon reservoir in New Zealand. The approach adopted for this study is to determine a suitable classification system for forest and scrub along with the necessary biomass relationships, and then apply the biomass figures to the total extent of each forest type, i.e. allometric equations for biomass, for the country.

To obtain the extent of total forest by forest and scrub type, a number of available databases were examined and allometric equations for biomass refined from biomass studies.

Fine-scale maps of indigenous forest in New Zealand before this study are scarce and often confined to small areas. The most comprehensive map set of both the distribution and composition of primary indigenous forests throughout New Zealand is the 1:250 000 forest class (FSMS6) that recognises 18 forest classes. These maps were produced before 1990 and needed substantial updating and additional coverage and were therefore not used to produce a 1990 baseline.

A previous study that produced the Vegetation Cover Maps (VCM) has national coverage for all woody vegetation but has been compiled at a scale of 1:1 000 000. Therefore, although it manages to resolve major plant communities, it only delineates map units greater than 500 ha.

Another key spatial database was the National Vegetation Survey (NVS) that contains data from plot-based studies collected on vegetation from 7083 plots around the country.

Hence to determine a 1990 baseline for indigenous forest and scrub, data were used from the NVS, relationships were made from a few biomass studies, and area data from the VCM used to generate a forest estimate for total biomass carbon stored in live trees.  We selected 574 plots that were as geographically dispersed as possible and had minimum bias in terms of regional sampling intensity, elevation, and mean annual precipitation.

5.2 Results

We estimated 1990 forest biomass carbon as 919, 939, 943, and 932 Mt, based on pooling the data from all 574 plots and stratifying these sample plots by VCM classes, elevation classes, and mapped soil-climate strata, respectively.  At most these estimates differ by 2.6% and have a mean of 933 Mt.  Of the national forest biomass carbon reservoir, 60.0% is stored in beech trees, 26.7% in other hardwoods, 13.2% in conifers, and 0.1% in other (e.g., tree ferns) taxa. The standard error of the national forest biomass carbon estimate is  25.1 Mt. The estimate of biomass carbon in scrub and other woody mixtures of indigenous vegetation were calculated as 527 Mt. (Goulding et al. 1998). Tate et al. (1997a) estimated total standing stock of indigenous forest litter carbon as 570 Mt.  All estimates are very sensitive to the actual size of the mapped area and heterogeneity within VCM classes.

Monitoring Carbon in Indigenous Forests and Scrub

6.1 System requirements

Indigenous forests cover approximately 6.4 million ha of land in New Zealand.  Contrary to the trend in recent history when bush was cleared and burned for farming, some land now recognised as economically unsustainable for farming, due to poor soils, steep slopes, or low commodity prices, is being left to revert to its original land cover. It progresses through a succession of bush eventually to mature forests.  Indigenous forests, with their large biomass, form a reservoir for carbon, second only to soil.

There are, however, several forces acting on the extent of the forests and the amount of biomass and therefore carbon they contain.  In part, as well as the natural reversion of pasture to scrub and, eventually, mature forests, our indigenous forests are under attack from pests such as possums, wild goats, and deer, and are subject to catastrophic events and limited harvesting.  In short, the indigenous flora of New Zealand is in a state of flux, even without active forest clearance or large-scale planting.  It is still debatable whether all these effects are ultimately classified as anthropogenic and therefore included in the Kyoto Protocol. However, the individual processes that contribute to forest biomass or carbon quantity changes need to be monitored so that we can understand and protect them, as required by the Framework Convention for Climate Change.

In considering the carbon balance of indigenous forests it is necessary to examine the entire ecosystem including the soil and the coarse woody debris that has fallen to the forest floor to form litter.  Fine litter is dealt with as part of the national soil monitoring system, but coarse litter with a diameter greater than 10 cm has been included as part of the forest system.

As well as biomass or carbon accumulation, it is also sensible in any monitoring system to consider other values inherent in those forests.  The members of the Steering Committee for this project add policy and management perspectives relating to biodiversity, forest condition, pest management, ecosystem processes and risk assessment.  Considerable research within the central-government-funded PGSF has, and is, being carried out.  This links directly or indirectly to this project, and science providers are strongly encouraged to ensure their research programmes contribute to the operational needs of the proposed monitoring system.

The baseline level of carbon stocks in indigenous forests is as reported for 1990, and a monitoring system needs to estimate the change in that figure and pull apart the various contributions to that change.  The system must be quantitative and give estimates of uncertainty, the goal being to achieve the 95% confidence level.  Any change in that level of uncertainty would influence the amount of direct sampling required and therefore alter the cost of the system.  The system needs to be appropriate to the New Zealand situation and will not necessarily mirror systems used in other countries with different land uses and management practices.  In particular it needs to incorporate regenerating scrub.

6.2 Consideration of alternative approaches for the monitoring system (Landcare Research New Zealand Ltd & Forest Research Institute Ltd 1998)

There are four distinct approaches to the monitoring system and various permutations of their combinations.  They vary in the technology available, their cost, and their ability to incorporate the non-carbon attributes required of the system.

The simplest and probably cheapest system is recommended by the IPCC for a tropical, undeveloped third-world country where carbon density is estimated for shrubs and for closed forest.  Changes in carbon density are determined to be directly proportional to land areas in either of those two categories.  Such a system would require an updated Land Cover Database that monitors the extent of land in different land uses over time.  

A simple plot-based sampling system could be used to give information specific only to the international reporting requirements without assessing other information on changes within the forests, or attributes relating to forest health and biodiversity. It would involve forest extent estimated through an LCDB with limited plot sampling of about 400–500 plots randomly distributed across New Zealand. For the requirements of the current contract this would be the preferred option. However, it would have to be modified to include the wider requirements of other stakeholders.

Remote-sensing techniques show promise and can clearly be used to estimate forest extent, but further precision in determining forest classes and estimating changes within indigenous forests is still under development. This option will become more available as technology develops, and may in time be able to substitute for some plot-based monitoring. The degree to which this occurs will eventually depend on what non-carbon values are incorporated into the CMS.  (Ranson et al. 1997; McNeil et al. 1998; Stephens et al. 2001)

The most comprehensive approach is to follow a model used in the Northern Hemisphere that would enable changes in carbon levels to be estimated along with information for key stakeholders interested in other forest attributes. The method uses a LCDB to measure forest extent, but then uses a comprehensive plot-based system to consider forest and scrub attributes. The plots would be located by a systematic random sampling regime across the country but would be precise enough to give information at a range of spatial scales and geographic, ecological, and regulatory boundaries. The information products of this system would anticipate multi-stakeholder requirements into the medium future and be on a par with other first-world countries.

Each of the above systems could be supplemented with simulation models that provide future predictive capability of productivity and biomass accumulation under different scenarios. These models are being developed within PGSF research programmes.

6.3 Designing the monitoring system

For the 1990 baseline carbon estimates for indigenous forest, 574 plots (20  20 m) from throughout the country were selected that were as geographically dispersed as possible and with minimal geographical and climatic bias. It has been assessed that if 95% confidence is to be achieved, any stratification of the forest should not reduce the number of sample plots. This number may need to be increased if other features or management outcomes, such as pest control, were to become part of the purpose for the monitoring system.

Testing for forest vegetation was carried out within a grid of 9  9 km, superimposed over the LCDB vegetation map within a South Island that had good historical data, during the summer of 1998/1999. This created 62 sampling points, 39 in forest and 23 in scrub. It gave the same 95% confidence level as had been proposed for the overall system.  The sampling intensity was based on procedures used extensively for NVS sampling and used to calculate the 1990 carbon baseline, providing data to answer the following stakeholders needs:

  • woody biomass as an indicator of carbon

  • tree death by species as an indicator of forest health

  • regeneration by species as an indicator of forest maintenance

  • invasions of exotic plant species as an indicator of intactness

  • browsing by introduced animals as an indicator of animal impacts

  • live and dead stem biomass as a habitat indicator

  • rare and threatened plants as an indicator of diversity maintenance.

To obtain a figure for biomass per hectare, the collection of biomass data was the prime consideration. A network of permanent plots (20  20 m) was proposed as the basis for the monitoring system and, where possible, existing NVS plots were to be used.

In addition to sampling methodologies for forests, sampling methods for scrub, which until now had not been well defined, and sampling procedures for coarse woody debris on the forest or scrub floor, were also required. Understorey (i.e. seedling samples) was identified and measured within each subplot.

6.4 Calculating carbon stocks

The work in the transect provided an opportunity to test the feasibility and accuracy of methods being developed to determine carbon stocks. Previous biomass studies from destructive sampling contributed to relationships between forest measurements, biomass, and carbon content. Refining the regression equations derived from those studies was a recommendation of the study, particularly for those species that composed over 95% of the forest carbon stock (Allen et al. 1998).

A volume-estimate approach to determining scrub was used that involved destructive sampling of two bushes from the side of each scrub plot.

Sampling of the transect estimated that 15% of the carbon in forests is from dead wood, but only 1% of the carbon from scrub is dead wood.

The study confirmed that around 500 forest plots would be required to estimate forest carbon at 95% accuracy. The same number for scrub plots would give ~90% accuracy. The mean carbon stocks per hectare were 203 t/ha for forest and 22 t/ha for scrub. Multiplying these figures by the estimated areas of forest and scrub gave carbon stocks of 67.3 and 2.9 million t respectively.

6.5 Estimating the spatial extent of indigenous forest and scrub

In any monitoring system, the extent of indigenous forest will need to be estimated through the updating of land-use databases that either focus on forests or include forests.  The two existing databases considered the best candidates for this task were the Vegetative Cover Map (VCM) and the Land Cover Database (LCDB).  Plot sampling results in the transect were used to determine which of those two databases should be recommended for the monitoring system.

The study showed that the mapping scale of the VCM, at 1:1 000 000, gave rise to mixed land-cover classes that could not provide sufficient precision of location and extent for the monitoring system. In addition, the VCM is now over 15 years old and would need to be updated to represent recent land-use cover.  The LCDB showed a high degree of accuracy in separating forest and scrub from other classes, but did not distinguish well between those two classes. This deficiency was not considered a problem for the monitoring system where both classes can be treated as one population. The recommendation from this study was that the LCDB be the primary mapping tool to determine the spatial extent of forest and scrub, with the VCM and a third classification system, known as the FSM6 series, being used to provide more detailed indigenous forest species information as required (Newsome et al. 1998).

6.6 Physiological modeling

There is scope to integrate physiological modeling into the system potentially to reduce the amount of sampling required, or to help target it better so that resource managers have a predictive tool to consider "what if" scenarios of climate variation or alternative management practices on forest condition and extent.

Work on these models is mainly being undertaken as part of PGSF programmes. When linked to remotely sensed information related to canopy condition and forest age, the programmes provide information on carbon uptake rates and biomass accumulation. This work will continue and will link strongly with plot-based work carried out in the development and implementation of the monitoring system so that over time it becomes an integral part of the system.

Data handling and management (Gibb et al. 1998)

An essential part of any ongoing monitoring system is the analysis and storage of data.  The requirements for the forest and scrub component of the overall system were:

  • a computer system capable of inputting field measurement data, analysing it to calculate the quantity of carbon per hectare, and storing the plot data for future measurements

  • a report detailing the data model, the potential to use existing systems, and the interactions with carbon-specific software.

The data file model for the South Island transect recorded 12 discrete types of data.  Most were collected for biomass carbon estimation but some addressed other end-user needs.  The data stored are:

  • tree diameter measurements by species (Diam)

  • seedling and sapling counts in subplots by species (Ustorey)

  • site information and stand characteristics (Site)

  • species composition and cover estimates (Recce)

  • upper 10 cm of mineral soil chemistry (Soil)

  • coarse woody debris in decay classes and, where possible, by species (CWD)

  • tree diameters on 20-m-radius plot by species (>60)

  • total height of representative trees (Height)

  • estimate of woody cover of scrub at 25 point (Cover)

  • width and height of individual scrub plants by species (Width)

  • mass data of individual scrub plants (Width biomass)

  • mass data from an area of scrub (Cover biomass)

To calculate the level of carbon stored in biomass on a forest plot, three biomass carbon values were calculated: coarse woody debris; trees on the 20  20 plot; and trees less than 60 cm on a 20-m-radius plot.  The data files required were tree diameters, height of representative trees, and species.

In considering a system for forest carbon data, existing databases were reviewed.  The closest existing systems were the NVS database and the production forestry permanent sample plot (PSP) system.  The first system, NVS, is primarily intended to determine spatial and temporal patterns in indigenous forest composition and structure for the surrounding environment. It can be used for carbon estimates but also for a variety of other data, required for forest health and biodiversity management. PSP is mainly intended to determine the effects of alternative silviculture methods on production forest growth.  

The data-management system for forests and scrub is only one part of the overall information system.  Also required, either as a separate component or integrated with the forest information, is information on soils, and a delivery system that transfers the required information to the client.

 

[ |