Skip to main content.

6 Recommendations on Methodology for a National Perspective

Introduction

This chapter describes options for refining the work reported on in this review to obtain better national (or in some cases regional) estimates of benefits and costs of afforestation and reversion measures.

Flood reductions and water yield

We do not recommend that MfE proceed with an attempt to quantify the nationwide value of carbon-sequestration-related flood reductions, or other changes in water yield.

To summarise our reasons:

  • Diffuse reversion / afforestation by individual owners, on parts of their own properties, will not appreciably alter even small flood waves (<1-year frequency) passing down the main channel of a catchment.

  • Widespread reversion / afforestation by owners changing land use on entire properties, will alter the magnitude and duration of small and moderate flood waves (1-to 10-year frequency). Consequential reductions in overbank flooding and associated damage are possible on unprotected rivers; but on protected rivers there will be little or no reduction.

  • Whole catchment reversion / afforestation by an agency intervening to change land use on all properties, will substantially alter the magnitude and duration of small to moderate flood waves; but this will only slightly alter large flood waves (> 10-year frequency). This scenario will not reduce large flood waves enough to avoid overbank flooding and associated damage, either on unprotected or protected rivers.

Should MfE wish to verify that flood reduction effects are indeed small in medium to large catchments, two hydrological models are currently available (summarised below). We suggest that it is unnecessary to undertake hydrological modelling for every catchment in the country. Such an exercise, apart from being expensive, is likely to replicate the same hydrological response many times. More information could be obtained from a single catchment, with a diffuse pattern of land use and ownership. Successive model runs, stepping up the percentage of afforested / retired land from 10 to 90% in 10% increments, would demonstrate what is likely to happen.

Watyield

The water balance model ‘Watyield’ was developed by Landcare Research, funded by MFE’s Sustainable Management Fund, initially for use by the Tasman District Council. It is intended to predict the effects that a land cover change may have on the water balance in a large catchment, where there are limited data on climate, soil and vegetation.

Input data required are:

  • daily rainfall

  • monthly evapotranspiration

  • estimates of% interception for each cover type

  • estimates of soil water parameters (for grouped soil types)

  • two base flow parameters.

Outputs are:

  • soil water storage

  • daily, monthly or annual water yields

  • minimum annual seven-day low flows.

It has been initially calibrated on three catchments ranging from 3 km2 through 23 km2 to 369 km2, with good results reported (Fahey et al, 2004). Instances of its subsequent use have not come to our attention, but may exist.

Topnet

The ‘Topnet’ model was designed by NIWA for continuous simulation of catchment water balance and river flow.

Model inputs are:

  • rainfall time series

  • temperature time series

  • digital elevation data for catchment

  • vegetation type (probably best sourced from LCDB2)

  • soil type (grouped, from NZLRI or NZ Soils Database)

  • sub-basin boundaries

  • drainage network branches (linking sub-basins).

Model outputs are changes in:

  • sub-basin plant canopy storage

  • sub-basin root zone storage

  • sub-basin saturated zone storage

  • sub-basin runoff time series

  • flow routed through the drainage network.

Although designed to model river flows, Topnet can also be used to model effects of land use change, as altering the averaged vegetation characteristics of a sub-basin also alters other model characteristics: canopy storage, aerodynamic roughness and albedo (used to calculate evaporation), and root zone storage (used to calculate transpiration and runoff).

Fahey et al (2004) report that Topnet has been used to model land use change in at least two large medium-sized catchments, Tarawera: 900 km2 (Woods and Duncan, 1999), and Shag: 544 km2 (He and Woods, 2001).

Topnet is used for flood routing investigations in the course of NIWA’s contractual work. They often form part of feasibility investigations for commercial agencies eg, power companies, irrigation ventures. Results do not appear in the public domain, unless the agencies’ proposals proceed to applications for resource consent. Reports may be public for a few instances of flood routing for regional councils.

Other models eg, forest cover flow change

Other models eg, forest cover flow change

Many other models are used overseas for rainfall-runoff assessment. One in particular, forest cover flow change, is specific to forest land cover changes and their effect on river flows. We have not discussed these as we are unaware of their use in the New Zealand landscape.

Monetary value of flood damage reduction

Our preceding recommendation is not to proceed with an attempt to quantify value of flood reduction, This relates to flood reduction in the hydrological sense ie, changes in the frequency, duration and magnitude of flood waves passing down a channel. Chapter 2 contains evidence (summarised above) that such changes are real, but too small to affect over-bank flooding.

Chapter 3 indicates that afforestation / reversion can reduce overbank flooding (and associated damage) by another mechanism: improved channel flood capacity as a consequence of reduced sediment yield. A monetary value could be attached to flood damage reduction through this process; the relevant recommendation appears under our sub-heading ‘Monetary value of sediment yield reductions’ (below).

Sediment yield

We recommend that MfE proceed with at least one study, to identify land in catchments nationwide, where sediment yield reduction could be a significant benefit of carbon-sequestration-related afforestation and reversion.

The rationale, from our reading of the publications, and also the information in several relevant unpublished reports, is that afforestation and reversion can cause substantial reductions in sediment yielded from river headwaters towards the downstream parts of medium to large catchments.

However, they do not invariably do so: the effect is not ubiquitous. It depends on afforestation and reversion being targeted towards parts of catchments that are both geologically unstable and high rainfall.

Where this combination exists, afforesting or reverting fairly small percentages of a catchment’s area can result in proportionately much larger percentage reductions in the catchment’s sediment yield.

The preceding point holds true so long as these small pockets are targeted onto erodible land. This is despite the likely diffusion of carbon-sequestration-related afforestation or reversion as numerous small pockets within larger areas of private farmland. This point is particularly important for any carbon sequestration initiative if sediment yield reduction is sought as a benefit.

We also caution that such a nationwide catchment study would reveal many catchments where much of the potential sediment reduction effect has already been achieved. Clough and Hicks (1993) point out that 5.9 million hectares of land were reserved as protection forests by the former New Zealand Forest Service between 1919 and 1987 (not just forest, but also scrub, tussock and alpine vegetation); and that a further 2.9 million hectares of similar land were reserved as national parks and scenic reserves by the former Department of Lands and Survey. This includes most catchments west of the South Island’s main divide, from Fiordland to Nelson; to catchments that have their headwaters in the North Island’s axial ranges from the Rimutakas to the Raukumaras; and to large western North Island catchments from the Wanganui to the Awakino.

Further reductions would have occurred in the last 20 years by land withdrawal from grazing through the tenure review process in the South Island high-country tussock lands, and the creation of a high-country conservation park system. In both the North and (to a lesser extent) the South Island, reductions would also have occurred through ongoing reversion in marginal hill country districts that are less accessible and not generally desirable for alternative land uses.

Nevertheless, our knowledge of the New Zealand landscape points at many other catchment headwaters where land is for the most part privately owned, with large areas farmed. In the North Island these include:

  • many (not all) catchments in the Northland and Auckland regions

  • Waikato west coast catchments north of the Awakino

  • small Taranaki and Wanganui catchments with headwaters close to the coast

  • sub-catchments west and north of the Rangitikei and Manawatu

  • catchments draining the hill country from East Cape to Cape Palliser.

In the South Island they include many sub-catchments that have their headwaters in tussock country well east of the main divide (though not main river headwaters that rise on the divide).

Should MfE wish to proceed with our recommendation, the necessary data are already to hand, stored in several public agencies’ geographic information systems. A database at NIWA’s Hydrology Centre in Christchurch (DM Hicks and Shankar, 2003) already contains:

  • boundaries for catchments draining to the sea (variable size from less than 100 km2, to in excess of 10,000 km2)

  • boundaries for medium-sized sub-catchments (variable size, but generally in the 100 to 1000 km2 range)

  • vegetation categories from the Landcover Database, second edition (LCDB2)

  • modelled sediment yields for different terrain-vegetation-rainfall combinations within each catchment.

With this database as a starting point, we suggest it would be worth:

  • overlaying cadastral information for land parcels, aggregated by ownership (public, Maori communal, company, private individual)

  • calculating percentage area in each category of ownership

  • then calculating percentage catchment area occupied by natural forest, planted forest, natural scrub and planted scrub, within each ownership category.

The GIS procedures could be undertaken nationwide by one of several agencies that possess the data (Linz, Terralink, Landcare Research, NIWA); NIWA might be in the best position given that it is the national research agency for measuring sediment yield. The procedures would provide MfE with reasonably recent and accurate figures for the percentage of each catchment presently occupied by privately owned land under woody vegetation (existing contributions to carbon sequestration). It could also provide the percentage occupied by farmland or tussock grassland (potentially available for carbon sequestration initiatives). These data could be provided both for terrains (see below) identified as having high sediment yields, and for terrains identified as not having these.

The percentage of medium-sized catchments are occupied by such reverted land, could also be derived by overlay of catchment boundaries on the Landcover Database (LCDB). Comparison of certain catchments between the first and second editions of the LCDB, or possibly between the Vegetation Map of New Zealand from the early 1980s (Newsome 1987) and the first edition of the LCDB (late 1990s), could more closely identify recent trends.

In undertaking this exercise, it should be kept in mind that private landowners will be receptive to afforesting or reverting only part of the area potentially available, regardless of the policy package available. However it seems to us that identifying where these potential areas are (ie, the target land) are essential precursors to any attempt at estimating what the sediment yield reductions might be. This is irrespective of whether the target lands are in fact on terrain with high sediment yields, and of what percentage of each catchment’s area they occupy,

We understand that MfE has nationwide estimates of land cover. If not done so already, a regional break-down of the figures would be most worthwhile in this regard, or a break-down by catchment within regions. It is the latter that is needed, to identify where the target land is.

Should MfE wish a more detailed pattern of spatial information than is available from the existing NIWA sediment yield model results (GIS layer and map), two others sources could be used.

CREAMS

The CREAMS model, operated by NIWA’s Water Quality Centre at Hamilton, was developed in the United States to predict changes in water quality as a result of agricultural land use practices. NIWA has successfully adapted it to New Zealand conditions, using the model to support research and contract investigations since about 1990. Its investigations have focussed on livestock exclusion from riparian zones, fertiliser application, and cultivation practices. As CREAMS incorporates a vegetation layer, it is capable of modelling catchment-wide land use change also. CREAMS is based on daily or event rainfall and utilises GIS data for slope, soil and vegetation cover on small areas within a catchment. It calculates runoff and water quality parameters for each element, by a modified version of the Universal Soil Loss Equation (USLE); it then aggregates them to catchment-mouth suspended sediment yield, phosphate load, nitrate load, and biological oxygen demand (by event or by year).

The data input and computer time required to run CREAMS would preclude its use for nationwide estimates. However as a tried-and-tested water quality model, it has value for detailed investigation of sediment yield reduction within particular catchments that can be regarded as test cases.

Sednet

The Sednet Model is a GIS-based model and designed to provide a sediment summary budget per sub-catchments within a watershed (measured as tonnes per year). The model was originally developed at the University of Canberra in Australia, and recently adapted by Landcare Research to fit New Zealand’s environmental parameters.

The model accounts for catchment sediment transport, various erosion types (earth-flows, hillslopes, banks and gullies) and is able to track sediment shifts downstream for an entire watershed. It predicts how much sediment will make it to the coast and from where the sediment and nutrients are coming. It also demonstrates how the sediment loads change with different land management, and over long time-scales – ie, pre-settlement sediment loads as compared to current sediment loads.

The model utilises a series of environmental parameters including landform information, floodplain information, daily stream / river flow rates, rainfall information, evapo-transpiration information, land use information, riparian vegetation information and known / modelled erosion information (eg, information from the New Zealand Empirical Erosion Model, NZEEM, see below). It has been applied for all the sub-catchments within the Manawatu River watershed under varying land use and erosion scenarios, but has yet to be tested or utilised elsewhere in New Zealand.

Monetary value of sediment yield reductions

We are less optimistic about recommending that MfE proceed with an attempt to quantify the nationwide value of sediment yield reductions. The rationale lies in our reading of the few available publications, plus our knowledge of many unpublished reports that have attempted to do this in past years. We conclude that while the monetary value of reductions in sediment yield has often been quantified (eg, as farm production from areas of pasture that are no longer silted) this was rarely done on a sound basis: the ‘saved’ production was hypothetical, not measured.

Should MfE wish to proceed with such a study, we recommend that it be done in a single region – perhaps in just one catchment within that region. A proposal along these lines (but for two full regions) has recently been developed by Ensis and New Zealand Institute of Economic Research (Ensis and NZIER, 2006b). We consider it better to have reliable values from a single thoroughly investigated catchment, than incomplete or unreliable ones from a study that spreads time and computer modelling resources thinly across too large an area.

A key element in calculating reliable monetary values would be to identify a catchment where local government agencies already have good records of historical costs for items such as silt clearance, regrassing, road repair, drain cleaning, channel dredging, gravel extraction. The Waipaoa (East Coast), Whareama (Wairarapa), Motueka (Tasman), and possibly Taieri (Otago) catchments, are promising candidate catchments.

Good historical records, and some monetary estimates, may be available for several of Environment Waikato’s Project Watershed sub-catchments (for example, Ritchie, 2000; Hill and Blair, 2005); also from current Environment Waikato initiatives for reducing sedimentation into Lake Taupo (currently not factored into benefit-cost analysis). Hill and Blair categorise the benefits resulting from soil conservation in the Middle Waikato Pilot Project, including economic benefits, but do not quantify these.

In Chapter 3 we mentioned improved channel flood capacity as a significant benefit. It does not appear to be an element in past cost-benefit analyses of sediment yield reduction (that focus on over-bank sedimentation or water quality issues). We suggest that it be included in a future catchment-based study.

Other catchments with records of bed level change following partial afforestation are the Waipaoa, Waiapu (East Coast), and Awhea (Wairarapa); possibly also the Esk (Hawkes Bay) and Pomahaka (Otago).

Terrestrial (on-site) erosion

Nationwide extent of carbon-sequestration-related reductions in terrestrial erosion could be estimated, if MfE wishes to do so. As with sediment yield, the identification of land potentially available, and whether it is on erodible or stable terrain, are essential first steps to take before attempting any estimates.

In our view, the only safe use of storm damage survey data for estimation at present would be to calculate the range of reductions observed (from Table 8), take the minimum value (about 50%), and say that this value will be equalled or exceeded during storms, where land is afforested or allowed to revert. This would be a crude approach, but better than nothing. The Institute of Geological and Nuclear Sciences (GNS) is also continues to maintain an as-yet unpublished database of studies of rainfall-induced landsliding (grouping the studies by erosion terrain) (Page in preparation). Once more readily available, this database may be a way to refine estimates by attaching different percentage reductions to ‘erosion terrains’.

Erosion surveys for regional state-of-environment reports seem a much better source of data about reductions attributable to afforestation or reversion. These surveys have the advantages of being region-wide, with large sample sizes, and tight error margins relative to sample means. Percentage reductions, estimated by ratio of sample means from Table 9, can be applied directly to each region where a survey was carried out: from this can be estimated what would happen if x% were afforested / allowed to revert. Survey data currently exist for seven of the country’s 15 regions / unitary authorities: Auckland, Waikato, Bay of Plenty, Gisborne, Manawatu-Wanganui, Wellington, and Tasman. A disadvantage of these surveys is that it would be statistically invalid to apply the percentage reductions to other regions (that don’t have any points included in the samples). Point sample surveys would need to be carried out in the other regions, before attempting estimates.

We do not recommend that MfE use soil conservation effectiveness surveys as a source of data for regional or national extrapolations: the sample sizes for these surveys are somewhat small, and error margins are large relative to sample means. They would create uncertainty, if percentage reductions were calculated from the data in Table 10 and applied elsewhere. The data are, however, useful for carbon sequestration initiatives, in the sense that they can be cited as demonstrating a consistent pattern of reductions in terrestrial erosion where trees are space-planted in pasture, as well as where land is afforested, or allowed to revert.

Two models are available from Landcare Research, which may be suitable for further work on nationwide estimates of erosion, as follows.

New Zealand empirical erosion model

The New Zealand Erosion Estimation Model appears to be the most promising way for estimating nationwide reductions in terrestrial erosion, if specific land types and percentages of land were to be afforested / reverted. This is based on recent work by Landcare Research with collaboration from other Crown Research Institutes (CRIs): it partitions the New Zealand landscape on the basis of rock type, landform (especially slope angle), and rainfall – the factors controlling erosion. The resulting groupings are called erosion terrains. The use of these terrains in combination with land use / land cover factors, as done in the New Zealand Empirical Erosion Model, is a promising development towards a more comprehensive and picture of erosion and sedimentation effects nationally.

NZEEM is an empirical model that predicts mean annual soil loss from annual rainfall, type of terrain and percentage of woody vegetation cover. It is calibrated on about 200 sediment yield data sets from most regions in New Zealand. NZEEM is claimed to be applicable to all types and sizes of catchments. It is a national development of an earlier study examining landslide susceptibility in the Manawatu-Wanganui region, validated by the incidence of erosion in the 2004 floods in that region (Dymond et al, 2006). It also builds on earlier work resulting in a database and digital map of mean specific sediment yield (kg / km2 / yr) produced by NIWA and Landcare as part of a project funded by the Foundation for Research Science and Technology (FRST), for studying carbon transfers associated with erosion (DM Hicks et al, 2004).

Land environments of New Zealand

LENZ (Land Environments of New Zealand) is an environmental classification intended to underpin a range of conservation and resource management issues. It is based on the fact that, rather than occurring randomly, species tend to occur in areas having similar environmental conditions. As a consequence, similar environments tend to support similar groups of plants and animals, provided they have not been substantially modified by human activity. LENZ uses these species-environment relationships by identifying and modelling climatic and landform factors likely to influence the distribution of species. LENZ uses these factors to define a landscape classification that groups together sites with similar environmental conditions, independently of current vegetation cover.

LENZ was originally envisioned as a framework for conservation management that would take advantage of the natural relationship between the environment and species distributions. However, it has had a much wider application. This is because the environmental factors that control the distributions of many land based plants and animals (temperature, water supply, availability of nutrients, etc) are also factors that provide major constraints on human land uses such as agriculture, horticulture, and forestry. Therefore it has been used to identify sites where similar problems are likely to arise in response to human activities, or where similar management activities are likely to have a particular effect.

Our supposition at the beginning of this project was that LENZ might be used in a similar way to group environments with particular erosion / sediment yield characteristics. However, the particular climate and landform parameters used for LENZ ie, those predicting dominant vegetation species distributions, are not necessarily the same ones determining erosion and sediment generation. A different set is needed for erosion. The approach described above for NZEEM and its associated terrain classification now seems more useful, unless a modification to the LENZ modelling is made: it should draw on the specific climate and terrain attributes that are critical in erosion or sediment generation (we understand these are also available on Landcare Research databases).

Terrestrial erosion reductions: nationwide estimates of monetary value

Attaching monetary values to reductions in terrestrial erosion is a difficult proposition. There may be ways to get around the absence of reliable nationwide information, for instance by using regional information, where it is of sufficient quality and resolution. Potentially usable studies (mainly recent ones commissioned by regional councils) are identified in our Chapter 6. They include benefit-cost analyses carried out for Environment Waikato’s Project Watershed, and an economic analysis of afforestation and poplar planting compared with farming on the East Coast of the North Island (McElwee, 1998). The recent proposal from Ensis and NZIER to MAF, to undertake one or two new regional assessments, may also come into this category.

Table 13 shows a preliminary and partial framework for setting out regional costs and benefits of soil conservation. It is based on unpublished work carried out by one of the authors (DLH) for Greater Wellington Regional Council in 1994. That work identified items for which data collection appeared feasible at the time. Table 13 shows in a matrix format (no actual data values are presented) how those data items could be adapted for evaluating the different afforestation and reversion scenarios discussed in this report.

McElwee’s approach is particularly interesting, even though it was taken without any consideration of the climate change mitigation potential of permanent afforestation and reversion. His study provides a good insight into multiple values of afforestation and reversion options on a farm and regional scale. It would be a useful model for further refinement to address the specific monetary estimates of benefits and costs of afforestation and managed reversion options for carbon sequestration.

Finally, we stress again the difficulty in providing the estimates requested in this project, ie, flooding and erosion benefits or costs associated with afforestation or reversion undertaken for the purposes of climate change mitigation or adaption. Estimates in either physical or monetary terms are difficult. However, it seems clear to us that for the greatest benefits to be realised, afforestation or reversion needs to the targeted at the type of land described at the beginning of Chapter 5. This will provide a full spectrum of ecosystem services, including greenhouse gas reduction, biodiversity protection, and soil and nutrient retention, as well as the erosion and flood reduction functions dealt with in the present report.

Table 13: Regional costs and benefits of soil conservation – a preliminary framework

View regional costs and benefits of soil conservation – a preliminary framework (large table).

[ |