• About this site|
• Site index|
• Contact us|
• Jobs

Quantification of the Flood and Erosion Reduction Benefits, and Costs, of Climate Change Mitigation Measures in New Zealand

Disclaimer

1 Introduction

2 Physical Effects of Afforestation and Reversion on Flood Occurrence and Other Hydrological Phenomena

3 Physical Effects of Afforestation and Reversion on Sediment Yield

4 Physical Effects of Afforestation and Reversion Effects on Erosion

5 Estimates of Monetary Benefits / Costs

6 Recommendations on Methodology for a National Perspective

7 Summary and Conclusions

8 Bibliography

9 Acknowledgements

2 Physical Effects of Afforestation and Reversion on Flood Occurrence and Other Hydrological Phenomena

How floods are altered by the planting or clearance of forests, is well known. For many decades now, from boreal forests to equatorial jungles, hydrologists have carried out scientific investigations of the topic. New Zealand investigations have been reviewed by Waugh (1980), Fahey and Rowe (1992), Maclaren (1996), and most recently by Fahey et al (2004). Rather than duplicate their reviews, our intent is to summarise key findings, and discuss their relevance to MfE’s brief ie, what do they actually tell us about the likely effect of afforestation or reversion carried out for the purpose of carbon sequestration?

Research studies in small catchments

These studies have been conducted in small catchments that have been entirely afforested or allowed to revert into scrub, or where standing forest or scrub have been cleared, specifically for the purpose of hydrological research. Relevant data come from two types of catchment investigation: where small catchments have been entirely afforested or allowed to revert into scrub; and where small catchments in standing forest or scrub have been cleared. Published summary data from these catchment investigations are grouped below into three tables, for water yield (runoff), floods, and low flows.

Table 1, from small catchment studies distributed throughout New Zealand, suggests a ubiquitous decrease in annual water yield when catchments are afforested or allowed to revert. A few of the reported decreases are quite small; most are in the 20 to 60% range; and some exceed 80%. All authors cited in Table 1 use the same hydrological statistic, averaged annual water yield. This enables direct comparison of their results. The only uncertainty when making comparisons, is the error margin inherent in averaging annual yields from a short record length. Few publications give the complete range of annual yield reductions.

Significant points to note about the published data in Table 1 are:

• water yields are averaged over several years (typically 10+ for standing forest, though as few as two for the pre-clearance vs post-clearance studies)

• in any one year, water yield reductions are greater or less than the average values

• most comment that reductions are greater in dry years and less in wet years are rainfall-driven.

Table 1: Effects of land use on water yield

View effects of land use on water yield (large table)

Table 2 (effects of land use on flood flows, from a subset of the same small catchment studies as Table 1) suggests very large decreases in flood peaks, but considerable variation in the magnitude of decreases measured in different parts of the country. Flood peak reductions are in the range >90% to 30% for small floods; 70% to 50% for annual floods; and 50% to 20% for large floods.

Table 2: Effects of land use on flood flows

View effects of land use on flood flows (large table)

Points to note about Table 2 are that:

• there is no uniformity in the nature of published flood reduction values

• some of the published figures are for annual floods ie, averaged values for the largest floods each year

• others are for small floods or ‘freshes’ ie, averaged values for floods that occur frequently during a year

• others again are for large floods or ‘storms’ ie, averaged values for floods that occur infrequently

• record lengths vary from 2 to 10+ years, but in all instances except Purukohukohu and Moutere, are too short for statistically reliable flood frequency analysis.

So a lot of the variation in magnitude of flood peak reductions relates to statistical presentation of results. All the authors have measured the same thing ie, a time series of flood peaks, but they have analysed them differently. This makes comparison of their summary values difficult.

An important non-effect, not apparent from Table 2, is that reduced flood peaks do not necessarily indicate reduced flood volumes. Some of the publications indicate that afforestation results in somewhat broader, lower flood waves. These discharge much the same total volume of floodwater downstream, but over a longer period of time. Other publications are silent on this point.

Table 3 is derived from a smaller sub-set of catchment studies, because few have published data about the effect of afforestation on levels of low flow. These few, mainly from the South Island, suggest residual low flows in late summer-early autumn decline by at least one-fifth (except for two short-duration records from Nelson).

Table 3: Effects of land use on low flows (from experimental catchments)

Significant points to note about Table 3 are:

• almost all catchments have low to moderate annual rainfall (mean annual ranging from 800 to about 1400 mm)

• the situation in high-rainfall catchments is either unknown or not analysed (probably the latter, as low flows would have been routinely measured in several other small-catchment studies).

Several publications about other research catchments state that their low flows declined after afforestation; or that these increased when forest was felled; but the publications do not give citable low flow statistics. It is unclear whether their statements are based on analysed data, or visual observation of hydrograph traces; or merely assume that, because annual water yields decline, residual low flows necessarily fall. Accordingly these catchments do not appear in Table 3.

Implications for forestry or reversion established primarily for carbon sequestration

At first sight, the New Zealand research findings might be seen as implying that forestry or reversion will result in:

• a substantial reduction in annual water yield

• an even greater reduction in flood peaks

• a modest reduction in low flows.

The research findings have been well publicised, perhaps leading planners and policy analysts to expect that carbon sequestration may have significant benefits in terms of flood reduction, or costs by way of reduced water availability for agricultural or industrial use.

We draw attention to what the authors of already-published reviews say. Maclaren (1996), after discussing several New Zealand extrapolations of research data to large, partly-forested catchments, states:

These examples illustrate the need to put land use in context. Vegetation may have a trivial influence on hydrological characteristics, compared with topography, the extent and influence of precipitation, and the structure of the soil and parent rock. Certainly, considerable caution needs to be exercised when extrapolating findings from one catchment to another.

To summarise: the benefit of forests in mitigating floods should not be overstated. Floods are common even in catchments of undisturbed native vegetation. Forests clearly can provide some smoothing of flood peaks in certain situations, but only in relatively small storm events and generally for small catchments and in areas close to afforested sub-catchments. Their main benefit lies in their ability to reduce sedimentation, if appropriate management practices are used.

Fahey et al (2004) state:

Afforesting close to 100% of small to medium size catchments that were previously in pasture or tussock grassland may reduce annual water yields by up to 55% and low flows by at least 20%, but the full effects will not be seen until canopy closure 5 to 10 years after planting.

Reversion of pasture to other forms of woody vegetation such as gorse, manuka or bracken will also reduce yields, but not to the same extent.

In larger catchments the effects of planned afforestation on water yields and low flows are likely to be less pronounced, because plantings will be at different stages of development throughout the catchment. Excluding high water yielding areas such as riparian zones from planting, coupled with careful management practices, will also help keep reductions in water yield to a minimum. Plantation forests on land previously in pasture or tussock grassland can also reduce flood peaks by a half to a third.

Fahey et al (2004) do not point out that the last sentence, like the others, is actually based on data from small catchments (generally less than 10 km²). They remain silent on whether flood peaks in large catchments are likely to be dampened as much. However, the italicized paragraphs are unaltered verbatim quotations from recent peer-reviewed reviews by leading researchers. They are the best summary of afforestation effects on flood peaks in the New Zealand landscape available at this time.

Large catchment investigations

Few large-catchment studies of afforestation effects on river flows have been carried out in New Zealand. Dons (1986) measured a 13% decrease in annual water yield from the 900 km² Tarawera catchment (Bay of Plenty), after 28% of its area was converted from scrub and bush into pine plantation between 1964 and 1981. However, he attributed just 5% of the decrease to afforestation; the other 8% being due to lower rainfall. Pearce (1987) estimated a 30% reduction in water yield from 120 km² of land afforested between 1960 and 1986 in the upper Mangatu and Waipaoa sub-catchments (East Coast). The afforested land was 36% of sub-catchment area (155 + 183 km²). Pearce’s method entailed subtracting evapo-transpiration estimates from rainfall records.

Mulholland (2006) forecast a 230% increase in peak discharge, from proposed deforestation of 225 km² in Waikato sub-catchments between Wairakei and Atiamuri (57% of four sub-catchments’ area). Volumes of flood runoff (into the Waikato River) were predicted to increase by 110–131 m3/s in a 20-year flood, and 222–239 m3/s in a 100-year flood. Mulholland’s predictions were made by applying flood runoff changes from the Purukohukohu experimental basins (headwaters less than 1 km² where flood runoff is surficial and its volume is just 2% of storm rainfall) to flow records from the Mangakara and Waiotapu (catchments 22 km² and 232 km² where the bulk of floodwater emerges by rapid translatory groundwater flow through pumice and ignimbrite, and its volume is a larger percentage of rainfall). The forecast flood runoff increases appear substantial (an additional 1 cumec per square kilometre of deforested area in a 100-year flood), but need to be scaled back (in proportion to the ratio of surface to subsurface floodwater contributions) before they can be applied to larger Waikato sub-catchments.

Turning specifically to low flows, Woods and Duncan (1999) used the Tarawera catchment to calibrate a hydrological model that produced similar outputs to Dons’ measurements. He and Woods (2001), using the same model to simulate a hypothetical 59% afforestation of the 544 km² Shag catchment (Otago), forecast a 45% reduction in low flows. In several medium catchments from Nelson through Canterbury to Otago, modelling studies by regional councils (unpublished but cited by Fahey et al, 2004) forecast that low flows will decline by more than 5%, if afforestation cumulatively amounts to more than 15% of catchment area.

These few investigations in medium or large catchments support Maclaren’s (1996) and Fahey’s (2004) caveats. They confirm it would be unwise to extrapolate absolute flows or percentage reductions in flows from small research catchments, where 100% of area has been allowed to revert or afforested, to the entire area of medium or large catchments.

Why this is so, can be elucidated by considering what happens when small-catchment data are extrapolated to unfarmed land, cumulatively less than 50% by area, diffused as small blocks through a larger area of farmed land. The runoff from a small block of afforested or reverting land reduces, by much the same percentage as small-catchment studies indicate. But this reduction is overwhelmed by ‘normal’ runoff entering the stream from a much larger area of surrounding farmland.

Taking a hypothetical example: if

• one-fifth of a sub-catchment is afforested, and the flood peak out of the afforested area reduces by 50%

then flood peak

• from the farmed four-fifths of the sub-catchment remains at 100%

• at the sub-catchment’s outlet reduces by 10% overall.

What actually happens to a flood wave passing down a large river is somewhat more complex, because its catchment’s flood response is not uniform. Quite apart from any vegetation effect, how fast a flood wave accumulates in the main channel, is influenced by:

• catchment shape (rounded, narrow, or regular)

• flow network topology (tributary branches numerous or sparse; long or short; junctions at wide or closely-spaced intervals)

• rainfall pattern (heavy rain falling on some sub-catchments but not others)

• hydro-geology (infiltration and storage of rainfall by soil and underlying rock, enabling either slow release of sub-surface runoff to channels, or fast surficial runoff).

Our example of a 10% reduction in flood peak at a partially afforested sub-catchment outlet, may diminish the catchment’s flood peak in the main channel, but alternatively it may have little or no effect. The magnitude of the effect will depend on whether:

• the sub-catchment is a large or a small part of the total catchment area

• its water enters the main channel close to other tributary junctions or tens of kilometres apart

• rain in the sub-catchment is heavy or light relative to what falls elsewhere

• the sub-catchment geology delays runoff to a greater degree than in other tributaries.

The question of what part of a catchment is being afforested becomes relevant here. In the New Zealand landscape, catchment headwaters are mountainous terrain or steep rangelands that intercept orographic rainfall (a narrow band of rain falling where a front is forced up over high ground). The headwater’s contribution to flood runoff is disproportionately large relative to their area; but little rain falls on middle catchments and floodplains, and these parts supply only a small proportion of flood runoff. Such orographic rainfall causes the frequent small floods that pass down New Zealand rivers each year, as well as many moderate floods (with return periods of 1 to 10 years). This is the circumstance where targeted afforestation of headwater sub-catchments will measurably diminish the floodwave that passes down a large catchment’s main channel in small to moderate floods.

However, large floods (>10 year return periods) tend to be caused when a wide moisture-laden air mass moves off the sea, dumping heavy rain on middle-catchment and downstream locations as well as on headwaters. These are the damaging floods that erode riverbanks, silt up unprotected floodplains, and breach stopbanks. In these circumstances, afforestation of a middle-catchment or a floodplain will have the same effect on runoff as afforestation of headwater areas. Regrettably, because of the other factors that determine passage of a large floodwave down a main channel, the vegetation effect will be quite small. This is so even where a significant proportion of the catchment is afforested.

The above simple outline of factors affecting flood response will demonstrate that vegetation is just one amongst many, and why vegetation change in just part of a catchment does not cause a substantial drop in main-channel flood peak. Rowe et al (2003) discuss these factors in more detail.

For medium-sized and large catchments, it is also necessary to discuss whether a vegetation-induced change to flooding in the hydrological sense (flood waves passing down a river channel), translates to a change in flooding in the risk sense (damage to people, buildings, infrastructure and economic activities) when flood waves are sufficiently large to spill across a valley bottom or floodplain.

Changes in flooding risk

It is tempting to speculate, from a planning or policy analysis perspective, that even a small reduction (a few percent) in peak flood flows through afforestation and reversion might have a significant economic implication. This view could be supported, in a hydrological sense, from observed changes to flood flow curves under different land uses. Swabey (personal communication, 2007)⁶ has commented:

Whenever a shift in the flow distribution curve occurs to a lower level, consequent issues are:

1. The floods may be slightly smaller, with most benefit coming at levels below the most extreme floods.
2. If fewer floods exceed protection levels (whether this is floodplain heights, floor levels, or stopbank levels), there will be a contemporaneous saving to the community.
3. Engineering design of future flood risk management methods will rely on the modified flow distribution curve, which now has a lower risk profile, meaning cost savings or enhanced protection levels will result.
4. The benefits of flood distribution translations to lower levels are not just about engineered protective works, but also about incremental change to natural systems like floodplains.
5. Floodplain flooding, which occurs at flow levels around the 100% to 10% annual return interval ie, once every 1–10 years on average, will be reduced too (even floodplain flooding of paddocks is damaging, particularly for farmers)

There are definitely New Zealand catchments where the above implications have been realised as a consequence of climate-induced shift in a flow distribution curve, or as a result of engineering-induced shift (damming or diversion of river flow). Some early examples are described by Poole (1981). Day et al (2007) give a recent summary of the topic (though the detail appears to be in background documents). Unfortunately, there is no published evidence – nor has any convincing unpublished evidence come to our attention – for any New Zealand catchment where the above consequences have been achieved by afforestation-induced or reversion-induced shifts in a flow distribution curve.

This negative finding is not what people expect. So the reasons why flood risk or damage reductions have not accrued from afforestation / reversion-induced hydrological changes, merit discussion here.

The discussion below is based on:

• accounts in Cowie (1957), who summarised damaging floods in New Zealand between 1920 and 1953 when very few rivers had comprehensive flood protection

• an analysis by Ericksen (1986) of floods from 1954 to 1985, decades when many large to medium-sized rivers had protective works installed along their downstream reaches

• because there is yet no nationwide summary of flood damage for 1986–2006, on personal recollection of news reports; discussions with regional council staff after events; and for some of the floods, personal observations.

1 Slightly smaller floods

Cowie’s (1957) account indicates regular inundation of farms, infrastructure and urban areas, on mostly unprotected valley bottoms or floodplains; including many moderate (1–10-year) floods as well as the large (>10 year) or extreme events (such as ‘Anzac rains’ of 1938). 1920–1953 follows the main period of European deforestation in most New Zealand districts, so it is possible that damage by moderate floods was more frequent as a consequence (though Cowie does not comment on this point, preferring to attribute instances of flood damage to occurrence of particular catchment rainfall patterns).

Ericksen’s (1986) account confirms that flood protection works (most designed to 20–50-year standard; some to 100-year or greater where rivers pass cities) have averted much damage to farmland and urban areas, by moderate floods (1- to 10-year) that formerly spilled over low-lying parts of now-protected floodplains. However he does not cite any instances of damage reduction on still-unprotected floodplains or valley bottoms, as a consequence of afforestation or reversion-induced changes in flood regime.

We are aware of several New Zealand catchments that had substantial vegetation change in their headwaters by 1986, either from deliberate afforestation (Waipaoa, Uawa, Waiapu, Esk, Awhea) or from reversion of abandoned farmland (Waitara, Whangamomona, Whenuakura, Waitotara). Our perception of the Wairarapa-Hawkes Bay-East Coast catchments is that rivers are aggrading, and there is ongoing damage to roads, bridges and farmland on valley-bottom floodways, even during moderate (<10-year floods). Our perception of the Taranaki-Wanganui catchments is that the river channels are incised, and damage during moderate (<10 year floods) is mostly bank scour or bank siltation; it takes a larger flood to spill over valley bottoms and cause flood damage (there have been three such events during the last 20 years).

2 Fewer floods exceed protection levels

Cowie (1957) reports instances of flood damage when stopbanks were overtopped eg, on the lower Waikato, the Manawatu, the Waimakariri, and the Clutha. It is clear from his account, that over-topping occurred because stopbanks were too low, poorly positioned, or incomplete, at a time when river protection was undertaken piece-meal by under-resourced local drainage boards.

Ericksen (1986) reports fewer stopbanks being overtopped. Again these were on older schemes where the standard of protection was low (20-year or less). He reports rather more instances of flood damage due to breach of stopbanks; and numerous occasions when there was considerable damage by back-up of tributary streams that were unable to discharge into the stopbanked floodways of main rivers. His analysis indicates that both causes of damage were associated with large floods (10 year or greater). Throughout his publication, he stresses that flood damage has decreased in frequency but paradoxically increased in magnitude along New Zealand’s protected rivers – because more assets have established on protected floodplains (eg, houses, factories, crop-growers, horticultural producers), and so there is risk of greater damage when stopbank failure or overtopping does occur.

Our personal recollection of major New Zealand floods since 1986 is, that flood damage continues to be caused by the same mechanisms identified by Cowie and Ericksen. For example:

• South Canterbury 1986: Road and bridge damage, gravel deposition on pasture, some damage to houses, where rivers either breached or over-topped small stopbanks (D Hicks personal observation).

• East Coast 1988: Waipaoa stopbanks did not overtop during Cyclone Bola. Extensive farm siltation, road blockage and house damage occurred in two circumstances: where tributary streams backed up on the Gisborne plains behind stopbanks; or where main rivers spilled across unprotected floodplains on the Tolaga flats and Waiapu valley (D Hicks personal observation).

• Hutt Valley 1990: Stopbanks did not overtop. Local blockage of roads and damage to houses where tributary streams and stormwater drains backed up (D Hicks personal observation).

• Taranaki 1990: Waitara river stopbanks did not overtop, but almost breached, during Cyclone Hilda. Widespread siltation and road blockage on unprotected valley-bottom floodways, throughout eastern Taranaki hill country (P Blaschke and D Hicks personal observation).

• Lower Waikato 1997: Stopbanks held where designed to 50 or 100 year standard, but over-topped where designed to 20 year standard. Widespread inundation of farmland, limited road blockage and house damage (Environment Waikato staff, personal communications and unpublished report).

• Manawatu-Wanganui 2004: Wanganui, Rangitikei and Manawatu River stopbanks did not overtop, but Moutoa floodway spilled over due to uncontrolled flow (damaged flood-gate). Extensive bank collapse, paddock siltation, road blockage, bridge collapse and house damage along floodways of Rangitikei and Manawatu tributaries, where stopbanks were either breached or absent (Horizons Regional Council staff, personal communications and unpublished report).

• Whakatane and Waimana 2004: River stopbanks did not overtop, but extensive paddock inundation occurred after groundwater flow (through pumiceous alluvium) locally under-mined stopbanks (Environment Bay of Plenty staff, personal communications and unpublished report).

• Wairarapa 2006: Ruamahanga stopbanks did not over-top. Paddock inundation, gravel deposition and silting were confined to the design floodway which includes farmland (Greater Wellington staff, personal communications).

• Northland 2007: Heavy rain in eastern Northland (March and July) caused extensive inundation and silting on the Hikurangi flats from back-up of tributary streams; but the Wairoa stopbanks did not overtop here, or on the western side of Northland (where neither the March nor the July floods were big events). Flood damage to houses in Kaitaia (July) was due to localised stopbank overtopping on an overflow channel (Tarawhaturoa), not the main Awanui or its spillway (Whangatane), which performed to design specifications. There was much riverbank damage, paddock silting and road blockage in valley bottoms upstream (middle and upper reaches of Awanui lack stopbanks), particularly on one tributary (Victoria) which is still adjusting to re-alignment of its channel some 40 years ago. Flood damage to houses and road blockage in Kaeo (July) included over-topping along a short length of diversionary stopbank (the only one in the catchment), but otherwise was due to buildings and roads being on or close to an unprotected floodplain, now 0.5–1 metre higher due to long-term siltation, than when many of the houses were built. (NRC staff and local residents personal communications).

• The Awanui river in particular is a graphic example of how flood damage still occurs in a medium-sized catchment, despite widespread reversion and afforestation in its headwaters over the past 20 years. Damaged houses, roads and farmland along the Kaeo river are in areas settled over a hundred years ago, so cannot be attributed to recent developments creating new risks. The cause of higher flood levels here appears to be not land use change (over half the catchment is bush and scrub), but floodway siltation arising from channel constriction and sediment trapping by vegetation. (Northland Regional Council, 2007a, b; R Cathcart and N Mark-Brown personal communications).

Our conclusion from the above summary is that flood damage during major New Zealand floods is rarely caused by floods exceeding design protection levels on stopbanks on rivers. It is more often caused by unpredictable breaches of a weak point in a stopbank when the river is below bank-full stage; or overflow of backed-up tributaries onto land behind stopbanks when the river is close to bank-full; or spillage across floodways. We agree with the observation that:

these risks do not change even when afforestation is factored in. Essentially they are a given, no matter what the scenarios in the catchment and thus can be discounted in risk change considerations (S Swabey personal communication.)

except that spillage across floodways may be reduced where siltation declines following afforestation.

3 Modified flow distribution curve with lower risk profile

We are not aware of any instances in flood design within New Zealand, where a river engineer has altered scheme design parameters or reduced scheme maintenance, in the expectation that flood peaks will reduce following headwater afforestation. There are certainly rivers where headwater afforestation has been carried out to assist river management (Waipaoa, Uawa, Waiapu, Awhea, Tauanui, Hiwinui). However scheme reviews, eg, Peacock et al (2000), and Gunn et al (2004), clearly state that afforestation has been carried out to reduce sediment yield, and thereby help maintain channel flood capacity.

These documents contain no evidence that afforestation has altered the hydrological parameters (peak, duration or frequency) of large floods; but the reviews (and additional scientific publications) confirm that in targeted catchments, channel aggradation has declined in downstream reaches since afforestation, and has reversed (started to degrade) in some (not all) headwater tributaries. They attribute maintained channel flood capacity and reduced flood damage to this effect (in part, alongside other factors such as good floodway operation). We shall discuss the evidence in Chapter 3 (sediment yield), confining our mention of it in the current chapter (hydrological changes) to an observation that afforestation-induced sediment yield reduction appears capable of reducing flood risk/damage.

4/5 Incremental change to natural systems, smaller floodplain floods

We acknowledge the possibility that afforestation of large areas within catchment headwaters may reduce hydrological parameters of moderate floods (1–10 years), enough for flood risk / damage to become less frequent on unprotected floodways and floodplains, simply through fewer moderate floods spilling across them. However, we must point out that there is currently no New Zealand flood study that can be cited either to support or disprove this possibility.

Conclusion: water yield, floods and low flows

Large changes in water yields, floods or low flows have been observed only in small catchments, where close to 100% of catchment area has been afforested or retained in native cover. The few published studies of partial afforestation in large catchments, all report much smaller changes in river flow.

Absolute flows or percentage reductions in flows, from small research catchments entirely under forest or scrub, should not be extrapolated to the entire area of medium or large catchments, if afforestation / reversion is patchy and occupies a small percentage of catchment area. Reduction in runoff from small blocks of afforested or reverting land is overwhelmed by ‘normal’ runoff from much larger areas of surrounding farmland.

Where afforestation / reversion occupies a large percentage of catchment area, flow changes can be forecast either by conventional hydrological techniques or by computer-generated flow models. However in this circumstance, vegetation-induced runoff reduction is dampened by other influences on river flow (notably channel network topology and basin hydrogeology).

Changes in flooding in medium to large catchments as a consequence of afforestation or reversion, are real in the hydrological sense. That means vegetation change does alter the frequency, magnitude and duration of small flood waves passing down a catchment’s main channel.

However, changes in flood hydrology do not translate to reductions in flood risk / damage (to people, buildings, infrastructure and economic activities), when moderate or large flood waves spill across a valley bottom or floodplain. There is documented evidence that much past flood damage in the New Zealand landscape has been caused by factors which afforestation or reversion cannot influence. These factors include under-designed flood protection schemes, unpredictable scheme failures (breached stopbanks, jammed floodgates, open or missing valves), and developments sited on flood-prone land.

Nevertheless, there are instances in the New Zealand landscape, where over-bank flooding (and associated flood risk/damage) has been reduced as a consequence of afforestation. In these instances, the mechanism is reduced sediment yield followed by channel degradation (which improves channel flood capacity). This mechanism will be discussed in Chapter 3.

Low flows

There is some evidence that changes in the frequency of low flows from partial afforestation or reversion in medium to large catchments, are sufficient to impact on water use. Decreases of 13% to 45% in minimum summer low flows have been measured or modelled, as a consequence of 2% to 59% of area being afforested, in several catchments over 500 km². In some other catchments (smaller but still medium-sized), modelling studies indicate that low flows will decline by more than 5%, if afforestation cumulatively amounts to more than 15% of catchment area.

In a catchment where a high proportion of flow is already allocated for commercial use (in the form of water permits), such a decrease would result in the regional council imposing temporary restrictions on take – something which already happens in dry summers – for longer than is currently the case. This would represent a real economic cost to permit-holders ie, would be an adverse effect of afforestation / reversion.

Summary of impact of afforestation on flooding under different scenarios

Diffuse reversion / afforestation, by individual owners on parts of their own properties (that continue to be farmed), will create a fragmented pattern of tree cover on just a small percentage of catchment area, in the midst of other land in pasture or tussock. This scenario 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 (dispersed amongst other properties that continue to be farmed), will create contiguous blocks of tree cover on a significant percentage of catchment area. This scenario will alter the magnitude and duration of small and moderate flood waves (1- to 10-year frequency) passing down the main channel of a catchment. Consequential reductions in overbank flooding and associated damage are possible on unprotected rivers; but for the few New Zealand catchments where there has been widespread afforestation / reversion, as yet there is no evidence that such reductions are attributable to vegetation-induced changes in flood frequency. On protected rivers there will be little or no reduction in overbank flooding and associated damage from small to moderate flood waves, because these events are contained by the flood protection works.

Whole-catchment reversion / afforestation by a public agency intervening to change land use on all properties (none of which continue to be farmed), will ensure tree cover on almost the entire catchment area. This scenario will substantially alter the magnitude and duration of small to moderate flood waves, but will just slightly alter large flood waves (>10-year frequency) passing down the main channel of a catchment. It will not reduce large flood waves enough to avoid overbank flooding and associated damage, either on unprotected or protected rivers, because vegetative retardation of runoff is outweighed by extreme rainfall during large events.

There are instances where extensive flooding and damage have been avoided in catchments following substantial but not complete headwaters afforestation (ie, a transitional scenario between widespread and whole-catchment planting. For these catchments there is evidence that the mechanisms are geomorphic, not hydrological (Kasai et al, 2005; Liebault et al, 2005). The rivers in these catchments pass more floodwater due to degradation of their channels, after sediment yield has been reduced by changes in land use. This mechanism can produce consequential reductions in flooding and damage during moderate floods on unprotected rivers; and during large floods on protected ones.

[ Previous|Next