Sediment budget to assess the geomorphic effect of a cyclonic storm, New Zealand

ELSEVIER

Geomorphology9 (1994) 169-188

Sediment budget to assess the geomorphic effect of a cyclonic storm, New Zealand M.J. Page, N.A. Trustrum, J.R. Dymond l_andcare Research New Zealand, Private Bag 11052, Palmerston North, New Zealand

(Received September 14, 1992; accepted November 23, 1992)

Abstract A short-term sediment budget was constructed to assess the erosion response of a soft-rock hill country watershed to an intense rainstorm event. The watershed is located in a landslide-prone area on the east coast of the North Island, New Zealand, During March 1988, 753 mm of rain was recorded over a four-day period with 320 mm and 329 mm on successive days. Known as Cyclone Bola, this was the largest rainstorm event in the Tutira watershed in the 93-year rainfall record. The budget quantifies the total sediment generated during the storm, the relative contribution of erosion processes involved, the amount of sediment held in storage and the amount discharged into two lakes within the watershed. A landform map of the watershed was constructed and the contribution of each landform type to the total budget was calculated. A total of 1.35 ( +_.0.13) million m 3 of sediment was generated during the storm at an average of 420 m3/ha. This is equivalent to a denudation value of 42 mm for the total watershed (3208 ha) and 83 mm for the landforms severely affected by landsliding ( 1427 ha). Of the sediment generated, 21% remained on hillslopes, 22% was deposited on valley floors, 51% was deposited on the lakebeds and the remaining 6% was discharged from the watershed via the lake outlet. Most of the sediment generated during the storm was from primary source areas on hill slopes, with sediment in secondary storage providing only a small contribution. Landslide erosion was the main process, accounting for 89% of the sediment generated. Channel, tunnel gully and sheet erosion were only minor contributors to the budget. Six hillslope landforms, which occupy only 44% of the watershed generated 90% of the sediment. The results of this sediment budget, when put in context with the storm magnitude-frequency history being analysed from lake cores, contribute to the identification of sustainable land use and management of soft-rock hill country.

1. Introduction Landslides are the most significant type of erosion occurring on the steep hilly landscapes of New Zealand. Covering more than 60% of New Zealand's land area, these steeplands were largely forested until about 150 years ago. The area o f forest has since been dramatically reduced by burning and logging and replaced by pasture for livestock farming (Blaschke et al., 1992). A major consequence of deforestation has been an increase in landside erosion (DeRose et al., 1993), with 0169-555X/94/$07.00 © 1994 Elsevier Science B.V. All rights reserved SSDIOI69-555X (93)E0058-K

subsequent reduction in productivity (Trustrum et al., 1990) and increasing flooding and sedimentation downstream. Large-magnitude rainstorms are the main cause of this land degradation. Such storms are the main agents of sediment generation, movement and discharge from steep watersheds and account for most of the slope failures on steep areas (Dyrness, 1967; Renwick, 1977; Lehre, 1982; O'Loughlin et al., 1982; DeRose et al., 1991; Trustrum and Page, 1992; DeRose et al., 1993). Extreme storms are also considered to be a major factor

170

M.J. Page et al. /Geomorphology 9 (1994) 169-188

affecting geomorphic processes and landscape evolution (O'Loughlin et al., 1982; Phillips, 1988; DeRose et al., 1993). The geomorphic importance of an event or its ability to affect the shape or form of the landscape has been defined by Wolman and Gerson (1978) as geomorphic effectiveness. High-intensity rainstorms are a feature of New Zealand's climate, especially on the east coast of the North

I E 176"50' Long.

Island where some of the most erosion-prone hill country occurs. The Tutira watershed (Fig. 1), north of Napier, is representative of much of this steep landslide-prone hill country. An excellent rainfall record from Lake Tutira, together with good historic documentation, identifies numerous erosion producing storms over a 93-year period. Cyclone Bola, which occurred in March 1988, is the largest rainfall event in

PHYSIOGRAPHIC AREAS ~Basin ~

and range terrain

Steepland terrain

39"10' S

1.at.

0

1

2

3kin

I

I

I

I

~

\

Study Are~_~

Fig. 1.LocationoftheTutirawatershedshowingphysiographicareasandmajorstreams.

M.J. Page et al. /Geomorphology 9 (1994) 169-188

that record. It provides a benchmark to assess the geomorphic effectiveness of such storms, together with the implications for sustainable land use. In this paper we use the sediment budget approach (Sutherland and Bryan, 1991; Reid, 1990) to quantify the total sediment generated by Cyclone Bola for each erosion process, the contribution from each landform type, the amount of sediment stored on the landscape, and sediment discharged into Lakes Tutira and Waikopiro. It is a short-term sediment budget spanning only one large-magnitude event, and therefore does not quantify rates of various sediment transfer processes and the residence times of slope and valley bottom sediment storage elements (Lehre, 1982; O'Loughlin et al., 1982; Sutherland and Bryan, 1991). However, the results allow both the on-site and off-site impacts of cyclone-induced erosion to be assessed. This sediment budget is part of a larger study to reconstruct the long-term erosion history of the Tutira watershed (Trustrum and Page, 1992). The overall goal of the study is to provide information which can be used to help assess sustainable land use in landslideprone soft-rock hill country. By determining the magnitude and frequency of storm events as preserved in stratigraphic cores taken from the sediments in Lake Tutira, Cyclone Bola can be put into context with the erosion history of the watershed. Integration of this storm record with the results of the sediment budget will aid in understanding the role that cyclonic storms play in landscape evolution and the effects that land use change has had on this role.

2. Study area

The Tutira drainage basin is situated in the northern Hawke's Bay region on the east coast of the North Island, New Zealand (Fig. 1). The region comprises an uplifted, gently dipping and dissected terrain of mildly deformed Late Pliocene and Early Pleistocene sediments. It is tectonically active, forming part of the "East Coast Deformed Belt" (Spi3rli, 1980). Uplift, which commenced in the Pleistocene, is still continuing, with subsequent downcutting by rivers and streams. The study basin is underlain by sandstones and siltstones (silty mudstones) interbedded with limestones and conglomerates (Lowe, 1987). The strata

171

dip uniformly at 2° to 10° southeast towards Hawke Bay. The Tutira basin has an area of 3208 ha and contains two lakes near its southern boundary, Lake Tutira ( 179 ha) and Lake Waikopiro ( 11 ha). Both are landslide dammed lakes (Adams, 1981; Lowe, 1987), formed by collapse of part of a hillside blocking the valley outlet at what is now the southern end of Lake Tutira. The bed of Lake Tutira is a drowned valley consisting of a series of interfingered spurs with steep slopes descending to a narrow, central meandering valley floor (lake basin). Lake Tutira has a mean depth of 20.8 m and maximum depth of 42 m. Lake Waikopiro has a less irregular lake bed morphology and a maximum depth of 16 m. Radiocarbon dates from cores of lake sediments indicate that the lakes were formed approximately 6500 yr B.P. (Trustrum and Page, 1992). Mean annual rainfall at Tutira is 1438 mm (19511980). Rainfall is highly variable in distribution with frequent lengthy droughts and high intensity rainstorms. July and August are the wettest months (32% of annual rainfall) and September to November the driest months (19% of annual rainfall). The Tutira district has been mantled by a number of tephras. A record of these has been preserved in swamp and lake sediments (Eden et al., 1993). Of these, Taupo and Waimihia Tephras are the two most distinctive and widespread on the landscape. Taupo Tephra ( 1850 yr B.P.) occurs as pumiceous lapilli mixed through the soil in the A and B horizons. Waimihia Tephra (3280 yr B.P.) occurs as a more persistent, discrete layer and is the thickest tephra present in the district. Beneath Waimihia Tephra on more stable surfaces there is a brown sandy layer which is considered to represent a number of Holocene rhyolitic and andesitic tephras which have been mixed and weathered so that individual layers are unrecognisable. All of these tephras are present on low angle stable landforms but have been largely removed from steeper hillslopes by erosion. The watershed comprises two distinct physiographic areas (Fig. 1): the area north of Lake Tutira (2021 ha), which consists of Pliocene sediments, and is characterised by a "basin and range" terrain, and the southern area surrounding Lake Tutira ( 1187 ha), consisting of Pleistocene sediments with a steepland terrain. The northern area consists of a series of NE-SW trending asymmetrical ridges or cuestas which alternate with gently dissected basins and downland. The cuestas have

rolling

~~

hillslop

-~ ~

_

Fig. 2. Major landforms of the steepland terrain surrounding Lakes Tutira and Waikopiro,

~,

Plateau

e~

M.J. Page et al. /Geomorphology 9 (1994) 169-188

long inclined dip slopes and shorter, steeper scarps. The stream system is entrenched for most of its length. Because of the generally subdued nature of the topography the tephra mantle is thicker and more uniform and landslides are less common than in the area surrounding Lake Tutira. By contrast the southern area is characterised by landslide-prone, moderately steep to steep dissected hills with a series of narrow alluvial valley floors which lead into the lakes (Fig. 2). The division of the watershed into northern and southern areas permits the sediment budget data to be analysed in terms of these two contrasting landscapes.

3. Land use history The Tutira drainage basin prior to Polynesian settlement (ca. 700-800 yr B.P) was covered with forest. Lowland conifer-broadleaved forest occurred in the central and southern part of the basin surrounding Lake Tutira, while in the northern part of the basin, at higher altitudes, beech-conifer-broadleaved forest predominated. At the time of European arrival in 1873 most of the forest had been destroyed by a succession of fires and replaced by bracken fern (Guthrie-Smith, 1953). This is supported by pollen analysis of lake sediments (J.M. Wilmshurst, pers. commun., 1991). Guthrie-Smith, who arrived on Tutira Station in 1882, described how evidence of the burnt forest increased inland from the coast. It appears that the steep hills in the southeast of the basin had been devoid of forest for at least several hundred years, while some areas in the north of the basin had only been recently burned. This would indicate a succession of fires, which were probably associated with initial Polynesian settlement along the coast, 11 krn to the east. Europeans began the conversion of fern to pasture by a combination of burning and fern crushing. This conversion has been spasmodic and affected by reversion. However, since the introduction of aerial topdressing in the early 1950s most of the watershed has been in pasture. At the time of Cyclone Bola, mixed sheep/beef cattle farming was the main land use within the watershed with only a few small blocks of second growth forest and scrub remaining.

173

4. Cyclone Bola Between 6 and 9 March 1988 Cyclone Bola moved southwestwards over New Zealand bringing heavy rain to the east coast of the North Island. During this fourday period 753 mm of rain was recorded at Lake Tutira, with 320 mm and 329 mm on successive days. This represents the largest rainstorm event at Tutira in the 93 years since rainfall records began. Rainfall records from Napier, 30 km south of Tutira, together with observations by local inhabitants, indicate that Cyclone Bola also exceeded anything in the preceding 70 years. This would make it the largest known storm in the district, with an average return period of well in excess of 100 years. Damage in the vicinity of Tutira was estimated at $NZ 12 million, while throughout the North Island repair costs exceeded $NZ 120 million (Trotter et al., 1989). High-intensity rainstorms are characteristic of the climate of the east coast of the North Island. Historical records document numerous storm events which produced severe erosion. Prior to Cyclone Bola the largest rainstorm was in 1938 when 692 mm fell in four days at Tutira. Analysis of the 93-year rainfall record shows that there have been 31 storms in excess of 250 mm, at an average of 3.3/decade. (Observations by the authors indicate that 250 mm in two to three days is the approximate threshold value required to induce significant erosion in landslide-prone hill country.) Of these, 20 exceed 300 mm, at an average of 2.1/decade and eight exceed 400 mm, at an average of 0.86/decade. The major erosion producing storms prior to Cyclone Bola were in 1938 (692 mm), 1924 (405 mm), 1917 (510 ram) and 1910 (420 mm).

5. Erosion impacts of Cyclone Bola The advent of Cyclone Bola provided a rare opportunity for directly observing and measuring the links between erosion and sedimentation and the geomorphic effects of such storms on the landscape. Four main erosion processes occur within the Tutira watershed: landslide, channel, tunnel gully and sheet erosion. Of these, landsliding is clearly the predominant process and therefore has the greatest geomorphic influence.

174

M.J. Page et al. / Geomorphology 9 (1994) 169-188

5.1. Landslide erosion During Cyclone Bola landsliding was widespread in the southern area of the watershed surrounding the two lakes where the majority of the steep terrain occurs (Photo 1 ). On these landslide-prone landforms much of the original tephra has been removed and tephra is now restricted to stable sites such as spur crests and hollows and only rarely occurs on steeper slopes. Landslides occurred on a range of landforms on slopes steeper than 18°. The scar depth ranged from 13 to 390 cm, with a mean of 89 cm. The shear plane was usually parallel to the former ground surface and on steeper slopes was at or near the bedrock contact, although this was often indistinct due to weathering of the soft parent material. On less steep slopes the shear plane was often located in the regolith. Any sheet and rill erosion which occurred on the landslide scars during the storm have been included in the landslide measurements. The typical hillslope morphology is one of a series of parallel or subparallel spurs and hollows leading away from the

ridge crest. Landslides commonly occurred within the steeper, upper two-thirds of the hillslope just below the ridge or spur crest, with the shallowest landslides tending to occur on the steepest slopes. The debris from the landslides formed a highly fluid, chaotic mix of soil, other regolith material and pasture which travelled considerable distances downslope from the scar sites. The elongated "debris tails" either entered directly into streams at the base of the hillslope, or more often accumulated on the colluvial footslope or in the interspur hollows. Cyclone Bola initiated a number of hillslope gully features which, because of their limited occurrence and similarity of form, were included in the landslide component of the sediment budget. These include the excavation of a number of colluvium-filled hollows in first-order basins (Photo 2). Following initial landsliding, deep (2 to 3 m) incisions (gullies) developed, exposing a number of buried soils overlain by colluvium. Taupo and Waimahia Tephras were present in the oldest of these soils.

Photo 1. Lake Tutira and surrounding steepland terrain following Cyclone Bola. Sediment on valley floor is largely derived from landslides. (Photo: N.A. Trustrum, March 1988).

M.J. Page et al. /Geomorphology 9 (1994) 169-188

175

Photo 2. Excavation of colluvium filled hollow within moderately steep hillslope landform. Note shallow landslide (centre) and buried soil (right). (Photo: N.A. Trustrum). A similar excavation of material occurred within the long, narrow ephemeral watercourses which drain the long, steep escarpment along the southeastern margin of the watershed. Here shallow ( 1 m) intermittent gullies developed down to bedrock, which in some cases initiated landslides by undercutting of the toe of the adjacent slope. 5.2. Channel erosion

Channel erosion occurred in all the narrow alluvial valleys which lead into Lakes Tutira and Waikopiro (Photo 3). Channel erosion refers to all the material removed by the stream and consists of lateral bank erosion, headward extension of the stream channel and downcutting of the streambed. Typically, lateral bank erosion removed 1 to 2 m of bank material, although the range was from 0 to 10 m. In some places lateral bank erosion toppled 10 m high willow trees in plantations established for stream stabilisation. All the material eroded from these channels entered the lakes

or was deposited on the valley floors, the stream beds having no significant storage capacity. Cores taken in these narrow valleys show that they were once arms of the lakes which have since been infilled with alluvial sediments. Prior to European settlement these valleys were poorly drained and swampy with unconfined stream courses. Subsequently, drainage ditches were dug, confining streams into straightened channels, and initiating channel erosion. Aerial photographs taken after the large 1938 storm show the development of only minor channel erosion. This has since increased significantly and channels are now 5 to 20 m wide. Stream beds are now typically 2 to 4 m, and in one case 8 m, below the valley floor, exposing a number of buried soils and, at stream level, a forest soil with in situ tree stumps which have been dated at 600 to 700 yr B.P. Areas of the valley floor adjacent to the streams continue to be the site of much deposition and during Cyclone Bola up to 2 m of sediment was measured at some sites, with a mean thickness of 28 cm. During the

176

M.J. Page et aL / Geomorphology 9 (1994) 169-188

Photo 3. Channel erosionon alluvial valleyfloor. Notelateral bankerosionand downcuttingof stream bed. (Photo:N.A.Trustrum). storm the lake level rose by 3.5 m and extended at least 200 m up several of these valleys, with the result that large amounts of sediment were deposited at this temporary lake margin.

approximately 40 cm, with an average thickness of 80 cm. Tunnel gullies were observed on slopes between 10 and 25 °. They were normally 1.5 to 2.0 m in width, with an average depth of approximately 1 m. The length of the feature was dependent upon slope length.

5.3. Tunnel gully erosion 5.4. Sheet erosion

Tunnel gullies in the study area are associated with the presence of tephra. For this reason most occurred in the northern area of the watershed where slopes are less steep and tephra deposits are thickest. They are initiated by subsurface concentration and flow of water above an impervious layer of compact weathered tephras. This causes scouring and removal of the overlying, poorly consolidated, low strength Waimihia Tephra, to produce narrow tunnels or pipes. Areas of the ground surface eventually collapse due to lack of support, producing open shafts. Normally a number of such "pot holes" appear in a line down slope, indicating the direction of the tunnel (Photo 4). The minimum thickness of porous, poorly consolidated tephra necessary for tunnel gully formation is

Sheet erosion, which refers to removal of surface particles by non-channelised overland flow of water, was not considered to be a major erosion process during the storm. Practically the entire watershed has a good pasture cover and no large areas of bare ground existed prior to the storm. As with landslides the majority of landforms susceptible to sheet erosion are in the steep southern area of the watershed. Here sheet erosion was in the form of diffuse movement of particles through the pasture sward. No accumulation of sediment due to sheet erosion was observed. Bare ground was limited to farm tracks, stock paths and areas around stock ponds. Rills were present on some farm tracks and, although these features may have been present prior to

M.J. Page et al. / Geomorphology 9 (1994) 169-188

177

Photo4. Tephracoveredhillslopewith tunnel gullyerosion (arrowed) and shallowlandslides (foreground).(Photo:N.A. Trustrum). Cyclone Bola, they would have been enlarged during the storm.

6. Sediment budget This study uses a sediment budget approach (Dietrich et al., 1982) to assess the geomorphic effectiveness of one storm while allowing the individual processes to be considered separately. The sediment budget presented is a simplified one which because of the short time frame, does not quantify erosion rates and residence time of sediment. However, the role of hillslopes and valley floors in the storage of storm sediment is identified. A sediment budget is a quantitative statement within defined spatial and temporal boundaries of the generation, storage and discharge of sediment. A simple equation expresses the relationship: sediment generation = sediment stored + sediment discharged

The budget was designed to calculate: 1. The amount of sediment generated during Cyclone Bola within the Tutira watershed. 2. The contribution of each erosion process to this total. 3. The contribution of each landform type to this total. 4. The amount of sediment deposited on the landscape. 5. The amount of sediment discharged into Lakes Tutira and Waikopiro. A short-term sediment budget for the Tutira watershed during Cyclone Bola can be expressed as: S=H+V+L+D

where S is the volume of sediment generated during Cyclone Bola, H is the volume of sediment deposited on hillslopes as debris tails, Vis the volume of sediment deposited on valley floors, L is the volume of sediment deposited in the lakes, and D is the volume of sediment discharged through the lake outlet. The volume of sediment generated during Cyclone Bola (S) was calculated by assessing the contribution made by the four main erosion processes operating within the watershed. Therefore:

178

M.J. Page et al. /Geomorphology 9 (1994) 169-188

s=&+&+s~+& where S~ is the volume of sediment generated by landslide erosion, Sc is the volume of sediment generated by channel erosion, St is the volume of sediment generated by tunnel gully erosion, and Ss is the volume of sediment generated by sheet erosion.

7. Measurement and results Sediment volumes generated by landslide and channel erosion were calculated from field measurements. For tunnel gully and sheet erosion (which field observations indicated were only minor contributors to the budget) estimates of maximum possible volumes were made.

7.1. Landslide erosion (Sl) Because of the very large number of landslides and their distribution within the watershed, a statistical sampling technique was necessary in order to calculate the volume of sediment produced by landslides (SO. S~ can be written as:

s,=ff

Similarly, a 10 mm dot grid was laid over the mosaic to estimate the area of landslide scars, as. The number of dots falling on landslide scars was multiplied by the horizontal area of a grid cell (which varied over the mosaic according to surface elevation above sea level) to estimate as as 1,330,000 m 2. The product of ~ and as gives the estimate of Sj as 1,184,000 m 3. An error analysis using a formula derived in the Appendix gives the standard error as 5% of $1.

7.2. Channel erosion (So) Six major alluvial valley floors were identified in which channel erosion was occurring. The volume of sediment eroded from each valley floor was calculated by a combination of field measurement (with a handheld tape) and comparison of pre and post Cyclone Bola aerial photographs. Morphological evidence, and where possible exposed roots and in situ tree stumps, were used to determine lateral bank erosion and bed incision. The total volume of sediment generated by channel erosion was calculated to be 28,000 m 3. Table 1 lists the six alluvial valley floors and their contribution to this total. A measure of the relative severity of channel erosion is given on a volume/ha basis.

Az(x,y) dxdy

scar area

where ffdxdy is the integration over the total landslide scar area; and Az(x,y) is the vertical depth of eroded soil as a function of easting and northing (x,y). The equation may be simplified to: S, = - ~ " as where ~ is the mean depth of soil eroded on landslide scars and as is the total landslide scar area in the watershed. A statistical technique was designed to estimate from a systematic sample of field measurements of landslide scar depth. A grid of regularly spaced dots located 15 mm apart was laid over an aerial photo: graphic mosaic of the watershed ( 1:5000 scale) and wherever a dot lay on a landslide scar, that point was selected for a field measurement of vertical eroded depth. The systematic sample comprised 126 points (designed to give a standard error of 5%) and gave a mean depth of 89 cm.

7.3. Tunnel gully erosion (St) Calculations of sediment produced by tunnel gullies are difficult to make due to the small, subsurface nature of the features and the absence of a significant eroded surface. Table 1 Estimates of sediment eroded from alluvial valley floors in the Tutira Watershed during Cyclone Bola. Alluvial valley floor

Surface (ha)

Volume (m 3)

Volume (m3/ha)

Hut Oporae Ridells Sullivans Sandy Creek Kahikanui

3.4 16.4 10.4 15.1 43.3

257.4 1015.8 4170.6 6516.0 7686.5

75.7 61.9 401.0 431.5 177.5

31.7 120.3

8740.5 28,386.8

275.7

M.J. Page et al. /Geomorphology 9 (1994) 169-188 Table 2 Sediment budget for the Tutira Watershed during Cyclone Bola Budget Location component

Input

Process

This is probably an overestimate, as it is recognised that some tunnel gullies were present prior to Cyclone Bola.

Sediment flux

Hillslopes Landslide ( 1° source areas) Sheet Tunnel Gully Valley floors Channel (2 ° source areas)

179

(volume, m 3)

(%)

1,200,000 96,000 25,000 28,000

89 7 2 2

1,349,000 Total Storage

Hillslopes Valley floors

Deposition Deposition

286,000 21 290,000 22 576,000 Total

Output

Lake bed Lake outlet

Sedimentation Stream/Channel transport

686,000 87,000

51 6

773,000 Total

Estimates of tunnel gully erosion have been made by applying results from a detailed storm damage assessment of a 161 ha study area at Otoi, 30 km north of Tutira (Harmsworth et al., 1987). This study area contained the same tunnel gully susceptible landforms as those at Tutira, and was subject to a similar storm rainfall (609 mm in three days). Landform units in the Tutira watershed affected by tunnel gully erosion were identified and the area calculated. Tunnel gully values from the Otoi study of 38 m2/ha area eroded, and 1.0 m average depth were then applied to calculate a volume eroded of 25,000 m 3.

7.4. Sheet erosion (Ss) No attempt was made to measure directly the contribution of sheet erosion to the budget. Unlike the other processes, sheet erosion does not produce erosion features which are readily measurable after a storm. In a study of post-deforestation soil loss from steep sandstone hillslopes in the Taranaki Region of New Zealand, DeRose et al. (1993) found that over an 85year period landsliding was the dominant erosion process, and that the contribution of other processes to soil loss is minimal. Lehre (1982) in a three-year study in a steep California drainage basin underlain by greywacke, estimated that hillslope sheet erosion occurs as a rainsplash-driven creep of a layer about 1.5 mm thick. Given the soft nature of the sediments at Tutira, together with the intensity and duration of the rainfall event, a figure of 5 mm is considered to reflect the upper limits of soil loss. Applying this figure to landforms with slopes > 10°, the estimated volume eroded was 96,000 m 3

7.5. Total sediment (S) From the values calculated for each of the four erosion processes, the total volume of sediment generated

Table 3 Comparison of denudation values for New Zealand storms Location

Source a

Storm date

Rainfall (mm)

Denudation value (mm)

Tutira Watershed Tutira (hillslope landforms) Tangoio, Hawke's Bay Pakaraka, Wairarapa Makahu, Taranaki Hapuakohe Range, Waikato Otoi, Hawke's Bay Tangoio, Hawke' s Bay Stokes Valley, Lower Hntt

[ 1] [ 1] [2 ] [3 ] [4] [5 ] [61 [2 ] [7 ]

1988 1988 1938 1977 b 1990 1973 1985 1971 1976

753 753 991 420 > 300 190 609 204 250-300

42 83 115 69 41 40 17 7.5 3

aSource: [1] Trustrum and Page (1992); [2l Eyles ( 1971 ); [ 3 ] Crozier et al. ( 1980 ); [ 4 ] DeRose et al. (1993); [ 5 ] Selby (1976); [ 61 Harmswort h et al. (1987); [7] McConchie (1980). bl0 weeks.

180

M.J. Page et al. / Geomorphology 9 (1994) 169-188

during Cyclone Bola (S) is estimated to be 1,349,000 m 3 (Table 2). This represents a surface lowering of 42 (__.4) mm for the total watershed. Table 3 provides a comparision with other New Zealand storms. 7.6. Colluvial and alluvial storage (H and V) Having calculated the total sediment generated during Cyclone Bola the next step was to calculate how much remained in storage on the landscape ( H + V) and how much entered the lakes (L). Because Lakes Tutira and Waikopiro act as a large sediment trap much of the sediment that would otherwise be removed from the watershed is deposited in the lakes. The sediment has an adverse effect, both in terms of the water quality, through nutrient inputs associated with the sediment, and lake infilling. The majority of the sediment deposited on the landscape was derived from landslides. Sediment was observed to be deposited both on hillslopes, in the form of landslide debris tails, and on valley floors. Although debris tails extended onto valley floors, most valley floor sediment was deposited by streams overtopping their banks. This sediment included material derived from all four erosion processes. The volumes of sediment deposited on hillslopes, H, and valley floors V, were estimated in the same manner as scar volume except that a 19 mm dot grid was used to select points for field sampling of sediment thickness. The volume for H was estimated to be 286,000 m 3 (standard error 5%), and the volume for V was 290,000 m 3 (standard error 7%). The volume of sediment that entered the lakes (L) was estimated to be 773,000 m 3 by deducting the sediment on hillslopes (H) and valley floors (V) from the total sediment generated (S) - - that is, L = S - H - V. Z 7. Lake sedimentation (L) The input and storage component of the sediment budget indicated that 773,000 m 3 of sediment entered Lakes Tutira and Waikopiro, representing an average thickness of 41 cm. In order to check this, sediment cores were taken from the lake beds during February 1992. Because of the variability in lake bed morphology and the entry points of sediment supplying streams, the lakes were stratified into three areas: a northern area where sedimentation rates are greatest, and within the

remainder, a shallow and steep area and a deep central basin area. Sampling was carried out on a 300 m × 300 m grid basis to provide 20 cores representing the various strata, from which the average thickness of Cyclone Bola sediment could be calculated. In order to prevent disturbance of the uppermost sediments, cores were obtained using a freeze-box sampler (Huttunen and Merilainen, 1978). Cores up to 0.5 m long and 2 to 4 cm thick were obtained. Previous cores taken with a Mackereth piston corer (Mackereth, 1958) showed that the sediment consists of slowly accumulating organic rich (gyttja-like) deposits alternating with layers of mineral-rich sediment, characterised by graded bedding, and representing the product of individual storm events. These "storm sediment pulses" have been correlated to a 93year storm history of the watershed (Page et al., in press). The Cyclone Bola sediment pulse was present in the 14 cores from the "shallow and steep" and "deep basin" areas. However, because sedimentation rates in the northern area are more rapid, either only the upper part of the Cyclone Bola sediment, or in some cases only post-Bola sediment was present. The 14 freezebox cores were therefore supplemented by eight cores taken several years earlier with a 3 m Mackereth corer in which the Cyclone Bola sediment was present. Four of these cores were from the northern area of the lake. The average thickness of Cyclone Bola sediment for the three strata within the lakes is as follows: shallow and steep area deep basin area northern area

6 cm 27 cm 98 cm

When these averages are applied over the area of each strata the average thickness of Cyclone Bola sediment in the lakes is 36 cm. This represents a volume of 686,000___ 170,000 m 3 The large uncertainty is due to the small number of samples. Therefore of the 773,000 m 3 of sediment estimated to have entered the lakes, 686,000 m 3 of sediment was found to have been deposited on the lake beds (L), representing 51% of the total sediment generated during Cyclone Bola. The difference of 87,000 m 3 is presumed to have been discharged from the lakes through the outlet (D). This represents a storage efficiency for the lakes of 89%, but due to the large uncertainty in the

M.J. Page et al. / Geomorphology 9 (1994) 169-188

volume of lake sediment this figure could range from 70% to 100%. The average bulk density of the Cyclone Bola sediment in the lakes ( 1.14 g / c m 3) was similar to that of uneroded material on the hillslopes (1.16 g / c m 3) (Eyles, 1971). The volume of Bola sediment in the lakes can therefore be directly compared with input volumes. In a New Zealand landslide-prone terrain with thicker tephra deposits and a more porous regolith, bulk density measurements of the fluid flow component of landslide deposits have been found to increase by up to 30% (M.J. Crozier, pers. commun., 1992). At Tutira bulk density increases can be expected to be less. Adopting an increase in bulk density of 20% would have the effect on the sediment budget (see Table 2) of increasing the amount of sediment stored on hillslopes by 3% and correspondingly decreasing the amount discharged from the lakes by 3%. In summary, of the 1,349,000 m 3 of sediment generated during the storm, 576,000 m 3 or 43% remained in storage on the landscape and 773,000 m 3 or 57% entered watercourses to be deposited in the lakes or discharged from the watershed via the lake outlet. The sediment volumes attributed to the various processes are summarized in Table 2.

Table 4 Landform types in the Tutira Watershed Process

Type

Stable

Plateau Ridge Crest Downland Relict Block Slide Debris Field Dip Slope ( < 10°)

Denudation

Accumulation

8. Landform analysis Calculating the total amount of sediment generated during Cyclone Bola within the Tutira watershed provides important information on regional denudation rates and sediment supply to rivers. It can also be correlated with the storm record preserved in lake sediments. However, it does not take into account the variability of land within the watershed and, therefore, the information is of limited value to land users and managers confronted with the issue of sustainable land use. In order to focus on these management issues a landform map of the watershed was constructed, consisting of 441 map units grouped into 24 landform types (Table 4). Landforms were identified by stereoscopic analysis of 1:15,000 scale colour infra-red aerial photographs ( taken four weeks after Cycl one B ola). Landform boundaries were then transferred to 1:5000 scale enlargements for checking in the field. Based on the

181

Total

Area % (ha) Watershed 30 85 369 32 29 352 897

1 3 11 1 l 11 28

Dip Slope ( > 10°) Scarp Slope (25-30 °) Scarp Slope ( > 30 °) Terrace Riser Bench Scarp Strongly Rolling Hillslope Moderately Steep Hillslope Steep Hillslope Escarpment Gorge Bluff

310 208 104 49 23 536 282 188 109 61 9 1879

10 6 3 1 1 17 9 6 3 2 < 1 58

Colluvial Footslope Remnant Alluvial Terrace Basin Alluvial Plain Swamp Fan Bench

43 21 138 146 53 2 29 432 3208

1 1 4 5 2 < 1 1 14 100

geomorphic processes operating, landforms were grouped into one of the following categories: stable, denudation, accumulation. The erosion response of each landform type was then assessed and its contribution to the total budget was calculated. The sampling scheme described earlier was designed to estimate sediment volumes to an accuracy of ___10% over the whole watershed. For individual landform types the standard errors are much higher and, therefore, an alternative method was used. Rather than estimate the area of landsliding within landform types using a dot grid, the relative area of landsliding was measured. Landform boundaries were transferred to a 1:50,000 scale rectified panchromatic SPOT image of the watershed, taken in January 1990 (Photo 5), and

182

M.J. Page et al. /Geomorphology 9 (1994) 169-188

Photo 5. January 1990 SPOT image of the Tutira watershed used to classifylandslide scars. then digitised using ARC INFO software. Each map unit was allocated to one of the 24 different landform types. Previous experience with SPOT imagery had shown that it is not possible to identify landslide scars and debris tails separately. For this reason a January 1990 image was chosen, where only scars were present, debris tails having revegetated in the intervening two years since Cyclone Bola.

A digital map of landslide scars was created by tagging all areas brighter than a given threshold. There is uncertainty in determining this threshold and thus total landslide areas may be inaccurate. However, the landslide areas within a landform expressed as a fraction of the total watershed landslide area (i.e. the relative area) is a more stable measurement. The relative area of landsliding for a landform type is multiplied by its

M.J. Page et al. /Geomorphology9 (1994) 169-188

183

TUTIRA WATERSHED EROSIONAL LANDFORMS

North

500m

0

Ill

T LEGEND Process

Landform Type

Erosional

Percent Sedimentfrom landslides

Sleep hills

33

Mod. sluep hills

22

m

Escarpments

m

Mod. steep scarps 13 Sleep scarps

5

Rolling hills

9

Others Depositional

8

10

NI NI

I I

lake

Fig. 3. Major erosional landforms in the Tutira watershed.

1 I

I

2 Ion I

I

184

M.J. Page et al. /Geomorphology 9 (1994) 169-188

Table 5 Percentagecontributionof landformtypes to sedimentgeneratedby landslides Landformtype

% contribution

Escarpment Moderately steep hillslope Steep hillslope Strongly rolling hillslope Scarp slope (25-30°) Scarp slope ( >30 °) Others

8 ( + 1) 22 ( + 4) 33 (_+4) 9 ( _+2) 13 ( _+7) 5 (_+2) 10

of sediment generated. Results summarised in Table 5 show that six landforms occupying 44% of the watershed contributed 90% of the sediment. This represents a surface lowering of 83 ( __+8) mm for the area of these landforms. Two of these landforms occupying only 15 % of the watershed contributed 55% of the sediment. A simplified landform map, showing the major erosional landforms is shown in Fig. 3.

9. Discussion 9.1. Erosion processes

Landsliding was the dominant erosion process during Cyclone Bola, accounting for 89% of the sediment generated. Hillslope morphology and historical aerial photography indicate that landslide erosion is the dominant hillslope-forming process in this terrain and that landslide events have short recurrence intervals. DeRose et al. (1993) also found landsliding to be the major erosion process in a similar New Zealand hill country environment over a period of 85 years since deforestation and also during a two-day high-intensity rainstorm. Other New Zealand studies where landsliding was the major erosion process initiated by high intensity or long duration rainfall are reviewed by Harmsworth and Page (1991) and include Pain (1971), Eyles (1971), Selby (1976), McConchie (1980), Crozier et al. (1980), Eyles and Eyles ( 1981 ), O'Loughlin et al. (1982), Harmsworth et al. (1987), and Trustrum et al. (1990). A number of studies in steepland terrain in other countries report similar findings. Lehre (1982), in a sediment budget of a steep drainage basin in north-central California, showed that in years with extreme rainfall and flow events, debris

slides and flows are responsible for over 80% of all sediment generated. Although only 2% of sediment was derived from channel erosion, this involved significant channel modification. Although alluvial valley floors are net sediment storage sites (290,000 m 3 gain versus 28,000 m 3 loss), there has been an increase in sediment remobilisation through channel erosion in the last 50 years. If the deepening and widening of channels continues an increasing percentage of sediment will be carried directly into the lakes. Observations four years after Cyclone Bola indicate that small storm events are continuing to enlarge channels. In constructing a sediment budget some processes can be quantified much more accurately than others. Kelsey (1981) points out that processes that are too complex to measure provide the unknowns into which the errors from other quantities are lumped in order to balance the budget. He also states that the least understood process should not be one of the crucial items for basin management. In this budget the most poorly quantified process is sheet erosion. However, the similarity of results regarding landslide contribution between the Tutira budget and other studies, and the similarity between the predicted and actual sediment in the lakes, indicate that sheet erosion is only a minor contributor to the budget and the uncertainty of the actual contribution is therefore not considered to be significant. 9.2. Sediment storage

Of the sediment generated during Cyclone Bola 57% entered watercourses to be deposited in the lakes or discharged through the lake outlet, with 43% remaining in storage on the landscape. Two nearby studies show similar results. Harmsworth et al. (1987), in a storm damage assessment of an area 30 km north of Tutira with a similar terrain and similar storm rainfall, calculated that between 50 and 55% of the sediment generated entered streams. For the same storm R.D. Black (pers. commun., 1990) applied these results to a similar nearby watershed to calculate the amount of sediment generated and the amount entering a small hydro lake. His estimate of 300,000 m 3 entering the hydro lake compares well with the figure of 288,000 m 3 derived by depth sounding.

M.J. Page et aL / Geomorphology 9 (1994) 169-188

Even for different terrains and triggering conditions similar values have been obtained. Lehre (1982), in a steep greywacke drainage basin, found that over a three-year period 53% of all sediment generated was discharged from the basin. Pain and Bowler (1973), following a large earthquake in Papua New Guinea, assessed the erosion effects on a steep forested greywacke mountain range. They state that about half of the material eroded entered the sea in the first six months after the earthquake. 9.3. Residence times

Although the scope of this budget has precluded quantification of the residence time of sediment and the rate of transport processes (Dietrich and Dunne, 1978), some general statements can be made. The presence of a number of tephras provides a method of dating land surfaces and identifying the relative stability of landforms. Hollows, basins and footslopes are the sites of hillslope sediment accumulation. Tephras and soils buried beneath these colluvial deposits indicate that these are long-term storage sites from which residence time of sediment can be calculated (Reneau et al., 1989). Deposition of this colluvium is largely episodic, consisting mainly of landslide debris. A minimum residence time of 3280 yr B.P. is indicated by Waimihia Tephra preserved in the excavated hollow shown in Photo 2. Valley floors are also major storage sites and here a more comprehensive chronology of sediment accumulation is preserved. This consists of a number of buried soils including the pre-European soil and, in the base of streams, the pre-Polynesian forest soil. The pre-Polynesian soil is approximately 1.75 m below the present surface and in situ remains of the forest have been dated at 660 yr B.P. 9.4. Lake sedimentation

The lakes act as highly efficient sediment traps. However, coring has shown that sediment accumulation within Lake Tutira is not uniform. This is not surprising given the irregular lake bed morphology (Irwin, 1978). With such a configuration little mixing of sediment within the lake occurs and this is confirmed in the shallower, northern end of the lake where major sediment entry points e×;~t. Here the thickness of Cyclone Bola sediment is approximately four times

185

greater than in the basins and 16 times greater than in the steep and shallow areas. The lake previously extended beyond its present northern margin into an area now occupied by a swamp. With the present rate and pattern of sedimentation the lake will continue to reduce in area as it infills from the north. 9.5. Geomorphic effectiveness

Cyclone Bola was a geomorphically effective event. The 1,349,000 m 3 of sediment generated is equivalent to a surface lowering of 42 mm for the entire watershed and 83 mm for the major erosional landforms. The figure of 83 mm is exceeded only by the 1938 storm when compared with denudation values for other documented New Zealand storms (Table 3). Such extreme events also account for most of the geomorphic work done in the Tutira watershed. Analysis of lake cores show that the amount of sediment generated by such events is disproportionately large and that little sediment accumulates between such events (Trustrum and Page 1992). This finding is consistent with the conclusions of a number of other studies reviewed by Pearce (1986). 9. 6. Magnitude-frequency-effectiveness relationships

In order to put Cyclone Bola into context as a geomorphic event an understanding of magnitude-frequency-effectiveness relationships is necessary. Cores taken from the lake beds, when correlated with a 93year rainfall record, have identified the magnitude and frequency of erosion producing storms (Page et al., in press). Although Cyclone Bola, with 753 mm, was the largest rainfall event in the 93-year record, the amount of sediment it deposited in the lakes was only 56% of that produced by the 1938 storm. Rainfall conditions do not account for the disparity in sediment produced by the two storms. Rainfall for the 1938 storm was 692 mm, also over a four-day period, while antecedent rainfall was similar. It has been suggested that an earthquake in 1931 ( magnitude 7.75 on the Richter scale ) predisposed hillsides to future landsliding and so contributed to the erosion caused by the 1938 storm. Guthrie-Smith (1953) who was at Tutira at the time of the earthquake stated that landslides were common although far less so than during a heavy rainstorm, and that numerous

186

M.J. Page et al. /Geomorphology 9 (1994) 169-188

cracks and fissures were formed. However, two rain storms in excess of 400 mm and one in excess of 300 mm which occurred between 1931 and 1938 could have been expected to trigger some of these incipient landslides. It seems likely, therefore, that although the earthquake may have contributed to the amount of sediment produced by the 1938 storm, it would not account for the large difference between the 1938 storm and Cyclone Bola amounts. The difference between the amounts of sediment produced by the 1938 storm and Cyclone Bola indicates that the magnitude of storm rainfall events and sediment yields are not simply related. Such magnitudefrequency-effectiveness relationships are complex (Beven, 1981) because the response of the landscape to any event may be very greatly affected by prior events, and the rate and form of recovery from that event may be affected by both prior and subsequent events (Pearce, 1986). At Tutira the 1938 storm was significantly larger than any previous rainfall event since the conversion of the watershed to pasture which began with European settlement. As such it removed a large proportion of the sediment that had been stored on the landscape under the previous forest and scrub cover. The cumulative effect of the 1938 and prior and subsequent storms was to reduce the amount of sediment available for mobilisation during Cyclone Bola. This progressive regolith stripping has also preferentially removed material from the most susceptible sites so that, even allowing for weathering rates, a greater percentage of remaining regolith is in more resistant sites. The result is a rise in the triggering threshold for landsliding episodes, often in the form of more extreme storm events. This has been described as "event-resistance" by Crozier (1986), and has also been reported for hillslopes in the Taranaki Region, New Zealand (DeRose et al., 1993). Numerous workers have referred to the difficulty in establishing magnitude-frequency-effectiveness relationships. Initial interpretations of lake cores indicate that analysis of the storm sediment record can contribute to an understanding of these relationships in steep landslide-prone terrain. 9. 7. L a n d use implication

In New Zealand cyclone-induced erosion is regarded as the main limitation to sustainable land use of much

soft-rock hill country. Within the Tutira watershed Cyclone Bola had major impacts on both the land and water resources. This sediment budget was designed to assess these impacts and to help resolve the issue of sustainable land use. To increase the usefulness of this budget for land management purposes, both the sediment source and storage areas were identified in terms of landforms which are readily recognisable by land managers. Results show that sediment generation is highly variable and dependent on landform type. Such landform analysis, together with studies on the effects of landslide erosion on productivity (Trustrum et al., 1990; Trustrum and Blaschke, 1992), allows management actions to be targeted. The integration of the results of this budget with the storm magnitude-frequency history preserved in lake sediments can be used to develop watershed-based models to predict the erosion response to various climate and storm scenarios. This will assist rational evaluation of environmental and economic risks of different land management strategies.

Acknowledgements The following people and organisations are acknowledged for their assistance. Ron DeRose, Landcare Research New Zealand, who assisted in the collection of field data and provided valuable discussion on hillslope processes; Ted Pinkney, Landcare Research New Zealand, who surveyed and supervised the coring programme; Janet Wilmshurst, University of Canterbury, who provided the freeze-box corer and assisted with lake sampling. The MacKereth cores were obtained as part of a joint coring programme with Gillian Turner and Eric Broughton, Victoria University of Wellington. Constructive comments on the text were made by Leslie Reid, USDA-Forest Service, Mike Crozier, Victoria University of Wellington, Doug Hicks and Paul Biaschke, Landcare Research New Zealand. We would also like to thank Robin Black, Hawke's Bay Regional Council, for his support of the study and to the Council for providing equipment and funding assistance; Department of Conservation for permission to take lake bed samples; Don Graham, manager Tutira Station, for access to the study site; and Paul Jennens,

M.J. Page et al. /Geomorphology 9 (1994) 169-188 G u t h r i e - S m i t h O u t d o o r E d u c a t i o n C e n t r e , for use o f a c c o m m o d a t i o n facilities.

Appendix. Variance of S V a r [ S ] = V a r [~-z - as ] = (~-~) 2 V a r [ a s ] + a2Var[-A--~]

(assuming errors are small and independent) = (-~)2A2Var[n]

+a2Var[-~]

w h e r e A is the area on the g r o u n d r e p r e s e n t e d by e a c h dot in the d o t grid and n is the n u m b e r o f t i m e s a dot lands on a l a n d s l i d e scar. n is a b i n o m i a l distribution w i t h v a r i a n c e n, therefore: Var[S] = (-~)2AZn +aZVar[ AZ]/n T h e r a n d o m e r r o r a s s o c i a t e d with e a c h d e p t h m e a s u r e m e n t in the field is m u c h s m a l l e r than the s t a n d a r d d e v i a t i o n o f l a n d s l i d e scar d e p t h and t h e r e f o r e can be ignored.

References Adams, J., 1981. Earthquake-dammed lakes in New Zealand. Geology, 9: 215-219. Beven, K.J., 1981. The effect of ordering on the geomorphic effectiveness of hydrologic events. Int. Assoc. Hydrol. Sci. Publ. 132: 510-526. Blaschke, P.M., Trustrum, N.A. and DeRose, R.C., 1992. Ecosystem processes and sustainable land use in New Zealand steeplands. Agric, Ecosyst. Environ., 41: 153-178. Crozier, M.J., 1986. Landslides: Causes, Consequences and Environment. Croom Helm, London, 252 pp. Crozier, M.J., Eyles, R.J., Marx, S.L., McConchie, J.A. and Owen, R.C., 1980. Distribution of landslips in the Wairarapa hill country. N. Z. J. Geol. Geophys., 23: 575-586. DeRose, R.C., Trustrum, N.A. and Blaschke, P.M., 1991. Geomorphic changes implied by regolith-slope relationships on steepland hillslopes, Taranaki, New Zealand. In: M.J. Crozier (Editor), Geomorphology in Unstable Regions. Catena, 18 (5): 489-514. DeRose, R.C., Trustrum, N.A. and Blaschke, P.M.,, 1993. Postdeforestation soil loss from steepland hillslopes in Taranaki, New Zealand. Earth Surf. Process. Landforms, 18 ( 2 ) : 131-144. Dietrich, W.E. and Dunne, T., 1978. Sediment budget for a small catchment in mountainous terrain. Z. Geomorphol. N.F., 29: 194-206. Dietrich, W.E., Dunne, T., Humphrey, N.F. and Reid, LM., 1982. Construction of sediment budgets for drainage basins. In: F.J.

187

Swanson, R.J. Janda, T. Dunne and D.N. Swanston (Editors), Sediment Budgets and Routing in Forested Drainage Basins. USDA Forest Service, Pacific Northwest Forest and Range Experiment Station, Gen. Tech. Rep., PNW-141, pp. 5-23. Dyrness, C.T., 1967. Mass soil movements in the H.J. Andrews Experimental Forest. USDA For. Serv. Res. Pap. PNW-42, 12 PP. Eden, D.N., Froggatt, P.C., Trustrum, N.A. and Page, M.J., in prep. A multiple source Holocene tephra sequence from Lake Tutira, Hawke's Bay, New Zealand. N.Z.J. Geol. Geophys., 36: 233242. Eyles, R.J., 1971. Mass movement in Tangoio Conservation Reserve Northern Hawke's Bay. Earth Sci. J., 5(2): 79-91. Eyles, R.J. and Eyles, G.O., 1981. Recognition of storm damage events. In: Proc, Eleventh New Zealand Geography Conf., Wellington 1981. New Zealand Geological Society, Wellington Branch, pp. 118-123. Guthrie-Smith, W.H., 1953. Tutira, the Story of a New Zealand Sheep Station. 3rd ed. William Blackwood, London. Harmsworth, G.R. and Page, M.J., 1991. A review of selected storm damage assessments in New Zealand. DS1R Land Resources. Sci. Rep. No. 9, 34 pp. Harmsworth, G.R., Hope, G.D., Page, M.J. and Manson, P.A., 1987. An assessment of storm damage at Otoi in Northern Hawke's Bay. Soil Conservation Centre, Aokautere Publ., 10. Huttunen, P. and Meril~iinen,J., 1978. New freezing device providing large, unmixed sediment samples from lakes. Ann. Bot. Fenn., 15: 128-130. Irwin, J., 1978. Lakes Tutira: Waikopiro: Orakai Bathymetry 1:5000 New Zealand Oceanographic Institute Chart. Lake series. Kelsey, H.M., 1981. Sediment budgets as a vehicle for understanding sediment transport and aiding in land managment. J. Hydrol. (New Zealand), 20( 1): 102-104. Lehre, A.K., 1982. Sediment budget of a small coast range drainage basin in North-Central California. In: F.J. Swanson, R.J. Janda, T. Dunne and D.N. Swanston (Editors), Sediment Budgets and Routing in Forested Drainage Basins. USDA Forest Service, Pacific Northwest Forest and Range Experiment Station, Gen. Tech. Rep. PNW-141, pp. 67-77. Lowe, D.A., 1987. The geology and landslides of the Lake TutiraWaikoan area, northern Central Hawke's Bay. M.Sc. Thesis, Victoria Univ. of Wellington, unpubl. Mackereth, F.J.H., 1958. A portable core sampler for lake deposits. Limnol. Oceanogr., 3: 181-191. McConchie, J.A., 1980. Implications of landslide activity for urban drainage. J. Hydrol. (New Zealand), 19( 1): 27-34. O'Loughlin, C.L., Rowe, L.K. and Pearce, A.J., 1982. Exceptional storm influences on slope erosion and sediment yield in small forest catchments, North Westland, New Zealand. In: E.M. O'Loughlin and E.J. Bren (Editors), Institute of Engineers, Australian National Conf. Publ., 82/6, pp. 84-91. Page, M.J., Trustrum, N.A. and DeRose, R.C., in press. A high resolution record of storm-induced erosion from lake sediments, New Zealand, J. Paleolimnol. Pain, C.F., 1971. Rapid mass movement under forest and grass in the Hunua Ranges, New Zealand. Aust. Geogr. Stud., 9: 77-84.

188

M.J. Page et al. / Geomorphology 9 (1994) 169-188

Pain, C.F. and Bowler, J.M., 1973. Denudation following the November 1970 earthquake at Madang, Papua New Guinea. Z. Geomorphol. N.F., 18: 92-104. Pearce, A.J., 1986. Geomorphic effectiveness of erosion and sedimentation events. J. Water Resour., 5( 1 ): 551-569. Phillips, C.J., 1988. Geomorphic effects of two storms on the upper Waitahaia river catchment, Raukumara Peninsula, New Zealand. J. Hydrol. (New Zealand), 27(2): 99-112. Reid, L.M., 1990. Construction of sediment budgets to assess erosion in Shinyanga Region, Tanzania. In: R.R. Ziemer, C.L., O'Loughlin, L.S. Hamilton (Editors), Research Needs and Applications to Reduce Erosion and Sedimentation in Tropical Steeplands. Proc. Fiji Syrup., June 1990, IAHS-AISH Publ. No. 192, pp. 105-I 14. Reneau, S.L., Dietrich, W.E., Rubin, M., Donahue, D.J. and Jull, A.J.T., 1989. Analysis of hillslope erosion rates using dated colluvial deposits. J. Geol,, 97: 45~53. Renwick, W.H., 1977. Erosion caused by intense rainfall in a small catchment in New York state. Geology, 5: 361-364. Selby, M.J., 1976. Slope erosion due to extreme rainfall: a case study from New Zealand. Geogr. Ann., 58A: 131-138. Sp~rli, K.B., 1980. New Zealand and oblique-slip margins: Tectonic development up to and during the Cenozoic. In: P.F. Ballance and H.G. Reading (Editors), Sedimentation in Oblique Slip

Mobile Zones. Int. Assoc. of Sedimentologists, Spec. Publ,, 4, pp. 147-170. Sutherland, R.A. and Bryan, R.B., 1991. Sediment budgeting: a case study in the Katiorin drainage basin, Kenya. Earth Surf. Process. Landforms, 16: 383-398. Trotter, C.M., Stephens, P.R., Trustrum, N.A., Page, M.J., Carr, K.S. and DeRose, R.C., 1989. Application of remotely sensed and geographic information system data to quantitative assessment of landslide damage. In: IGARSS '89 12th Canadian Symp. on Remote Sensing. Remote Sensing: An Economic Tool for the Nineties, July 10-14, 1989, Vancouver, Canada. Trustrum, N.A. and Blaschke, P.M., 1992. Erosion and land productivity. In: Proc. 44th Ruakura Farmer's Conf., 9-10 June 1992, Ruakura~ New Zealand. Trustrum, N.A. and Page, M.J., 1992. The long-term erosion history of Lake Tutira watershed: implications for sustainable land use management. In: Proc. Int. Conf. on Sustainable Land Management, November 17-23, 1991, Napier, New Zealand. Trustrum, N.A., Blaschke, P.M., DeRose, R.C. and West, A., 1990. Regolith changes and pastoral productivity declines following deforestation in steeplands of North Island, New Zealand. Trans. 14th Int. Congr. Soil Sci. Kyoto, Japan, 1: 125-130. Wolman, M.G. and Gerson, R., 1978. Relative scales of time and effectiveness of climate in watershed geomorphology. Earth Surf. Process., 3: 189-208.