EXECUTIVE SUMMARY

This report, triggers of extreme erosion-sedimentation events on hillslopes in the Nattai catchment, follows on from Technical Report 1 provided to the Catchment Authority (SCA) in early 2005. In the first report, we established a number of important findings, namely:

• The hillslopes in the incised valleys of the Nattai catchment (and other valleys incised within the Blue Mountains Plateau) clearly differ from the typical hillslopes formed wholly on Hawkesbury Sandstone such as the Hornsby Plateau to the north and the Woronora Plateau to the south. • The Nattai hillslopes contain extensive gravelly colluvial mantles reflecting debris flows, rock falls and landslides i.e. mass movement or extreme erosion-sedimentation events; • Event based movement of large volumes of sediment on hillslopes within the SCA Special Areas, particularly those slopes adjacent to the foreshores could pose a serious threat to water quality and quantity within the major water supply reservoirs; • The triggers of extreme events on hillslopes are likely to be a combination (coupling) of wildfires, severe storms, catastrophic floods, earthquakes and breach of intrinsic (geomorphic) thresholds; • However, there is a need to analyse these triggers in more detail to quantify their role and relative importance in sediment mobilisation and transport, to determine relationships or feedbacks between triggers such as the fire-rainfall-runoff relationship, as well as the return periods of events of different magnitudes in order to evaluate the risks to water quality and quantity.

The main aims of this technical report are (i) to provide more detail on the morphology and dominant erosional processes on hillslopes in the incised valleys of the Nattai catchment including the outcomes from an investigation into the landslide below Tumbledown Mountain located in the lower Nattai valley; and (ii) to assess each erosional trigger in terms of the role, importance, feedbacks and return intervals, placing particular emphasis on wildfires and data collected after the 2001-02 fires which burnt the largest area of the Nattai catchment since the construction of Warragamba in 1960. Our goal is to determine the most significant trigger(s) of extreme erosion-sedimentation events in the SCA Special Areas.

The sites used to conduct field work for this investigation are Blue Gum Creek, a tributary of Little which in turn flows into the and, Tumbledown Landslide which is located in the lower Nattai catchment adjacent to the section of the Nattai River that is flooded by . Our methods include a detailed reconnaissance investigation of the Nattai catchment using maps, aerial photographs and site visits to gain an understanding of catchment topography and morphology, then selection of field sites based on access and suitability for investigation to achieve the above aims. The Blue Gum Creek sites were used to collect data on post-fire erosion and characterise hillslope deposits. The Tumbledown Landslide was used as a case study to provide information on the size and volume of rock and failure mechanism which could lead to landsliding in the Nattai catchment. Longer records provided by the SCA on rainfall, stream flow, river suspended sediment, turbidity and fire history were used to identify significant events i.e. catastrophic floods, severe wildfires, and analyse frequency-magnitude relationships. Additional rainfall records obtained from the Australian Bureau of Meteorology were used to supplement the SCA data. Seismic records obtained from Geoscience were used to analyse the spatial distribution of earthquakes compared with known areas of landslides in the SCA Special Areas.

The hillslopes in the incised valleys of the Nattai catchment are characterized by deposits formed through extreme erosion-sedimentation events. These include thick gravelly colluvial mantles which extend from the upper to lower slopes and occasionally infill the valley floor. In some locations the gravelly material infill depressions in bedrock or previous drainage lines with little or no surface expression. Where the have incised through the upper sequence and into the lower Permian strata, significant valley widening takes place and lobes or mounds of gravels can often be seen preserved on the slopes resulting from landslides. The dominant

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment ii erosional processes on hillslopes in the Nattai catchment include rainsplash, slope wash, debris flows, rock fall, rock avalanching and rotational slumping. Rainsplash and slope wash occurs widely across hillslopes resulting in erosion of surface material. Debris flows form in tributary streams and drainage lines on hillslopes where concentrated runoff mobilizes coarser material including boulders, transporting it downslope and into the major rivers. Rock fall and rock avalanches result from collapse of the steep or vertical cliff faces formed in the Hawkesbury and Narrabeen Group sandstones, whilst rotational slumping in bedrock appears to occur in the valleys incised into the weaker Permian rocks.

The triggers of extreme erosion are wildfire, extreme rainfall, catastrophic floods, earthquakes and intrinsic thresholds, although three scenarios can result: (i) extreme erosion results from a single trigger, (ii) extreme erosion results from multiple triggers, or (iii) extreme erosion results from the coupling of triggers. Wildfire is often thought of as a major trigger of extreme erosion- sedimentation events in southeastern Australia. The evidence from the Nattai catchment, however, suggests that the nature and extent of erosion following wildfire is determined by the characteristics of rainfall in the post-fire period. Heavy rainfall events can trigger significant rainsplash erosion and transport of coarse material, but these may or may not occur within the critical period depending on weather patterns which bear little relationship to wildfire. Extreme rainfall events including heavy rainfall and severe thunderstorms are important triggers of extreme erosion particularly if the former is coupled with catastrophic floods which can export the equivalent of decades of sediment load from catchments within a few days. High magnitude earthquakes (> 6) are a likely trigger of large landslides. Although earthquakes occur at much longer return intervals compared with the other triggers, the impacts resulting from a landslide event along the foreshores of Lake Burragorang will be substantially greater.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment iii Table of Contents

EXECUTIVE SUMMARY...... II

1. INTRODUCTION ...... 1

2. NATTAI CATCHMENT ...... 3

3. METHODS, FIELD SITES AND SOURCES OF DATA...... 4

3.1. FIELD INVESTIGATIONS ...... 4 3.2. EXISTING RECORDS...... 6

4. CHARACTERISTICS AND DOMINANT EROSIONAL PROCESSES ON HILLLOPES IN THE NATTAI CATCHMENT...... 9

4.1. BLUE GUM CREEK...... 9 4.2. TUMBLEDOWN LANDSLIDE ...... 14

5. TRIGGERS OF EXTREME EROSION EVENTS ...... 19

5.1. WILDFIRES...... 19 5.2. EXTREME RAINFALL...... 21 5.3. CATASTROPHIC FLOODS ...... 23 5.4. EARTHQUAKES...... 26 5.5. INTRINSIC THRESHOLDS...... 28

6. DISCUSSION ...... 30

6.1. ROLE OF WILDFIRES COMPARED WITH OTHER TRIGGERS...... 30 6.2. COUPLING OF TRIGGERS LEADING TO EXTREME EROSION-SEDIMENTATION EVENTS ...... 31

7. CONCLUSIONS ...... 33

8. ACKNOWLEDGEMENTS ...... 34

9. REFERENCES ...... 34

Cover: Nattai River looking downstream, with Tumbledown Mountain and the Tumbledown Landslide scar seen on the cliffs in the background.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment iv

1. INTRODUCTION

Hillslopes characterise most of the land surface, playing an important role in sediment supply and transport. As a result, contemporary hillslope processes have been well studied. In Australia, and more specifically southeastern Australia, hillslopes are thought to be dominated by the processes of weathering, rainsplash and slope wash leading to gradual erosion and lowering of the landscape (e.g. Walker 1964; Williams 1972; Williams 1978). Under conditions which result in exposure of the soil surface such as following severe wildfire (Figure 1) the resistance to erosion is decreased, which enhances the effectiveness of mechanical processes and facilitates potentially higher rates of erosion and sediment mobilization into stream networks (Prosser and Williams 1998). Hence wildfires, along with severe storms, and floods are considered to be the major triggers of extreme erosion events in southeastern Australia. The 2003 fires provide a good example where increased sediment and nutrient loads from tributary catchments after the fires significantly affected the quality of water supplied to reservoirs downstream (Environment ACT 2004). Despite this, very little investigation has been undertaken to quantify the role these erosional triggers i.e. determine if one is more important than another, or to establish the relationships (coupling) and feedbacks between triggers such as the fire- rainfall-runoff relationship, or to evaluate the return intervals of triggers and hence determine how often we might expect extreme erosion events to occur. This information, however, is critical for effective management of water supply catchment areas where the provision of clean, adequate water is foremost. This report focuses on the triggers of extreme erosion-sedimentation events in the Nattai catchment which drains into Lake Burragorang, Sydney’s principal water supply reservoir (Figure 2). Our aim is to investigate the role, relationships and return intervals of triggers to inform the managing agency, the Sydney Catchment Authority (SCA) who commissioned this research.

(a) (b)

Figure 1(a) Typical conditions immediately following wildfire including exposure of the soil surface and (b) erosion and transport of sediment by the processes of rainsplash and slope wash with sediment trapped in litter .

Other key hillslope processes which result in extreme erosion-sedimentation events are mass movement or landslides as they are more commonly termed. Examples include debris flows, rock falls and rock avalanches (Table 1). Mass movement is often reported from areas where mass wasting and erosion is dominant over weathering and soil formation. Usually these landscapes have at least one of the following characteristics: active tectonic uplift, high relief, intense rainfall (typhoons/cyclones), and/or recent episodes of glaciation. Examples with one or more of these traits include the European Alps, New Zealand, Taiwan and the Himalayas.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 1

Figure 2. Location of the Nattai catchment showing the Blue Gum Creek (BGC) field sites (red circles), Tumbledown Landslide (yellow circle), Nattai and flow gauges (blue triangles) and rainfall gauges (green squares) used in the study.

Table 1. Classification and characteristics of some of the major types of mass movement (Source: Summerfield 1991) MECHANISM Slide Fall Flow Type of mass Translational Rotational Rock fall Debris flow Debris (rock) movement slide slump avalanche Material Unfractured Rock and/or Jointed rock Mix of rock, Rock rock soil sand and mud Moisture Low to Low to Low High Low content moderate moderate Rate of Very slow to Extremely Extremely Very rapid Extremely rapid movement extremely slow to rapid rapid moderate

Diagram

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 2

Mass movement in southeastern Australia is generally not expected given the stable, low relief terrain and temperate climate. A few small-medium contemporary mass movement events including small rock falls and cliff failures have been reported along the Illawarra Escarpment near Wollongong and in the Blue Mountains Plateau near Katoomba and Nattai, but most are related to underground coal mining and ground subsidence close to the escarpment/cliff edge (Pells et al. 1987; Cunningham 1988). Preliminary investigations in the Nattai catchment, however, revealed a different situation, with hillslopes widely blanketed by mass movement deposits (Tomkins et al. 2004b). A brief description was provided in Technical Report 1, including preliminary radiocarbon ages which indicated that deposits along Blue Gum Creek, a tributary flowing from the east of the Nattai catchment (Figure 2), were Late-Pleistocene to Holocene in age (< 15,000 years BP). The importance of these deposits is three-fold. Firstly, they contain significant stores of sediment which if remobilised and transported to the stream network could have major effects on water quality and quantity in the reservoirs. Secondly, they provide an indication of the magnitude, distribution and consequences of a mass movement event on hillslopes in the Nattai catchment. Thirdly, they suggest that large-scale mass movement could occur under present day conditions since climate conditions are similar to that throughout the last 10,000 years. In response, we extended our investigations to consider mass movement events in the Nattai catchment and to determine the triggers or conditions which may facilitate them. This also involved examining the seismic record since earthquakes are a known trigger of mass movement in Europe, USA and Asia (Keefer 1984; Keefer 1994; Keefer 2002).

In summary, the following research questions are addressed in this report:

1. What are the characteristics and dominant erosional processes on hillslopes in the incised valleys of the Nattai catchment? What are the key issues in terms of sediment supply and threats to water quality and quantity?

2. How important are wildfires in triggering extreme erosion and can we quantify the role of wildfires compared with other triggers including catastrophic floods, extreme rainfall events, earthquakes and intrinsic thresholds1?

3. What are the relationships (coupling) between different triggers and hence how does this affect the size and frequency of events?

The report is structured as follows. Firstly, a description of the Nattai catchment, methods, field sites and sources of data is provided. The morphology and dominant erosional processes in the Nattai catchment are then described. An analysis and discussion of each erosional trigger follows then final conclusions and implications for the SCA.

2. NATTAI CATCHMENT

The Nattai River catchment (701 km2) drains the southern end of the Blue Mountains Plateau, a sandstone dominated tableland on the western margin of the Sydney Basin. There are three primary geological units in the catchment (Rose 1966; Ray 2003). The uppermost is the heavily jointed, medium-coarse grained Triassic Hawkesbury Sandstone which is exposed on the plateau, ridge tops and upper valley slopes. Underlying the Hawkesbury unit are the interbedded sandstones, siltstones and claystones of the Triassic Narrabeen Group which form the lower slopes and valley floor of some tributaries including Blue Gum Creek. In the main Nattai valley and lower sections of the Little River and Blue Gum Creek the streams have incised into the shale and siltstone dominated rocks of the Permian Illawarra Coal Measures. In these parts the valley widens significantly.

1 In this report was not considered due to the difficulty in defining the temporal and spatial boundaries of drought affect land within the Special Areas. The resulting low runoff is a major issue for the SCA and is likely to have been addressed in other SCA commissioned projects.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 3

Topography exerts a strong influence on soils and vegetation. Soil depth on the plateau is mostly less than 1 m with widespread rock outcrop (King 1994; Henderson 2002; Wilkinson et al. 2005). In the valleys, the upper slopes are characterized by vertical cliffs, the mid- to lower-slopes are mantled by gravelly colluvium whilst thicker deposits (> 3 m depth) occur on the valley floors. Vegetation is dominated by fire-adapted open eucalypt forest which grades into tall open-forest in moist sheltered valleys and woodland on drier west-facing slopes (Fisher et al. 1995). Owing to the characteristics of the terrain, only small areas on the plateau have been cleared for agriculture and urban settlement. The majority (around 70 %) of the Nattai catchment is forested crown land protected through State Forest, National Park estate and the Greater Blue Mountains World Heritage Area.

The present-day climate of the Sydney region is humid to sub-humid warm temperate. Mean daily minimum and maximum temperatures in the Nattai catchment are around 2.0 ˚C (July) and 27 ˚C (January) respectively (Bureau of Meteorology 2006b). Average annual rainfall is 864 mm with a slight summer maximum and occasional light snowfalls at higher elevations in the upper catchment during winter. However, total annual rainfall often shows large variations between years (range of 400 mm to 1800 mm) due to conditions associated with the El Niño-Southern Oscillation (ENSO).

3. METHODS, FIELD SITES AND SOURCES OF DATA

Data was collected through field investigations and ground truthing, combined with extensive analysis of existing records. This commenced in 2004 and continued for the duration of the project. At the same time, other studies were being carried out by the CSIRO (Titled Impacts on water quality by sediments and nutrients released during extreme bushfires) and by colleagues from the University of Wales Swansea and University of Plymouth, UK. In all cases, investigation focused on post-fire erosion, the result being that we were able to pool together a large and complementary body of research on the topic. At times research from these other groups will be mentioned or utilized (with permission) but none is specifically discussed in this report.

3.1. Field investigations

Two field sites were selected along Blue Gum Creek (catchment area 44.6 km2) to analyse post- fire erosion and to characterise hillslope deposits and catchment geology (Figure 2). The sites are accessed via Fire Road W41 from Thirlmere Lakes, 7 km south-west of Picton. The sites, known as high-burn and low-burn were chosen based on fire intensity during the 2001-02 wildfires (Chafer et al. 2004). The high-burn site (34˚ 13.19 S, 150˚ 29.63 E) is a hillslope on the south- west side of the valley extending directly up from the Blue Gum Creek. Investigations primarily focused on the mid and lower slopes, footslopes and valley floor (Figure 3a). The low-burn site (34˚ 13.60 S, 150˚ 30.36 E) is located within a small (1.9 km2), unnamed tributary catchment of Blue Gum Creek. The site includes a hillslope and spur at the junction of two drainage lines and a large terrace which extends along the valley floor. A third site, medium burn, which is situated between the high and low-burn sites was used for investigations by the CSIRO.

At the high and low-burn sites, soil pits or augers of several meters depth or to bedrock were excavated in transects down the slope commencing at the lowest exposure of bedrock and, in selected geomorphic units to describe the soil mantle and alluvial deposits (Figure 3b). The elevations of the pits were surveyed using a tape and clinometer tied to known benchmarks established through differential GPS. Soil units were identified and described according to Northcote (1979). Selected charcoal samples were submitted for radiocarbon dating at the University of Waikato Radiocarbon Dating Laboratory, New Zealand. Additional information on geology was added through analysis of bedrock outcrop and bedrock encountered in the pits. At approximately the same locations on the hillslopes, erosion pins and bioturbation plots were used to measure post-fire ground surface changes and rates of soil mounding by ants, small mammals and other invertebrates (Shakesby et al. 2003; Shakesby et al. 2006).

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 4

(a) (b)

Figure 3(a) Blue Gum Creek high-burn site showing the steep mid-slopes and (b) gravelly colluvial mantle exposed in a pit.

A landslide situated on the lower slopes below Tumbledown Mountain in the lower Nattai valley was used as a case study to investigate landslide morphology and failure mechanisms (Figure 4). The landslide was first identified using aerial photographs (1934 onwards) and parish maps (1900 onwards). On further investigation, other potential landslides in the Nattai valley and in the Wollondilly valley were found to have been previously mapped by McElroy and Relph (1958). However, no research had been conducted into the timing, reasons and mode of failure. The Tumbledown Landslide was chosen for investigation due to accessibility including a road which extends across the site from Sheehys Creek Road, and the unusual circumstance of large boulders sitting within the Nattai River channel adjacent to the site. Initially it was considered that these boulders might be linked to the landslide or to an earlier slide which crossed the Nattai River. Such an occurrence would dam river flow and have severe consequences for water quality and quantity downstream. Data collection at the Tumbledown landslide site included surveying and mapping of geomorphic units using a differential GPS, analysis of bedrock and soil characteristics in major units through shallow pits and exposures, and a geophysical investigation of soil thickness on pull-apart structures identified below the landslide. Soils were characterized according to McDonald et al. (1990). Sediments were described using Folk (1974)

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 5

Plateau N

Tumbledown Mountain

Dissected valley

TL

Nattai River

Gillans Creek

Steep lower slopes (5 – 60 %)

Figure 4. 1956 aerial photograph of the Lower Nattai valley showing the location of the Tumbledown Landslide (TL) marked in red and boulders in the Nattai River shaded yellow. The Nattai River and Gillans Creek are indicated as a blue dashed line. The upper edge of the cliff line is marked as a white dashed line.

3.2. Existing records

Records of wildfires (and hazard reduction burns, HRB), rainfall, discharge, river sediment concentration and earthquakes were obtained from various sources including the SCA, Bureau of Meteorology (BoM), Geoscience Australia (GA) as well as published works. These were analysed as follows.

Wildfires

• Fire records from the Nattai catchment were used to determine significant wildfire events (> 10 km2) since 1964 (limit of record) and all events which burnt the field sites at Blue Gum Creek (Tables 2a,b).

Rainfall

• Daily rainfall records from eight gauges across the catchment (Table 3) were used to determine individual rainfall events, identify extreme rainfall events, analyse the return intervals of different magnitude events and calculate average catchment rainfall. • 15-minute radar reflectivity data from the BoM Doppler Radar facility located at Kurnell, south of Sydney, were used to determine rainfall intensity, duration, distribution and type (i.e.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 6

drizzle, showers, thunderstorms) for each rainfall event. A GIS layer showing sub- catchments in the Nattai (upper-Nattai, mid-Nattai, lower-Nattai, Little River, Blue Gum Creek) was also used to calculate maximum rainfall accumulation per sub-catchment over 15 and 60 minute time periods. • Synoptic charts sourced from the BoM analysis chart archives (Bureau of Meteorology 2006a) and records of east coast cyclones and tropical cyclones sourced from the Kitamoto Laboratory Japan http://agora.ex.nii.ac.jp/digital-typhoon/search_name.html.en were used to determine weather patterns responsible for extreme rainfall events.

Discharge

• Hourly instantaneous discharge records from the Nattai and Little River catchments (Table 4) were used to determine flood frequency and magnitude since 1965 and 1990 respectively, as well as the hydrological response to individual rainfall events particularly those following the 2001-02 wildfires. Catastrophic flood events were identified using the definition of Erskine and Saynor (1996) which is 10 times the mean annual flood (MAF) peak discharge (MAF in the Nattai River, 5246 ML day-1; and Little River, 1195 ML day-1). • Additional records of floods in the Hawkesbury- at Windsor since 1799 (Sourced from Riley 1980) and from the at Nepean Junction (gauge number 212240) were used to extend the Nattai records to gain a longer perspective on flood frequency and determine maximum historical flood heights.

River sediment concentration

• Turbidity and suspended sediment records from the Nattai River (Table 5) were used in conjunction with the 40-year discharge data to estimate contemporary river sediment load and evaluate the contribution of sediment from wildfires and catastrophic floods. Turbidity and suspended sediment records from the Little River were also obtained, however, the record was found to be too short and discontinuous to be used to reliably estimate contemporary river sediment load.

Earthquakes

• The GA earthquake database http://www.ga.gov.au/oracle/quake/quake_online.jsp was used to determine earthquake distribution and magnitude within or close to the Nattai catchment and SCA Special Areas since 1872. • Estimated recurrence intervals of earthquakes in the western Sydney Basin seismic zone which includes the Blue Mountains Plateau, Nattai catchment and Lake Burragorang were obtained from Berryman and Stirling (2003). • Earthquake records were compared with landslide distribution mapped by McElroy and Relph (1958) to ascertain whether a cause-effect link might exist.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 7

Table 2a. Major wildfires and hazard reduction burns (HRB) (> 10 km2) in the Nattai catchment since 1964. (Data source: SCA. Winter and Watts 2002) Date of fire Type of fire and location Area burnt (km2) 9 Feb 1965 Wildfire on the Wanganderry Tableland and Lower 97 Nattai catchment 28 Nov–2 Dec 1968 Wildfire in the Mid Nattai, Little River and Blue Gum 423 Creek catchments 9 Sept 1980 Wildfire in the Lower Nattai catchment at 12 with the 5 Sept 1985 HRB on the Wanganderry Tableland 14 1988/89 Wildfire in the Upper Nattai and headwaters of the 23 Little River catchment 24 Aug 1993 HRB along Blue Gum Creek 12 26 Nov–18 Dec 1997 Wildfire on the Wanganderry Tableland 63 3 Dec 2001–14 Jan 2002 Wildfire in the Mid Nattai, Little River, Blue Gum Creek 530 and Lower Nattai catchments

Table 2b. Wildfires and hazard reduction burns (HRB) at the Blue Gum Creek field sites since 1964. (Data source: SCA. Winter and Watts 2002) Date of fire Type of fire and location Area burnt (km2) 22 Aug 1964 Escape from HRB which burnt down the Blue Gum 4.3 Creek valley including low-burn site 28 Nov–2Dec 1968 Major wildfire which burnt all field sites 423 1971/72 HRB on the plateau (Buxton Ridge) above the field 4.3 sites 24 Aug 1993 HRB in the valley including all the field sites 12 3 Dec 2001–14 Jan 2002 Major wildfire which burnt all field sites 530

Table 3. Rainfall records for the Nattai catchment (Data source: SCA (station number 568); BoM (station number 068)) Rainfall gauge Identification number Period of record Nattai Causeway 568140 1981–2005 Hilltop Nattai Tableland 568164 1990–2005 Hilltop Starlights Track 568050 1981–2005 Leicester Park 568099 1946–2005 Mittagong High Range 068030 1945–2003 Mittagong Kia Ora 068033 1902–2003 Buxton Amaroo 068166 1967–2004 Oakdale Cooyong Park 068125 1963–2004

Table 4. Discharge records for the Nattai catchment (Data source: SCA) Flow gauge Identification Period of Gauged catchment number record area (km2) Nattai River at the causeway 212280 1965–2005 446 Little River at Fire Road W41 2122809 1990–2004 104

Table 5. Turbidity and suspended sediment records for the Nattai catchment (Data source: SCA) Water quality monitoring site ID Period of record – Period of record – Suspended number Turbidity sediment Nattai River at Smallwoods E210 1961–2005 1991-2005 Crossing Little River at Fire Road W41 E243 1991–2005 1991-2005

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 8

4. CHARACTERISTICS AND DOMINANT EROSIONAL PROCESSES ON HILLLOPES IN THE NATTAI CATCHMENT

4.1. Blue Gum Creek

Hillslope morphology and stratigraphy

The hillslopes along the Blue Gum Creek valley are characterized by distinctive gravelly colluvial mantles, which are continuous i.e. not interrupted by bedrock controlled breaks in slope, and thicken downslope. Such mantles seem to be common throughout the Nattai valley, but are not typical of other sandstone hillslopes in the Sydney Basin.

Investigation at the high-burn site along Blue Gum Creek revealed an asymmetrical valley with the mantle accumulating on the north-east facing slope (concave profile) and in the valley floor of the creek and side tributaries, compared with mostly outcrop on the south-west facing slope (convex profile) (Figure 5). On the north-east facing slope, angles range from 30° – 40° on the upper slopes, 10° – 30° on the mid slopes and 5° – 10° on the lower slopes, whereas the foot slopes and floodplain vary locally with slopes ranging from 0° – 7°. Assessment of the geology at the site showed an easterly dip on the Hawkesbury and Narrabeen rocks of 6˚, implying that slopes with an easterly aspect are more likely to be susceptible to failure.

Figure 5. Cross-section of the Blue Gum Creek valley at the high-burn site, showing the locations of soil pits, valley shape and geology (VE = 2.2).

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 9

The sediments on the north-east facing slope and in the valley floor at the high-burn site can be broadly classified into five geomorphic units (see Figure 6):

1. Colluvial mantle, which occurs from the upper slopes to the lower slopes (Pits HMU to HBL). The mantle is characterized by large poorly sorted angular gravels within a sand- mud matrix (bulk density of the matrix material = 0.88 – 1.49 g cm-3). The uppermost 5–7 cm is highly organic but this may be related to the fires in 2001-02. Surface gravels decrease in size from average 18 cm to 6 cm downslope on the mid slope (with occasional larger boulders). On the lower slopes, the smaller gravels are absent and only partially buried large boulders can be found. A gully which dissects the mantle adjacent to the transect has large boulders (B-max > 1.5 m) composed of Hawkesbury sandstone chaotically arranged on the slope. Adjacent to and below the gully, are lobes of poorly sorted gravels and boulders (and sands) with steep slip faces.

2. Fluvial sands, which occur in the base of the valley from around 170 cm below the surface to a depth of 3.5 m in the auger. The sands are typically coarse, moderately well sorted and clean, with only a slight brown or grey organic stain (clay content < 8 %). Rounded to sub-rounded gravels of B-max = 48 mm were found in the unit in Pit HFSL. A lens (10 cm thick) of medium clay coarsening upwards to a sandy clay loam was encountered in Pit HFC. The clays indicate a much lower energy depositional environment compared to the surrounding sands. The water table was intersected at a depth of 2.8 m despite prolonged drought conditions. Mottling of the clays indicates the permanency of wet conditions.

3. Floodplain deposits, which overlie the fluvial sands and form the surface of the valley floor (Pit HFC and Auger HFP). These deposits consist of massive brown-black clay loams and sandy clay loams, 140 – 174 cm in thickness. The sediments are typical of overbank, floodplain deposition whereas the modern channel is transporting a sheet of coarse sand similar to the fluvial sand unit below.

4. Intermixed deposits, which consist of massive orange clayey sands (sub-angular to sub- round) with small well rounded pebbles dispersed throughout, alternating with a thin angular gravel lens (B-max = 4 cm). This unit is termed intermixed as it has proven difficult to resolve the origin of the material. It could be fluvial or colluvial in origin or form part of an alluvial fan.

5. Older valley fill, which was encountered in the base of Pits HBL and HFSU below 150 cm. The thickness of this unit is at least 170 cm (as the pit did not reach bedrock). The sediments are characterized by sands and sub-angular to sub-round gravels in a sandy clay loam matrix, which become lithified at depth. The average bulk density of the unit is 2.006 g cm-3, which is much higher than the clayey sands above (1.46 – 1.62 g cm-3) and other units in the valley.

At the low-burn site, hillslope stratigraphy includes a gravelly colluvial mantle with similar characteristics to the high-burn site. In the floor of the valley, however, there are thick (2–8 m) sandy deposits preserved between the two tributary streams as a spur and adjacent to the right valley margin as a terrace. Soil texture is sand to sandy clay loam with very few gravels encountered. Bulk density ranges from 1 to 1.5 g cm-3 whilst colour is strongly red and yellow due to oxide coatings on the sand grains. Inset within the terrace is a narrow floodplain composed of sandy loam and sandy clay loam. Bedrock and/or gravels were encountered 1–2 m below the surface, as well as exposed in the creek bed. The gravels can be traced downstream and into Blue Gum Creek, but not upstream of the confluence, indicating that they are part of a distinct deposit sourced from the tributary. The morphology and arrangement of the gravels and thick sandy units suggests that the units once formed part of a much larger alluvial fan, which is now being incised and reworked to form an inset floodplain.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 10

Figure 6. Profiles for each pit/auger hole at the high-burn site. Interpretation of geomorphic units is based on sediment texture, character and position in the landscape. Radiocarbon ages (years BP) and sample locations are indicated by solid squares (VE = 1.8).

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 11

In addition to the gravelly colluvial mantle, the surface morphology at both sites is characterized by a thick, but patchy leaf litter layer with abundant bioturbation including ant mounding, small mammal scrapes and tree throw (Figure 7a-c). In the first years following the 2001-02 wildfires soil pedestals, litter dams and small alluvial fans were observed indicating significant post-fire erosion and sediment movement. However, only the low density and finer material such as charcoal, leaf litter and clays were observed as drapes on the bed of the Nattai River. Transport of the coarser sands and gravels appeared to be confined to the hillslopes i.e. localized redistribution rather than net export.

(a) (b) (c)

Figure 7a) Small alluvial fan (~50 cm length) composed mainly of organics and charcoal blanketing the gravels on the hillslope, (b) tree throw after a severe storm, and (c) mound of Aphaenogaster longiceps, the funnel ant.

Dominant erosional processes

The valley morphology and stratigraphy of sediments on the hillslopes at Blue Gum Creek indicates that the dominant erosional processes (in terms of the volume of sediment moved) are event-based rock fall and debris flows (terminating as alluvial fans), with rainsplash and slope wash particularly after wildfires playing a lesser role. Rock fall is primarily the result of weathering and detachment of individual blocks along bedding planes and joints in the sandstone cliffs (Figure 8a). These can occur at anytime, with the most recent example in the Nattai catchment being a small rock fall which blocked the Nattai North access road on 8–10 July 2005 (Figure 8b). Debris flows on hillslopes occur primarily along tributary streams and in small drainage paths. At the Blue Gum Creek high-burn site, debris flow deposits are prevalent on the surface in the form of built-up lobes of rock and coarse sediment (Figure 8c) with distal fining of material downslope possibly represented by the very sandy deposits in pits HFSU and HFSL. Older debris flow deposits appear to be preserved as colluvial hollows. Colluvial hollows infill depressions in bedrock but often have no defined surface expression and lay undiscovered until exposed in road cuts. Good examples can be seen along Sheehys Creek road in the south of the Nattai catchment, where the gravelly fills are cut into the Narrabeen Group rocks (Figure 8d).

Preliminary radiocarbon dating of charcoal from deposits at the high-burn site (Figure 6) indicates that the colluvial mantle and most of the units infilling the Blue Gum Creek valley are less than 6 ka (mid-late Holocene) (Tomkins et al. 2004a) in age and therefore, formed under similar climatic conditions to the present and from material sourced from the adjacent slopes and side tributaries. However, one of the problems with radiocarbon dating of charcoal is the influence of bioturbation and mixing of the soil. Younger charcoal may be incorporated into older sediments at depth via burrows, tunnels and ant galleries. As a result caution needs to be applied until more precise dating of sediments is obtained. The resolution of dating also precludes establishing the number of individual erosion-sedimentation events and the size of those events.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 12

(a) (b)

(c) (d)

Figure 8(a) example of a small rock fall from the sandstone cliffs, (b) disturbance caused by a small rock fall on 8–10 July 2005 at Nattai North, (c) debris flow deposit at the Blue Gum Creek high-burn site, and (d) colluvial hollow exposed along Sheehys Creek Road.

Key issues for sediment supply and water quality

One of the major controls on mass movement in the Nattai catchment appears to be geology, both in terms of the composition of the bedrock and the structure of units. Mass movement seems to become a dominant process where rivers have incised through the resistant but highly jointed Hawkesbury Sandstone and into the underlying Narrabeen Group and Permian rocks. These locally weaker lithologies weather and erode more easily undermining the overlying massive sandstone rocks. The result is an abundant supply of large volumes of sediment from hillslopes. The lack of sediment accumulation along the Nattai valley suggests that most is exported from the catchment or only temporarily stored on the lower slopes and valley floor. This implies that a significant mass movement event within the lifetime of the reservoir would produce a dramatic increase in sediment and nutrient load during the event and afterwards through reworking of any stored materials. Minor events such as the recent small rock falls are unlikely to have any effect on water quality or quantity in the Nattai catchment.

The role of wildfire in reducing vegetation cover and exposing the soil to rainsplash erosion and sheet wash is also an issue for sediment supply and water quality, particularly in terms of the low density and finer material (organics, charcoal and clays) which are more easily transported into and through the river network as floating and suspended load. Investigations undertaken by the CSIRO study have been able to quantify this following the 2001-02 wildfires (Wallbrink et al. 2004; English et al. 2005; Wilkinson et al. 2006). Transport of the sandy and gravelly material, however, seems to be limited despite conditions which suggest considerable erosion potential i.e. steep slopes, abundant sediment supply, evidence of rainsplash erosion and sheet wash, and connectivity between hillslopes and rivers. The reasons for this may relate to rainfall and runoff in the post-fire period, which forms the focus of Section 5.1 in this report.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 13

4.2. Tumbledown Landslide

Landslide morphology and stratigraphy

The Tumbledown Landslide is one of a number of pre-European mass movement deposits identified in the Nattai valley and around the foreshores of Lake Burragorang (Figure 9). The landslide is a composite feature consisting of a rock avalanche (or sturzstrom) sitting on top of multiple rotational slumps. The rock avalanche is composed of chaotically arranged large sandstone boulders sourced from the failure of a 260 m long section of cliff face in the lower Nattai valley. The deposit is estimated to have a volume of 1.5 million m3, a runout distance of 570 m and is within 320 m of the maximum water level of the Nattai arm of Lake Burragorang.

Figure 9. Known landslide distribution around the foreshores of Lake Burragorang, including several post-European landslides related to underground coal mining and ground subsidence. Tumbledown Mountain is shown in the lower-right of the figure.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 14

Mapping of the landslide site revealed a complex morphology both in plan view and in cross- section (Figures 10 and 11). The rock avalanche itself has several lobes or mounds of deposits giving rise to an undulating surface topography with tens of meters of relief including a steep (35– 55˚), thick (5–15 m) leading edge. The avalanche appears to have preferentially flowed down and infilled prior drainage lines and depressions (Figure 12a) except on the front right where a large distal pressure ridge has formed through forward motion of the rocks and friction with the underlying surface. In front of and on the lower sides of the avalanche are several unusual and extensive very low relief almost flat areas, each with a slight (average 4˚) backwards tilt in towards the cliff and declining in elevation forming a step-like or benched morphology. The front of each is marked by a steep head scarp up to 40 m in height, which extend almost parallel to the Nattai River over distances of 200 m to > 1 km (Figure 12d). Drainage lines have formed along the base of some of the head scarps and where enclosed, small swamps have developed ringed by paperbark trees (Melaleuca sp). A seismic survey conducted along a cross-section through the head scarps revealed that soil thickness above the bedrock was around 10 m and reasonably uniform down the slope (± 3 m). Soils are predominantly clayey reflecting weathering of the underlying Permian rocks. In some units angular sandstone gravels are also present. On either side of the upper part of the rock avalanche are talus deposits which thin away from the cliff face. The deposits consist of smaller sandstone gravels within a sand-mud matrix similar to the colluvial mantle at the Blue Gum Creek sites. On the talus slope above the avalanche are several large rotated columns of sandstone up to 50 m in height which appear to have slid down the slope (Figure 12e). These probably occurred at the same time as the rock avalanche failure, but remained in tact rather than disaggregating into blocks through fracture along bedding planes.

Figure 10. Geomorphic map of the Tumbledown Landslide showing the area of failure on the escarpment, the rock avalanche, head scarps indicating rotational slumping in the bedrock and boulders in the Nattai River sourced from Gillans Creek.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 15

The landslide site also includes numerous large (> 3 m diameter) sub-rounded sandstone boulders situated within a 650 m section of the Nattai River (Figure 12b), as well as a similar looking deposit which extends up to 70 m above the river bed on the opposite bank. As mentioned earlier, it was initially thought that the boulders in the river might have been part of the Tumbledown landslide or an older, larger landslide that crossed the Nattai River. On further investigation, however, the boulders were found to have come from Gillans Creek, a tributary which flows into the Nattai River just upstream of the Tumbledown Landslide (Figure 4). The boulders increase in size downstream, which is unlike typical landslide deposits (chaotic arrangement) or fluvial deposits (downstream fining), and instead suggests that they formed part of a large debris flow that originated in Gillans Creek and flowed down the Nattai River for a limited distance as channel slope declined. The coarsest boulders now form a lag within the river bed appearing to be beyond the competence of contemporary floods.

Dominant erosional processes

Failure of the cliff face at the Tumbledown Landslide site appears to be the result of a number of complex and inter-related erosional processes which have not been identified within the Nattai catchment or elsewhere in this region before. Firstly, the morphology and seismic survey of the head scarps and flat areas behind suggests that deep-seated rotational slumping is occurring along failure planes within the exposed weaker Permian bedrock in the valley (Figure 11). The slumping may be related to incision of the Nattai River and trimming of the toe of the bedrock which leads to on-going slippage of each block as material is removed. Slumping within the bedrock is likely to be slow but with time can lead to significant displacement as evidenced from the head scarps. Secondly, it appears that slumping within the Permian bedrock must extend beneath or close to the valley sidewalls, reducing basal support for the overlying sandstones. Changes in the stress distribution across the cliff face are evident as tensional cracks which propagate vertically through the sandstone preferentially following jointing patterns (Figure 12c). The rock avalanche at the Tumbledown site appears to have resulted from failure of the cliff face through translational sliding of sandstone columns downslope, then fracturing along bedding planes as the mass extended across the slumped terrain. The exact timing of rotational slumping and rock avalanching is unknown. It is likely that the slumping is continuous or related to floods within the Nattai River. The rock avalanche is noted on parish maps dating from 1900 and has a good tree cover indicating at least a pre-European age and probably substantially older. An estimate of age based on valley denudation rates is older than 2,500 years and less than 250,000 years.

The boulders situated within the Nattai River imply that debris flows are a third dominant and inter-related erosional process in the Nattai valley. Debris flows were discussed earlier in Section 4.1 but the Tumbledown Landslide site indicates that they may form a critical mode in the transport of coarse material from the plateau and hillslopes into the river network. That is, debris flows mobilize material made available on hillslopes through rock fall, rock slide and rock avalanching, transporting it either directly or in stages to the major rivers via tributary streams and drainage lines, where it can then be exported from the catchment. The frequency of debris flows must be equal to or greater than the supply of sediment from the hillslopes otherwise the valleys would infill and sidewalls would decline, which is clearly not the case in the Nattai valley.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 16

W E

Figure 11. Cross-section through the Tumbledown Landslide showing the rock avalanche and interpretation of rotational slumping in the Permian bedrock.

(a) (b) (c)

(d) (e)

Figure 12a) Tumbledown Landslide, showing a drainage line at the side of the landslide which is partly infilled with boulders; (b) boulder lag in the Nattai River (note person for scale); (c) tensional cracking within the sandstone cliff face; (d) head scarp indicating failure planes within the Permian bedrock, and (e) rotated sandstone block on the hillslope above the rock avalanche.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 17

Key issues for sediment supply and water quality

Both the discovery of landslides within the Nattai and Wollondilly valleys and investigation of the failure mechanisms at the Tumbledown site has raised several key issues for sediment supply and water quality within SCA Special Areas. All of the known landslides have been found along the immediate foreshores and adjacent slopes of Lake Burragorang. This may be coincidence because these areas are more accessible by road and form the focus of this investigation. The evidence from the Tumbledown site however, suggests that the landslides are directly linked to exposure of the Permian strata through river incision. The geology of the Sydney Basin has been well mapped and shows that most of the Burragorang reservoir (including the lower Coxs, lower Wollondilly and Nattai Rivers) is located within the Permian Illawarra Coal Measures and Berry Formation. In these locations rotational slumping, rock fall and rock avalanching has led to the development of the broad valleys, compared with for example, the Warragamba gorge, which flows through Hawkesbury and Narrabeen Group strata. The exposed Permian strata around the foreshores of Lake Burragorang, means that landsliding is expected to continue. The implications of such an event on sediment influx, dam capacity and wave generation have been well documented from examples of landslide failures into reservoirs in Italy (Vajont Dam, Pontesei Reservoir) (e.g. Panizzo et al. 2005) and more recently associated with the Three Gorges Dam in China (e.g. Liu et al. 2004). However, there are many unknowns regarding the existing landslides around the foreshores of Lake Burragorang including the time of failure, the prevailing climate conditions or triggers (earthquakes are often suggested), and thresholds which lead to collapse. Further investigation is also warranted to establish the frequency and rates of rotational slumping, as well as the effects of reservoir impoundment on pore-water pressure within the Permian bedrock and older landslides which extend below maximum water level e.g. Bimlow Landslide. It is possible that all the currently preserved landslides occurred at similar times or under similar conditions such as during a period of enhanced fluvial activity or as a result of a high magnitude earthquake. Alternatively, it is possible that landslides occur randomly throughout time depending on internal thresholds and forces and only the most recent are still preserved in the landscape today.

The role of debris flows in transporting large volumes of sediment into the major rivers is also an issue for sediment supply and water quality. To date, the importance of these processes has been poorly recognized compared with for example, post-wildfire erosion. The Gillans Creek debris flow deposit indicates both the size and volume of sediment able to be transported from smaller tributary streams. The implications of an equivalent magnitude event occurring in the future are significant, not only in terms of sediment input into the reservoir but also through damming or disruption of river flows. The effects of landslide-damming have been well documented globally (e.g. Webb et al. 2003; Dai et al. 2005; Korup 2005).

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 18

5. TRIGGERS OF EXTREME EROSION EVENTS

5.1. Wildfires

Wildfires are often thought to be a major trigger of extreme erosion in the Australian landscape. The reasons for this are two-fold: (i) wildfires usually result in very high to extreme burning of the vegetation cover which exposes the soil to geomorphic processes including rainsplash and slope wash; and (ii) research on wildfire impacts from North America has shown that severe erosion (including debris flows, rilling, gullying and sheet erosion) follows severe wildfire through the increased effectiveness of the first post-fire rainfall events (Robichaud 2000; Cannon et al. 2001; Moody and Martin 2001). As a consequence, the results from the North American studies have been applied to southeastern Australia leading to the expectation of severe erosion within the post-fire period and the view that wildfires are a major cause of extreme erosion in fire-prone landscapes globally.

The results from the Blue Gum Creek study sites following the 2001-02 wildfires, however, suggests a more complicated response with limited erosion and transport of the sandy material in the post-fire period compared with the export of organics and finer material. Several issues are raised as to why this might be so, including the characteristics of rainfall (and runoff) after the fires, effects on soils and rates of vegetation recovery. A further issue is whether the 2001-02 wildfires were typical of all fires or if the lower-than-expected rates of sediment transport are unusual. Examination of rainfall in the period following the 2001-02 wildfires showed below average totals for three years after the fires with a small number of events of mostly patchy, low- medium intensity rainfall (showers or drizzle) and short lived high intensity summer thunderstorms (Table 5, Figure 13). In the case of the former, very little runoff and thus little increase in river discharge resulted indicating that most rainfall infiltrated into the soil, whereas the thunderstorms triggered a small, but sharp rise in river height indicating rapid runoff from the hillslopes.

Table 5. Summary of rainfall events in the first three years after the 2001-02 wildfires Rainfall type Year Long duration, low Patchy, low–medium Localised, short duration intensity (< 10 mm hr-1) intensity (10–30 mm hr-1) storms with well defined stratiform rainfall stratiform rainfall (showers, cells where intensity > 75 (showers, drizzle, 1a) rainfall, 1b) mm hr-1 (thunderstorms, 2) 2002 1 11 6 2003 1 9 9 2004 0 10 11 Total 2 30 26

1a 1b 2

Figure 13. Examples of rainfall type occurring in the Nattai catchment within the first years following the 2001-02 wildfires. Type 1a and 1b is low-medium intensity rainfall. Type 2 is short- duration high-intensity thunderstorms. Sub-catchments within the Nattai catchment are labeled as follows: 1. upper-Nattai, 2. mid-Nattai, 3. Little River, 4. Blue Gum Creek, 5. lower-Nattai. Source: Bureau of Meteorology and Corporation.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 19

The short-lived nature of the thunderstorms and localized distribution however, meant that the effects were rarely sustained or catchment wide. In areas where the storms tracked, significant rainsplash erosion resulted from rainfall intensities of > 75 mm hr-1.

Examination of rainfall within the first 12 months following the 1965, 1968 and 1997 wildfires, were compared to the 2001-02 fires (Figure 14). The result reveals a variable pattern. Rainfall following the 1965 fires was below average with only a small number of low magnitude rainfall events. In contrast the 1968 and 1997 wildfires were characterized by above average rainfall and a larger number of events including at least one very large event each of > 60 mm day-1 average rainfall across the catchment, which generated significant runoff and increase in discharge in the Nattai and Little Rivers. It is possible that these larger rainfall events in the post-fire period could trigger extreme erosion on hillslopes including stripping of topsoil, but the relationship between fire and large rainfall events compared with large rainfall events in non-fire periods is not certain i.e. does fire enhance the impact of these events or have no effect? A comparison of discharge resulting from three similar sized rainfall events, two of which occurred after the 1968 and 1997 wildfires and the third in 1986 indicated increased post-fire discharge with respect to rainfall, suggesting that fires increased runoff rates from hillslopes. Analysis of turbidity records from the Nattai River following the 1968 and 2001-02 fires compared with non-fire periods however, showed elevated turbidity at low discharge (but within the range of normal scatter) and no difference at higher discharge (> 5000 ML day-1) suggesting that at least with respect to suspended sediment, wildfire has no apparent effect on sediment yield rates (Figure 15). Measurements of bedload transport are needed to determine the effect on total sediment transport. In summary, (i) the characteristics of rainfall in the post-fire period provides a likely explanation for the lower-than-expected erosion observed at the Blue Gum Creek sites following the 2001-02 wildfires; (ii) there is no consistent pattern in post-fire rainfall therefore large magnitude rainfall events may or may not occur in the post-fire period; and, (iii) erosion potential will also vary both between fires and in the years after fire, becoming lower with time after fire as the vegetation recovers.

Figure 14. Rainfall totals and number of events within the first 12 months after the 1965, 1968, 1997 and 2001-02 wildfires and 1985 hazard reduction burn. Average annual rainfall, 864 mm is indicated.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 20

The above results suggest that wildfires are a less important trigger of extreme erosion than otherwise thought, but the nature and extent of erosion is highly dependent on the characteristics of rainfall in the post-fire period. Severe and widespread wildfires are estimated to occur around once every 33 years based on the 1968 and 2001-02 wildfires. Smaller wildfires seem to occur around once every 10 years depending on fuel loads and burn history. Using these recurrence intervals, an estimate of the contribution of severe wildfires to contemporary sediment yield in the Nattai catchment is around 5 %.

Figure 15. Pre- and post-fire turbidity data from the Nattai catchment. Red circles indicate data collected after the 1968 wildfire (to 1974). Yellow circles indicate data collected after the 2001- 02 wildfire (to 2003). Crosses and line indicate pre-fire data and trend. For scaling purposes, three pre-fire values of > 600 NTU are not shown including one datum collected during the 1978 flood (> 1000 NTU).

5.2. Extreme rainfall

Extreme rainfall is likely to be major trigger of extreme erosion on hillslopes in the Nattai catchment, primarily through (i) rainsplash, especially where the vegetation cover is low such as under woodland, on steep upper slopes or following fire; (ii) slope wash especially where concentration of runoff forms rills and gullies; (iii) debris flows along drainage lines; and, (iv) flooding which leads to sediment mobilization and transport along river channels. Extreme rainfall in the Nattai catchment is defined in two ways:

1. Rainfall of > 60 mm day-1 (hereafter referred to as heavy rainfall) using an average from up to eight rainfall gauges located across the catchment. This amount is equivalent to a one year average recurrence interval at each gauge. Rainfall is generally widespread, medium to high intensity and long duration (> 24 hours) producing significant surface runoff from hillslopes leading to flooding in the major river systems.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 21

2. Rainfall with intensities exceeding 100 mm hr-1. This captures severe thunderstorms which can produce significant rainfall in short periods of time depending on cell size, location, speed and direction.

The synoptic conditions which are most likely to produce heavy rainfall are east coast cyclones (ECC) and tropical cyclones that move south into the higher latitudes. ECC originate on the east coast of Australia between latitudes 20–40 degrees, occurring mostly in the autumn and winter months when the land-sea temperature difference is greatest (Hopkins and Holland 1997). Intense ECC occur around once per year (Holland et al. 1987) producing large-scale orographic rainfall sometimes with embedded thunderstorms providing a portion of the heavy rainfall (Lynch 1987). Between 1958 and 1992 there were 22 ECC (Hopkins and Holland 1997). The most significant rain days in the Nattai catchment occurred on 20 March 1978 (177.3 mm day-1), 12 June 1964 (156.2 mm day-1) and 6 August 1986 (139.5 mm day-1) (Table 6). During the 1986 ECC, the areas of maximum precipitation were recorded at Robertson on the Woronora Plateau and Katoomba on the Blue Mountains Plateau due to topographic uplift, and north of Bankstown due to convective heating of the atmosphere (Hess 1990). Heavy rainfall also results from tropical cyclones formed in the or Gulf of Carpentaria during summer, which move southwards along the east coast of Australia or inland decaying to become rain depressions. At least seven heavy rainfall events in the Nattai catchment have resulted from southward moving tropical cyclones, with the most significant rain days occurring on 19 January 1950 (137.3 mm day-1), 16 April 1969 (126.4 mm day-1) and 3 February 1990 (101.9 mm day-1) (TC Nancy, shown in Figure 16a,b). Monsoonal depressions, which develop over Western Australia and move east across the continent are a third source of heavy rainfall. One known example occurred in 1983 producing a maximum on the 22 March of 112.4 mm day-1 in the Nattai catchment.

Table 6. Significant heavy rainfall events in the Nattai catchment since 1945 Date Maximum rainfall Total rainfall Total number of Synoptic (mm day-1) (mm) rain days conditions1 17–27 March 1978 177.3 466 11 ECC 5–13 June 1964 156.2 325 9 ECC 4–10 August 1986 139.5 281 7 ECC 19–20 January 1950 137.3 147 2 TC 15–16 April 1969 126.4 162 2 TC 15–23 March 1983 112.4 241 9 MD 1–13 February 1990 101.9 278 13 TC 1ECC, east coast cyclone; TC, tropical cyclone; MD, monsoonal depression

Intense thunderstorms are often the result of convection associated with inland troughs or prefrontal troughs during the summer months (November to April). Storm cells track north- easterly or south-easterly across the Sydney Basin as single isolated cells or multi-cells (Figure 13 (Type 2)). Maximum rainfall intensity in the cell core can exceed 75 mm hr-1 with 15-minute rainfall accumulations within sub-catchments exceeding 10 mm depending on the size, speed and duration of the storm cell. In the Nattai catchment rainfall associated with thunderstorms can last for 2–15 hours, but more often is between 6–8 hours. Maximum rainfall recorded at a single gauge is < 60 mm day-1, or < 45 mm day-1 averaged across the catchment. The frequency of thunderstorms varies per year (Table 5). The most significant storms in terms of maximum rainfall accumulation per sub-catchment over 15 minutes (A15) and 60 minutes (A60) since 2002 are shown in Table 7.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 22

(a) (b)

Figure 16. Tropical Cyclone Nancy moving down the east coast of Australia on (a) 3 February 1990 and (b) 4 February 1990, bringing rainfall of > 100 mm day-1 to the Nattai catchment. Source: National Institute of Informatics, Kitamoto Laboratory, Digital Typhoon Project, http://www.digital-typhoon.org

Table 7. Significant thunderstorms in the Nattai catchment since 2002 Date Duration No. Max A15 per Max A60 per Max rainfall Max recorded (hours) of sub-catch. sub-catch. intensity at a gauge cells (mm) (mm) (mm hr-1) (mm) 5 Dec 2004 7.75 1 13.7 (MN) 41.7 (MN) > 75 25.5 13 Dec 2004 8.75 3 14 (MN) 30.5 (LR) > 75 43.5 1 Feb 2005 7 1 13.8 (LR) 31 (BGC) > 75 56.5 2 Feb 2005 2.5 1 20.7 (LR) 39 (LR) > 75 43 19 Feb 2005 8.5 1 11.4 (LR) 30.4 (LR) > 75 48 20 Feb 2005 16 1 16.8 (BGC) 38 (LN) > 75 47 Sub-catchments are indicated as follows: MN, mid-Nattai; LR, Little River; BGC, Blue Gum Creek; LR, lower-Nattai.

The frequency and magnitude of extreme rainfall events suggests that they are an important trigger of extreme erosion within the Nattai catchment, and probably more so than generally recognized. The recurrence intervals of events based on existing records range from 1–15 years, due to variations in controls including regional climate, synoptic-scale weather patterns and large- scale atmospheric-ocean phenomena such as the El Niño-Southern Oscillation (ENSO). Hence future extreme rainfall events are difficult (if not impossible) to predict except in the sense that an event(s) could in occur in any year.

5.3. Catastrophic floods

Catastrophic floods in southeastern Australia have been found to generate up to 283 times the mean annual sediment yield of a catchment (Erskine and Saynor 1996) through reworking of colluvial and fluvial material stored in the channel bed and banks, and net transport of sediment from the system. The cause of flooding is extreme rainfall (widespread, medium-high intensity, long-duration events), hence the two are intrinsically linked particularly in terms of frequency and magnitude. However, not all extreme rainfall events will result in catastrophic floods. This is because of on-ground factors which affect runoff such as antecedent soil moisture, vegetation cover and soil infiltration rates, as well as rainfall variables such as duration and intensity.

There have been three catastrophic floods in the Nattai River since 1965 (Figure 17, Table 8) using the definition of Erskine and Saynor (1996) (ten times the mean annual flood (MAF) peak discharge). The first occurred in April 1969, five months after the 1968 wildfires. As mentioned

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 23

earlier, the fire appears to have increased runoff and discharge relative to rainfall with the flood hydrograph showing a very sharp single peak which had receded within a few hours. The largest flood was in March 1978. This flood generated almost eight times the mean annual sediment yield within six days or 21 % of the total sediment yield produced over 40 years. Since 1978 flood frequency and magnitude has declined particularly after 1992 which may be reflection of repeated drought conditions across southeastern Australia. No catastrophic floods have occurred in the Little River since records commenced in 1991.

Figure 17. Flood record from the Nattai River since 1965 showing three catastrophic floods in 1969, 1975 and 1978.

Table 8. Catastrophic floods in the Nattai River since 1965. Date Peak Return Cause Max No of times greater discharge interval rainfall than mean annual (ML day-1) (years) (mm day-1) sediment yield 15–22 April 1969 68,269 > 20 Tropical 126.4 2.1 cyclone 21 June–2 July 1975 55,106 > 20 Unknown 150.7 3.3

18–30 March 1978 77,500 ~ 50 East coast 177.3 7.8 cyclone

A good indication of catastrophic floods in the Nattai catchment pre-1965 can be gained from the Hawkesbury-Nepean River flood record which commenced in 1799 (Riley 1980). Using the height of the 1975 flood as the lower limit, there have been 34 catastrophic floods recorded within the last 207 years. Most of these occurred in June and between February to April with none occurring in the months of October, December or January (Figure 18). The largest catastrophic flood (by far) occurred in June 1867 with a return interval of around 200 years. The cause of the flood is thought to have been a severe ECC (NSW State Emergency Service 2005) with continuous heavy rain reported over six days and maximum totals of > 100 mm day-1 (Bracewell and McDermott 1985). The estimated peak discharge in the Nattai River during this event was 215 000 ML day-1 or 41 times MAF. Sediment yield is estimated to have been 52 times greater than mean annual or equivalent to 55 years of contemporary sediment yield in a single event.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 24

Figure 18. Monthly frequency of catastrophic floods in the Hawkesbury-Nepean River since 1799.

Catastrophic floods are important triggers of extreme erosion events within the Nattai catchment, particularly in terms of their ability to mobilize extremely large volumes of sediment over short periods of time (measured in hours to days). The contribution of catastrophic floods to contemporary sediment yield in the Nattai catchment based on the 40-year record is estimated to be around 35 % (seven times greater than wildfire). Using the longer Hawkesbury-Nepean record of floods over 207 years the amount increases to 55 %.

The return interval of a catastrophic flood in the Nattai catchment is at least 20 years. The record from the Hawkesbury-Nepean River however, shows an uneven distribution over time (Figure 19). Three flood-dominated periods were identified by Erskine and Warner (1988): 1799–1819, 1857–1900 (including the 1867 flood) and 1947–1978. The period since 1978 has seen relatively few floods and no catastrophic floods. Floods in southeastern Australia are thought to be more likely during La Niña years or when the southern oscillation index (SOI) is positive (Chiew et al. 1998). However, SOI data available from 1876 when compared to the Hawkesbury-Nepean record shows a less clear pattern with six catastrophic floods occurring when the SOI was positive (the classic La Niña association), three occurring when the SOI was negative and four occurring when the SOI was neutral. Of the catastrophic floods reported in Table 8 only the 1975 event conforms to a classic La Niña pattern (Bureau of Meteorology 2007). Hence it seems that catastrophic floods, like rainfall, could occur in any year. In fact there are four years on record (1806, 1809, 1819 and 1956) where two catastrophic floods occurred in a single year, which has significant implications for erosion potential.

Figure 19. Annual frequency of floods in the Hawkesbury-Nepean River since 1799 showing three flood-dominated regimes (FDR) between 1799-1819, 1857–1904 and 1949–1978, and three intervening drought-dominated regimes (DDR).

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 25

5.4. Earthquakes

Earthquakes are a known trigger of landslides in Europe, USA and Asia, particularly in tectonically active regions (Keefer 2002; Dadson et al. 2003). As described earlier, several large landslides that predate European settlement have been found in the Nattai and lower Wollondilly valleys. Others probably exist within the Coxs valley and tributaries incised into the Permian rocks but are difficult to detect due to inaccessible terrain and thick vegetation cover. The question is whether earthquakes in the Sydney Basin could trigger landslides in the incised valleys of the Blue Mountains Plateau.

The Sydney Basin, which extends to the north to Newcastle, south to Wollongong and inland to the continental divide has been sub-divided into five seismic source zones by Berryman and Stirling (2003). The most active zone and one which has also experienced the largest earthquakes on record commencing in 1872 is the western Sydney Basin which extends west from the Lapstone Structural Complex (LSC) under the Blue Mountains Plateau to Lithgow. It includes the Nattai catchment and area beneath Lake Burragorang. Within the zone, a cluster of earthquakes has been recorded along the Wollondilly River between Warragamba and Upper Burragorang (Figure 20). The cluster includes 16 earthquakes and 291 aftershocks recorded since 1960 (Table 9). The majority of earthquakes were between magnitudes 1–3 and occurred at mid-crustal depths, with three earthquakes recorded at 0 km depth indicating that they are probably related to coal mining (blasting). The largest earthquakes in the zone occurred at Robertson in 1961 (Magnitude 5.6) and, Burragorang in 1973 (M 5.5) (Drake 1976). The epicenter of the Burragorang earthquake was located 6 km east of , close to the Wollondilly River although the area of seismicity extended in a north-west to south-east direction for 8 km. The earthquake is thought to have resulted from movement along a steeply dipping thrust fault which extends 8–24 km below the surface within the Palaeozoic basement rocks (Mills and Fitch 1977). Indeed several authors have linked the earthquakes in the western Sydney Basin to movement along the west dipping Lapstone Fault which extends underneath the lake (Berryman and Stirling 2003; Gibson in press). Recent evidence of active faulting has been discovered in rock outcrop along a small stream which cuts through the LSC (D. Clark, pers. comm. July 2006).

Table 9. Timing, depth and magnitude of earthquakes within the south-west cluster located at upper Burragorang. Data source: Geoscience Australia Earthquake Database http://www.ga.gov.au/oracle/quake/quake_online.jsp. Date of earthquake Depth (km) Magnitude Notes 02/09/1960 10 1.7 23/06/1962 0 1.9 10/07/1962 8 3.2 09/03/1973 21 5.5 Burragorang / Picton earthquake 09/03/1973 to 23/07/1973 Max 38 Max 3.9 Aftershocks 16/11/1974 0 1.2 06/03/1975 0 1.5 18/08/1976 16 2.4 26/01/1978 17 2.3 15/01/1979 24 2.8 19/04/1979 20 2 16/03/1982 23 2.2 31/12/1982 19 2 15/02/1986 24 1.7 02/04/1987 22 2.7 06/06/1996 20 2.5 26/06/2001 10 1.9

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 26

-33

C Gosford Lithgow -33.5 Richmond

h) Hornsby

ut Penrith so

s, Warragamba Sydney e Liverpool e r g e -34 B d L

al N m i T Picton c

e Upper Burragorang d ( de tu

i Wollongong t Mittagong a L -34.5

-35

150 150.5 151 151.5 Longitude (decimal degrees, east) Figure 20. Seismic events recorded in the Sydney Basin since 1872 showing the cluster of earthquakes at Upper Burragorang (open blue circles). Landslides are indicated as follows: T – Tumbledown, N – North Nattai, L – Lacys, B – Bimlow, C – Carne. Locations of towns/cities are indicated by a red triangle. Data source: Geoscience Australia Earthquake Database http://www.ga.gov.au/oracle/quake/quake_online.jsp)

The known landslides in the Nattai and Wollondilly valleys occur within or close to the earthquake cluster suggesting a possible causal relationship (Figure 20). There is similar agreement from another landslide along Carne Creek in the to the north, which is located within a seismically active area near Lithgow. However, all of the landslides are of pre-European age whereas the earthquake record is post-European, which precludes establishing a direct link. Furthermore, the recorded earthquakes appear not to have triggered any new or additional landslide failures. Thus if a link exists, it implies that only larger magnitude earthquakes are involved i.e. greater than magnitude 5.5. This is supported by Reynolds (1976) who notes that a magnitude 5.5 earthquake is not likely to fracture hard rock at the surface, whereas a magnitude 6.5 earthquake is very likely to fracture surface rock. The strong joining patterns in the Triassic sandstones suggest that the vertical cliff faces around the incised valleys could fail under a magnitude 6.5 earthquake especially where undermined by rotational slumping in the Permian rocks.

Whilst we cannot establish a cause-effect link, it is possible that large magnitude earthquakes are important triggers of extreme erosion events in the Nattai catchment and other valleys incised within the Blue Mountains Plateau. The return interval of a magnitude 6 earthquake or greater in the western Sydney Basin zone has been calculated by Berryman and Sterling (2003) as ≥ 1200 years with the maximum probable earthquake estimated to be magnitude 8 (return of around 100,000 years) (Figure 21). Though infrequent, the threat of a large magnitude earthquake occurring within the lifetime of Lake Burragorang and the other major water supply reservoirs is still ever-present.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 27

Figure 21. Estimated return intervals of earthquakes across the Sydney Basin. The return interval of a magnitude 4, 5.5 and 7 earthquake in the western Sydney Basin zone is shown. (Data source: Berryman and Stirling, 2003).

5.5. Intrinsic thresholds

On hillslopes, intrinsic (or geomorphic) and extrinsic thresholds play a key role in determining slope morphology and stability. Thresholds are defined as the quantitative point at which an action is triggered and are usually represented by either maximum or minimum variables, such as maximum slope angle before failure. Hence changes in variables can occur gradually over time up until the quantitative point, beyond which is known as a breach or exceedence of thresholds and often results in a significant change in morphology. Extrinsic thresholds are determined by external factors which act upon hillslopes to promote changes such as rainfall, floods and earthquakes. Whereas, intrinsic thresholds are determined by the internal factors of the hillslope relating to the colluvium (sediment particle size, character, volume, packing, cohesion, rate of sediment supply), bedrock (composition, structure, jointing patterns), weathering (rates, degree, type (chemical, physical)), relief (slope angle and height) and vegetation (percentage cover, type). Hillslope properties can vary both locally and at larger scales, as well as with time, hence determining intrinsic thresholds can be notoriously difficult except in the broadest sense of hillslope stability. Hillslopes are determined as stable if long periods of time proceed where thresholds are rarely exceeded. Conversely, where many significant changes occur over short timeframes such as a succession of landslides, hillslopes may be assessed as geomorphically unstable.

Based on field observations, rock falls from cliff faces formed in the Hawkesbury sandstone appear to be the most common example of exceeding intrinsic thresholds within the Nattai catchment. Rock falls are usually localized (i.e. single block), though significant disturbance down slope can occur including damage to vegetation and rock outcrop through impact of the falling debris as seen at North Nattai in July 2005. The main process of rock fall is weathering along joints and bedding planes within the bedrock leading to collapse of individual blocks. Failure

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 28

appears to commence at the bottom of the strata, proceeding upwards (undermining effect) resulting in shear vertical walls with a jagged appearance (Figure 22a). Similar slope failures have been observed along sections with outcropping Narrabeen Group rocks. The inter-bedded nature of the strata and rapid weathering of the softer shale lenses produces steep slopes mantled by fragmented loose material with little understorey vegetation cover. In some areas such as along the western (Wollondilly) side of the Wanganderry Tablelands whole slope segments formed on thinly bedded Narrabeen Group rocks have collapsed (Figure 22b). A third example of failure through intrinsic thresholds involves rotational slumping in the Permian rocks. Once exposed, it is likely that weathering occurs along joints and fractures within the Permian bedrock that leads to failure planes and thus facilitates slippage of blocks. Where slumping occurs in the Permian bedrock that is close to the valley side walls such as at the Tumbledown Landslide, the overlying sandstone cliffs then collapse through unsupported weight. The actual extent of rotational slumping within the Nattai catchment and along the foreshores of Lake Burragorang is unknown. The Tumbledown site indicates that head scarps are a surface expression of the failure planes, but these are difficult to detect on aerial photos and topographic maps due to thick vegetation cover and coarse scale resolution.

(a) (b)

Figure 22a) Jagged walls along the Nattai valley formed through rock fall. (b) hillslope failure within Narrabeen Group rock on the western side of the Wanganderry Tableland.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 29

6. DISCUSSION

In the introduction of this report we posed three key research questions which guided our investigation of the Nattai hillslopes. The first related to characterizing the hillslopes, determining the dominant processes and identifying issues and threats to water quality and quantity. This was achieved in Section 4 using Blue Gum Creek and the Tumbledown Landslide as representative sites. The second and third questions related to the triggers of extreme erosion events, including assessing the role of wildfires compared with other triggers and, examining the relationships or coupling between triggers along with the effect on recurrence intervals and the size of events. The following discussion now addresses these questions.

6.1. Role of wildfires compared with other triggers

Wildfires are commonly thought to be a trigger of extreme erosion and at the start of this project the same belief was maintained. During the course of the project, however, as data was collected on various aspects of the 2001-02 wildfire including fire severity, soil water repellency, sediment movement, bioturbation, rainfall and runoff this view began to change. The 2001-02 wildfire appeared to trigger very little erosion within the Nattai catchment apart from transport of ash, charcoal and burnt organics from the soil surface. Though this was significant in terms of water quality, it was apparent that most of the mineral soil (i.e. derived from the colluvium) remained on the hillslopes trapped in litter dams, infilling depressions or forming a surface lag. Examination of rainfall in the post-fire period provided the answers with the characteristics of rainfall being a primary determinant of the amount and extent of rainsplash erosion, runoff and sediment transport. In addition, water repellency and bioturbation were found to play a key role in determining infiltration and storage of rainfall within the soil.

Rainfall is highly variable in the years (before and) after fire. This affects the number, size and type of rainfall events. Heavy rainfall events (> 60 mm day-1) are required to generate significant runoff and transport of sand from the hillslopes into the river network. Small to moderate rainfall events and those resulting from thunderstorms appear to produce sufficient runoff to transport the low density and finer material as floating and suspended load. It is probably this material particularly the charcoal and ash that gives the impression of extreme erosion from within burnt catchments. The likelihood of a heavy rainfall event(s) occurring in the post-fire period is determined by synoptic scale weather patterns which bear little relationship to fire or to broad scale climate dominated by ENSO. This affects the frequency of extreme erosion events arising from wildfires. In terms of return intervals, extreme erosion from wildfires could still occur but probably just not as frequently as is often believed due to rainfall variability.

The role of wildfire as a trigger of extreme erosion-sedimentation events in the Nattai catchment compared with other triggers is substantially less (Figure 23). Instead extreme rainfall appears to be the most common trigger with severe thunderstorms and heavy rainfall events resulting in widespread impacts including rainsplash erosion, slope wash, debris flows and where large enough generation of catastrophic floods. Catastrophic floods, although less frequent, transport the largest volumes of sediment. These can result in decades worth of sediment being mobilized and exported from the catchment within a few days. Earthquakes are the least frequent trigger of extreme erosion events although the effects of such an event could be more significant depending on whether landslides result and if they occur around the foreshores of Lake Burragorang. Lastly, intrinsic thresholds are probably the most poorly known and least credited trigger of extreme erosion events. Intrinsic thresholds play a fundamental role in landscape evolution over short and long timescales.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 30

Figure 23. Impact of wildfires on extreme erosion events compared with other triggers.

6.2. Coupling of triggers leading to extreme erosion-sedimentation events

The coupling of triggers refers to situations where one trigger follows another, or occurs at the same time as another (concurrently), or leads to another. In some instances an extreme erosion event will result if one of those scenarios occurs. Situations identified where the coupling of triggers leads to extreme erosion in the Nattai catchment are outlined in Table 10. The relationships and feedbacks between triggers however, are likely to be complex given the number of variables associated with a single trigger, which are then increasingly magnified when multiple triggers are involved.

The implications for extreme erosion-sedimentation events, resulting from the coupling of triggers is as follows. With respect to recurrence intervals, in most cases it is likely that the frequency of an extreme event would decrease since the chance of two triggers occurring at the same time or soon after one another is less than the frequency of each individual trigger. As outlined earlier, a good example is extreme erosion resulting from wildfires which depends on the timing of wildfire and rainfall in the post-fire period. The exception however, is where an external trigger is coupled with intrinsic thresholds. In these cases, it is likely that extreme erosion events would increase especially on hillslopes that are at or close to critical limits and hence are susceptible to the smallest change imposed by an external trigger. A period of hillslope instability may arise when hillslopes close to thresholds are subject to frequent external triggers, including those which may usually have a lesser effect. The magnitude of extreme erosion events are also likely to be affected by the coupling of triggers. Events may be larger or have a greater impact particularly in terms of sediment supply and transport. Extreme rainfall events coupled with catastrophic floods are the best examples of this. The rainfall facilitates sediment movement on hillslopes into the river network whilst the floods transport the material downstream exporting it from the catchment.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 31

Table 10. Coupling of triggers leading to extreme erosion events in the Nattai catchment 1. Wildfire followed by extreme rainfall (thunderstorms) Extreme erosion Rainsplash and slope wash Examples Eight high intensity thunderstorms within the first three years following the 2001-02 wildfires, with the first on 11th April 2002. Role of each • Wildfire reduces the protective vegetation cover leading to exposure of trigger and the soil. erosional effects • High intensity thunderstorms generate raindrops with high kinetic energy which upon impact results in significant rainsplash erosion, formation of soil pedestals and generation of overland flow.

2. Wildfire followed by extreme (heavy) rainfall and floods Extreme erosion Debris flows and reworking of colluvial material Examples 1. November 1968 wildfire followed by April 1969 tropical cyclone and 1969 catastrophic flood 2. November 1997 wildfire followed by August 1998 ECC and 1998 flood Role of each • Wildfire increases runoff rates from burnt hillslopes during heavy rainfall trigger and events leading to increased peak discharge of floods and increased erosional effects sediment transport capacity on slopes and in streams.

3. Extreme (heavy) rainfall and catastrophic floods Extreme erosion Debris flows and reworking of colluvial material Examples 1. 1867 ECC and 1867 catastrophic flood 2. 1978 ECC and 1978 catastrophic flood 3. 1986 ECC and 1986 flood Role of each • Heavy rainfall facilitates transport of material from hillslopes via slope trigger and wash and debris flows in the major streams. erosional effects • Catastrophic floods then transport and export this material from the catchment.

4. Extreme (heavy) rainfall and breach of intrinsic thresholds Extreme erosion Slope failure and debris flows Examples Gillans Creek debris flow Role of each • Heavy rainfall saturates thick soil/colluvial deposits or weak fractured trigger and bedrock, increasing pore-water pressure and decreasing shear strength. erosional effects • Failure of the slope occurs as a slurry of material or debris flow.

5. Catastrophic floods and breach of intrinsic thresholds Extreme erosion Rotational slumping within the valley floor Examples Rotational slumping at the Tumbledown Landslide site Role of each • Failure planes develop within fractured rock outcropping in the valley trigger and floor. erosional effects • Removal of material at the toe during floods facilitates slumping along the failure planes.

6. Earthquakes and intrinsic thresholds Extreme erosion Rock fall and landslides Examples No specific examples as recent earthquakes have been minor. Role of each • Weathering of bedrock occurs along joints and bedding planes. trigger and • Large magnitude earthquakes cause ground movements which dislodges erosional effects weathered and unsupported bedrock.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 32

7. CONCLUSIONS

Hillslopes in the Nattai catchment are widely blanketed by thick colluvial mantles formed through extreme erosion-sedimentation events including rock fall, debris flows and landslides during the Holocene. The major controls on hillslope process and morphology appears to be geology and river incision. Where rivers have incised into the weaker Permian rocks, valley widening proceeds via failure of hillslopes through rotational slumping within the exposed Permian bedrock and rock avalanching resulting from collapse of the vertical sandstone cliffs above. Debris flows play a key role in transporting coarse material from the plateaux and hillslopes into the stream network.

There are several triggers of extreme erosion events in the Nattai catchment including severe wildfire, extreme rainfall, catastrophic floods, earthquakes and intrinsic thresholds. Extreme erosion can result from any of three scenarios; (i) a single trigger which occurs in isolation, (ii) multiple triggers, where more than one trigger leads to extreme erosion, and (iii) the coupling of triggers, where one follows another, or occurs at the same time as another, or leads to another.

Despite causing significant damage to vegetation, wildfire alone does not lead to extreme erosion. Instead it is the coupling of fire with rainfall (and runoff) in the post-fire period that determines the nature and extent of sediment entrainment and transport. Large rainfall events following fire are likely to trigger substantial erosion of the coarser sandy sediment whereas minor rainfall events appear sufficient to transport floating and suspended material such as ash, charcoal, leaf litter and clay resulting in temporary increases in river sediment yield. Rainfall is highly variable across southeastern Australia, which means that erosion in the post-fire period will also vary becoming less likely with time after fire. Hence predictions of erosion after wildfire are difficult and uncertain. The role of fire in facilitating extreme erosion events is also somewhat less than is commonly perceived.

The most significant type of extreme rainfall in the Nattai catchment is long duration, continuous, medium-high intensity (heavy) rainfall where > 60 mm day-1 falls across the catchment. These events are commonly triggered by east coast cyclones and tropical cyclones which move down the east coast of Australia. Both can result in sustained runoff from hillslopes, significant sediment transport and major flooding in the river systems. Less significant is rainfall triggered by severe thunderstorms, although rainfall totals, runoff and the extent of erosion is determined by the intensity, location, speed, size and direction of storm cells.

Catastrophic floods are major triggers of extreme erosion events, mobilizing and exporting decades equivalent of sediment yield from catchments within a few days. The major cause of catastrophic flooding is heavy rainfall over one or more days leading to significant catchment- scale runoff. Since 1965, there have been three catastrophic floods in the Nattai catchment. The longer Hawkesbury-Nepean River record however, shows multi-decadal oscillations in flood frequency and magnitude indicating that there are periods when extreme erosion triggered by catastrophic floods has been more frequent.

High magnitude earthquakes (> 6) are a likely trigger of landslides preserved along the lower Nattai River and around the foreshores of Lake Burragorang, although the disparity in ages precludes establishing a direct cause-effect link. The area beneath the Blue Mountains Plateau is the most seismically active in the greater Sydney Basin with earthquakes clustering in a region to the south-west centered upon Upper Burragorang. High magnitude earthquakes are relatively infrequent compared with other triggers however, the impacts on water quality and quantity from an earthquake-triggered landslide around the foreshores of the lake are likely to be significantly, if not profoundly greater than other extreme erosion events.

Intrinsic thresholds are a primary control on hillslope stability. Breach of intrinsic thresholds results in rock fall, rock avalanching and rotational slumping within Permian bedrock as well as failures in weathered Narrabeen Group rocks. In contrast with external triggers, which are mostly event driven, breach of intrinsic thresholds occurs through gradual changes in hillslopes up until a quantitative point. Often multiple failures or effects result through positive feedbacks.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 33

8. ACKNOWLEDGEMENTS

We thank the following people for their assistance or input to the project: – Macquarie University undergraduate students (Heather Skeen, Scott Kirk, Anne Ashworth, Clay Finnemore, Scott Reyes, Chris Ives, Warren Lal, Rod Winton, David McConville) for assistance with field work and data collection. – James Ray and Chris Chafer from the SCA for access to the Nattai catchment rainfall, discharge, turbidity, suspended sediment and fire records. – Alan Seed from the Bureau of Meteorology for the Mapview program and Kurnell radar data (processed by the Sydney Water Corporation and BoM). – Staff at the BoM Climate Services Centre for information relating to historical weather patterns. – Kelvin Berryman for access to data on earthquake recurrence intervals within the Sydney Basin. – Dan Clark for useful discussions on seismic activity and photographs of faulting along the Lapstone Structural Complex.

We also thank the following colleagues for their direct contribution to the various themes: Anna Gero and Paul Hesse (Macquarie University), Rick Shakesby and Stefan Doerr (University of Wales Swansea), Will Blake (University of Plymouth), and Peter Wallbrink (CSIRO).

9. REFERENCES

Berryman, K., and Stirling, M., 2003, Earthquake ground motion and fault hazard studies at the Lucas Heights Research Reactor Facility, Sydney, Australia: evolution of methods and changes in results.: International Symposium on Seismic Evaluation of Existing Nuclear Facilities, Vienna, August 2003. p. 43-47. Bracewell, M. G., and McDermott, G. E., 1985, Report on the 1867 flood, Metropolitan Water Sewerage and Drainage Board, Sydney. Bureau of Meteorology, 2006a, Australian Regional MSLP Analysis Chart Archives, available at http://www.bom.gov.au/nmoc/MSL/index.shtml. Date accessed 26 September 2006. —, 2006b, Climatic averages for NSW and ACT sites, available at http://www.bom.gov.au/climate/averages/tables/ca_nsw_names.shtml, date accessed 12 October 2006. —, 2007, Australian rainfall patterns during La Niña events, available at http://www.bom.gov.au/climate/enso/ninacomp.html, date accessed 6 March 2007. Cannon, S. H., Bigio, E. R., and Mine, E., 2001, A process for fire-related debris flow initiation, Cerro Grande fire, New Mexico: Hydrological Processes, v. 15, p. 3011-3023. DOI: 10.1002/hyp.388. Chafer, C. J., Noonan, M., and Macnaught, E., 2004, The post-fire measurement of fire severity and intensity in the Christmas 2001 Sydney wildfires: International Journal of Wildland Fire, v. 13, p. 227-240. DOI: 1071/WF03041. Chiew, F. H. S., Piechota, T. C., Dracup, J. A., and McMahon, T. A., 1998, El Nino/Southern Oscillation and Australian rainfall, streamflow and drought: Links and potential for forecasting: Journal of Hydrology, v. 204, p. 138-149. Cunningham, D. M., 1988, A rockfall avalanche in a sandstone landscape, Nattai North NSW.: Australian Geographer, v. 19, p. 221-229. Dadson, S. J., Hovius, N., Chen, H., Dade, W. B., Hsieh, M.-L., Willett, S. D., Hu, J.-C., Horng, M.-J., Chen, M.-C., Stark, C. P., Lague, D., and Lin, J.-C., 2003, Links between erosion, runoff variability and seismicity in the Taiwan orogen: Nature, v. 426, p. 648-651.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 34

Dai, F. C., Lee, C. F., Deng, J. H., and Tham, L. G., 2005, The 1786 earthquake-triggered landslide dam and subsequent dam-break flood on the Dadu River, southwestern China: Geomorphology, v. 65, p. 205-221. Drake, L., 1976, Seismic risk in : Bulletin - Australia, Bureau of Mineral Resources, Geology and Geophysics, v. 164, p. 5-8. English, P., Wallbrink, P., Chafer, C., Humphreys, G., Shakesby, R., Blake, W., Doerr, S., and Vigneswaran, B., 2005, Report 2: Tracer assessment of post-fire sediment and nutrient redistribution on hillslopes: , NSW, CSIRO Land and Water Client Report. Environment ACT, 2004, ACT Water Report 2003-2004, Canberra, Environment ACT. Erskine, W. D., and Saynor, M. J., 1996, Effects of catastrophic floods on sediment yields in southeastern Australia: Erosion and Sediment Yield: Global and Regional Perspectives. IAHS Publication 236, p. 381-388. Erskine, W. D., and Warner, R. F., 1988, Geomorphic effects of alternating flood- and drought- dominated regimes on NSW coastal rivers, in Warner, R. F., ed., Fluvial Geomorphology of Australia, Sydney, Academic Press, p. 223-244. Fisher, M., Ryan, K., and Lembit, R., 1995, The natural vegetation of the Burragorang 1:100 000 map sheet: Cunninghamia, v. 4, p. 143-215. Folk, R. L., 1974, The petrology of sedimentary rocks: Austin, Texas, Hemphill Publishing Co., 182 p. Gibson, G., in press, Seismicity of the Sydney Basin Region: Geoscience Australia Journal, p. In press. Henderson, L., 2002, Soil Landscapes in the hydrological catchments of the Sydney Catchment Authority: Burragorang and Wollongong Map Sheets: Department of Infrastructure, Planning and Natural Resources. Hess, G. D., 1990, Numerical simulation of the August 1986 heavy rainfall event in the Sydney area: Journal of Geophysical Research, v. 95, p. 2073-2082. Holland, G. J., Lynch, A. H., and Leslie, L. M., 1987, Australian East-Coast Cyclones. Part 1: Synoptic overview and case study: Monthly Weather Review, v. 115, p. 3024-3036. Hopkins, L. C., and Holland, G. J., 1997, Australian heavy-rain days and associated east coast cyclones: 1958-92: American Meteorological Society, v. 10, p. 621-635. Keefer, D. K., 1984, Landslides caused by earthquakes: Geological Society of America Bulletin, v. 95, p. 406-421. —, 1994, The importance of earthquake-induced landslides to long-term slope erosion and slope- failure hazards in seismically active regions: Geomorphology, v. 10, p. 265-284. —, 2002, Investigating landslides caused by earthquakes - a historical review: Surveys in Geophysics, v. 23, p. 473-510. King, D. P., 1994, Soil Landscapes of the Katoomba 1:100 000 Sheet: Department of Conservation and Land Management,. Korup, O., 2005, Geomorphic hazard assessment of landslide dams in South Westland, New Zealand; fundamental problems and approaches: Geomorphology, v. 66, p. 167-188. Liu, J. G., Mason, P. J., Clerici, N., Chen, S., Davis, A., Miao, F., Deng, H., and Liang, L., 2004, Landslide hazard assessment in the Three Gorges area of the Yangtze river using ASTER imagery: Zigui-Badong: Geomorphology, v. 61, p. 171-187. doi:10.1016/j.geomorph.2003.12.004. Lynch, A. H., 1987, Australian east coast cyclones: case study of the storm of August 1986: Australian Meteorological Magazine, v. 35, p. 163-170. McDonald, R. C., Isbell, R. F., Speight, J. G., Walker, J., and Hopkins, M. S., 1990, Australian Soil and Land Survey Field Handbook: , Australia, Inkata Press, 198 p. McElroy, C. T., and Relph, R. E., 1958, Explanatory notes to accompany geological maps of the inner catchment, Warragamba storage: Department of Mines, Geological Survey of NSW.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 35

Mills, J. M., and Fitch, T. J., 1977, Thrust faulting and crust - upper mantle structure in East Australia: Geophysical Journal of the Royal Astronomical Society, v. 48, p. 351-384. Moody, J. A., and Martin, D. A., 2001, Initial hydrologic and geomorphic response following a wildfire in the Colorado Front Range: Earth Surface Processes and Landforms, v. 26, p. 1049-1070. DOI: 10.1002/esp.253. Northcote, K. H., 1979, A factual key for the recognition of Australian soils: Adelaide., Rellim Technical Publications. NSW State Emergency Service, 2005, Hawkesbury Nepean Flood emergency sub plan (HNFESP), Sydney, NSW State Emergency Service, p. 131. Panizzo, A., De Girolamo, P., Di Risio, M., Maistri, A., and Petaccia, A., 2005, Great landslide events in Italian artificial reservoirs: Natural Hazards and Earth System Sciences, v. 5, p. 733-740. Pells, P. J. N., Braybrooke, J. C., Mong, J., and Kotze, G. P., 1987, Cliff line collapse associated with mining activities, in Walker, B. F., and Fell, R., eds., Soil Slope Instability and Stabilisation, Rotterdam, Balkema, p. 359-385. Prosser, I. P., and Williams, L., 1998, The effect of wildfire on runoff and erosion in native Eucalyptus forest: Hydrological Processes, v. 12, p. 251-265. Ray, H., 2003, New South Wales 1:250 000 Statewide Geology: Sydney Region: NSW Department of Mineral Resources,. Reynolds, R. G., 1976, Coal mining under stored water: report on an inquiry into coal mining under or in the vicinity of the stored waters of the Nepean, Avon, Cordeaux, Cataract and Woronora Reservoirs, New South Wales, Australia: Department of Public Works. Riley, S. J., 1980, Aspects of the flood record at Windsor: Papers of the Newcastle (16th) conference of the Institute of Australian Geographers., University of Newcastle. p. 325- 339. Robichaud, P. R., 2000, Fire effects on infiltration rates after prescribed fire in Northern Rocky Mountain forests, USA: Journal of Hydrology, v. 231-232, p. 220-229. Rose, G., 1966, Wollongong 1:250 000 Geological Series Sheet SI 56-9: NSW Department of Mines. Shakesby, R. A., Blake, W. H., Doerr, S. H., Humphreys, G. S., Wallbrink, P. J., and Chafer, C. J., 2006, Hillslope soil erosion and bioturbation after the Christmas 2001 forest fires near Sydney, Australia, in Owens, P. N., and Collins, A. J., eds., Soil Erosion and Sediment Redistribution in River Catchments, Wallingford UK, CAB International, p. 51-61. Shakesby, R. A., Chafer, C. J., Doerr, S. H., Blake, W. H., Wallbrink, P., Humphreys, G. S., and Harrington, B. A., 2003, Fire severity, water repellency characteristics and hydrogeomorphological changes following the Christmas 2001 Sydney forest fires: Australian Geographer, v. 34, p. 147-175. DOI: 10.1080/0004918032000108446. Summerfield, M. A., 1991, Global Geomorphology: Singapore, Longman Publishers, 537 p. Tomkins, K. M., Humphreys, G. S., Shakesby, R. A., Doerr, S. H., Blake, W. H., and Wallbrink, P., 2004a, Mass movement events in the south-west Sydney Basin during the Holocene: Regolith 2004., CRC LEME. Canberra. p. 365-369. Tomkins, K. M., Humphreys, G. S., Skeen, H. J., Taylor, G. M., Farwig, V. J., Shakesby, R. A., Doerr, S. H., Wallbrink, P., Blake, W. H., and Chafer, C. J., 2004b, Deciphering a colluvial mantle: Nattai catchment: SuperSoil 2004: Program and Abstracts for the 3rd Australian and New Zealand Soils Conference, University of Sydney, Australia. p. www.regional.org.au/au/asssi/supersoil2004. Walker, P. H., 1964, Sedimentary properties and processes on a sandstone hillside: Journal of Sedimentary Petrology, v. 34, p. 328-334. Wallbrink, P., English, P., Chafer, C., Humphreys, G., Shakesby, R., Blake, W., and Doerr, S., 2004, Report 1: A review of literature pertaining to the effect of fire on erosion and erosion rates, with emphasis on the Nattai catchment, NSW following the 2001 bushfires, CSIRO Land and Water Client Report.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 36

Webb, R. H., Fenton, C., and Cerling, T., 2003, Late Quaternary debris-flow aggradation of the Colorado River in Cataract Canyon, USA: XVI INQUA congress, Shaping the Earth: a quaternary perspective, Reno, NV, United States, July 23-30, 2003. p. 63-64. Wilkinson, M. T., Chappell, J., Humphreys, G. S., Fifield, K., Smith, B., and Hesse, P. P., 2005, Soil production in heath and forest, Blue Mountains, Australia: influence of lithology and palaeoclimate: Earth Surface Processes and Landforms, v. 30, p. 923-934. DOI: 10.1002/esp.1245. Erratum 30: 1683-1685. DOI: 10.1002/esp.1311. Wilkinson, S., Wallbrink, P., Humphreys, G., Blake, W., Shakesby, R., and Doerr, S., 2006, Report 3: Post-fire sediment and nutrient redistribution to downstream waterbodies, Nattai National Park, NSW, CSIRO Land and Water Client Report. Williams, M. A. J., 1972, The influence of slope, soil and plant cover on runoff and erosion in the upper Shoalhaven area, 1966-1968: Journal of the Soil Conservation Service of NSW, v. 28, p. 51-63. —, 1978, Late Holocene hillslope mantles and stream aggradation in the , NSW.: Search, v. 9, p. 96-97. Winter, J., and Watts, M., 2002, Fire reports 11 Wollondilly: Bushfire Bulletin, Journal of the NSW Rural Fire Service, v. Commemorative Edition Christmas Fires 2001, p. 92-96.

Technical Report 3: Triggers of extreme erosion events on hillslopes in the Nattai catchment 37