Landscape Processes of Moonaree Station: a Pilot Study Gresley Wakelin-King DISCLAIMER

Government of South Australia South Australian Arid Lands Natural Resources Management Board

July 2009 South Australian Arid Lands Natural Resources Management Board Landscape processes of Moonaree Station: a pilot study Gresley Wakelin-King DISCLAIMER

The South Australian Arid Lands Natural Resources Management Board, and its employees do not warrant or make any representation regarding the use, or results of use of the information contained herein as to its correctness, accuracy, reliability, currency or otherwise. The South Australian Arid Lands Natural Resources Management Board and its employees expressly disclaim all liability or responsibility to any person using the information or advice.

LANDSCAPE PROCESSES OF MOONAREE STATION: A PILOT STUDY

GRESLEY WAKELIN-KING Wakelin Associates Pty. Ltd. Geology – GIS – Geomorphology

July 2009 South Australian Arid Lands Natural Resources Management Board

This work is copyright. Apart from any use permitted under the Copyright Act 1968 (Commonwealth), no part may be reproduced by any process without prior written permission obtained from the South Australian Arid Lands Natural Resources Management Board. Requests and enquiries concerning reproduction and rights should be directed to the General Manager, South Australian Arid Lands Natural Resources Management Board Railway Station Building, PO Box 2227, Port Augusta, SA, 5700

Landscape Processes of Moonaree Station: A Pilot Study

Table of Contents

Summary 5

Part 1 – Introduction 7

Background 7

Aim, Output, Outcome 8

Study Area: Moonaree Station 8

Methods 11

Results: Assessment of Methods and Resources 14

Part 2 – Report to Landholders 25

Geology 25

Geomorphology and Landscape Process 39

Implications for Land Management: Summary 57

Part 3: Controls on Surface Water in SA Arid Lands 59

Low Priority Given to Understanding Drylands Catchments 59

Different Features of Drylands Rivers 61

Human Activities that Affect River Flow 64

References 67

Tables 1 Comparison of cost and utility of images for geomorphic mapping. 24

Figures 1 Location of Moonaree Station. 10

2 Landform-process units and field sites. 13

3 Narrow-scale comparison of air-photo, photomosaic, LANDSAT, and DEM images for a small valley. 17

4 Broad-scale comparison of photomosaic, LANDSAT, ALOS, and DEM images for a wide area. 19, 20, 21

5 Comparison of in-house photomosaics created from 800dpi scans of existing air photos. 22

6 The Yardea Dacite. 25

7 Geological structure at Moonaree, shown by fracture-valley and hilltop trends. 26

8 Digital Elevation Model of the Gawler Ranges, showing the Acraman Impact Structure and the regional fracture pattern. 27

9 Fracture spacing influences hillslope weathering and gnamma formation. 28

10 Observed outcrops of white claystone and weathering profile. 29

11 Near Gorge Creek, white claystone, mottled with weathering profile overprint. 30

12 Tertiary rocks and weathering profile. 32

13 Calcrete and silcrete. 33

14 Calcrete distribution at Moonaree. 34

15 Modern soil and sand, old and modern gypsum. 35

16 Distribution of gypsum and gilgai at Moonaree. 36

17 Elevation profiles of the southern sand sheet. 37, 38

18 Landscape elements. 39

19 A discontinuous drainage gutter in a rocky fracture-valley. 40

20 Chenopod plains in Belt Hill and Bond Paddocks. 41

21 Looking across the drainage axis of a chenopod plain in Mt Cooper paddock. 42

22 Bare, sheet-eroded patches along the stock route at Kallinta. 43

23 A swale with trees; a sketch of a line of discontinuous creeks. 44

24 A discontinuous creek in Mt Cooper Paddock. 45

25 Channel incision in Yeltabinna Creek. 46

26 Catchments of the Moonaree creeks. 47

27 Gilgai pits. 49

28 One of Moonaree’s largest gilgai pits. 50

29 Gilgai in the bare ground around Cornish Well. 52

30 Downslope sediment transport in the rocky hilltops of Parkers. 53

31 A stock pad may concentrate runoff and promote erosion. 54

32 A deep gully system in Spearfelt traces upslope to a station track. 55

33 Gullying along station tracks. 56

34 Less accuracy in understanding remote catchments. 60

35 A gully formed along a station track in NSW. 65

Maps (A3 size) 1 Paddocks and major creeks. End pages

2 Station tracks and watering points (Landholders and SAAL-NRM only). End pages

3 Paddock boundaries over the LANDSAT image. End pages

4 Moonaree Digital Elevation Model as if lake-full to 200m ASL. End pages

LANDSCAPE PROCESSES OF MOONAREE STATION: A PILOT STUDY

Summary This pilot study aimed to map landscape processes (geomorphology) on a grazing property, to gain new information contributing to sustainable rangelands management. Moonaree Station (Alastair and Catherine McTaggart), in the Gawler Ranges, was the chosen site. The method was an initial desktop and remote- sensing study followed by field work and consultation with the landholders. The project began in early February, field work was undertaken in mid-June, preliminary results verbally presented to the landholders in late June, and project documentation was completed to first draft by end August. The method worked well, however affordable satellite images are not particularly suitable for geomorphic investigation of this type, and in South Australia aerial photography is difficult to access. Timing of field work proved to be a difficulty, as the landholders could not engage with the project until after shearing, whereas the project had to be acquitted before the end of the financial year; the time in the field and the post-field reporting were thus shortened.

Method recommendations for subsequent work of this kind are 1) that an arrangement be made with SA-DEH for access to stereo pairs of aerial photography; 2) that LANDSAT and ALOS be used for base map construction; 3) that strong representations be made to State and Federal policymakers that range managers are increasingly required to manage for whole-country goals (e.g. biodiversity), yet a fundamental source of information (aerial photography) is no longer being affordably supplied by government; 4) that project funding not be tied to the financial year; 6) that extra time should be allocated (as non-project time interval: at no extra project cost) between draft and final report to give landholders opportunity to comment.

Moonaree Station's landscape is high rounded hills of Yardea Dacite separated by wide straight fracture-valleys, with playa lakes to the east and west. The geological history is as follows: emplacement and later fracturing of volcanic rock (<1,600 and <580 million years ago), partial submergence of the rocky hills by Lakes Gairdner and Acraman, deposition of fluvial sediments in valleys, and deep weathering (~65-2

million years ago), and during the last ~6 million years, sand transport, soil formation, and repeated episodes of valley sediment deposition and duricrust (calcrete and silcrete) formation. The most recent events have been smaller increments of rising and falling lake water levels (>2 million years), the arrival of Aboriginal people at ~40,000 years, and the commencement of European-style stock grazing in 1862.

Dominant landscape processes include: links between soil type and underlying geology, differing degrees of runoff (rocky hilltops versus upper hillslope), the importance of intense localised rainfall as landscape formation element, discontinuous ephemeral stream processes in many of the valleys, channelised streams in the longer valleys with bigger catchments, spatially discontinuous development of gilgai soils, and landscape control of the groundwater amount and salinity.

Implications for land management include: strong links between geology and landscape productivity, particularly with respect to gilgai areas; gully erosion along tracks; strong links between landscape productivity and the discontinuous ephemeral stream landforms; grazing management and vegetation preservation as an important factor in landscape maintenance; the random occurrence of rain-induced erosion patches; the dominance of local subsurface conditions in the success (or otherwise) of drilling for groundwater; the need for good information on track creation and management; and the need for effective techniques of rangeland rehabilitation along the old stock route.

This project demonstrates the strong links between landform and ecology, and the diversity of landscape types in the Arid Lands. The information that came out of this project has given the landholder new information about water quality issues, extra- productive “sweet spots” in the landscape, gullying, rangeland rehabilitation, and the relationships between grazing history, river type, and chenopod plain productivity. Issues highlighted for further consideration generally in the SA Arid Lands include 1) identification of land in near-original condition, 2) the need for low-impact track creation and maintenance techniques, 3) links between palaeochannels and gilgai, and 4) the potential loss of a valuable information source with reduced access to aerial photography for land managers. Future projects of this type in different types of country will add to the known landform-ecology relationships and expand the

information which is specifically targeted to land management. As a general principle, ecology starts with land and water, and geomorphology – landscape process – is fundamental to both.

This report concludes with some general observations on management of surface water (including groundwater extracted from alluvium) use in South Australia. At present there is little legislative framework governing the use of surface waters in the SA Arid Lands. Control of surface waters generally aims to ensure all users (including the environment) have reasonable use of the resource. It is important to recognise that dryland rivers and drainages do not necessarily operate in the same way as perennial rivers, nor necessarily look like them: management policies should be developed from understanding South Australia's creeks, rather than derived from policies developed for temperate-zone rivers. Small discontinuous and/or ephemeral creeks, such as are the dominant creek type on Moonaree, could be regulated by limiting the amount of water captured in lower-order tributaries. Effective and fair regulation is not straightforward. Further consideration should be given to characterisation of the Arid Land’s waters. The ultimate aim of such regulation must be defined: is it to preserve the status quo, to work towards some approximation of pre-European function, or to balance ecological and economic requirements (which in their turn must then be defined).

Part 1: Introduction

Background Plants and animals do not live in isolation from their physical surroundings, and knowledge of an area’s landscape processes (the geological science of geomorphology) should be a key component of any rangeland management plan. Despite this, landscape is often dealt with descriptively, and its processes are not well understood; rangeland monitoring tends to focus on plant communities rather than effective landscape processes.

In the rangelands the most important physical element is water. Drylands rivers are water’s most obvious manifestation, but they are not the only pathways of water transport, nor do they behave in the same way as “normal” (perennial) rivers. Recent

innovators in rangeland management (Peter Andrews of Natural Sequence Farming; Bob Purvis of Atartinga Station; Hugh Pringle and Ken Tinley of the Ecosystem Management Project) have used drylands river processes to improve the condition of their lands. Similar projects need to be undertaken in a range of landscapes so that general principles can be extracted from landscape-specific solutions.

Aim, Output, Outcome This project aims to map and document landscape processes in a property in the SAAL-NRM area. A suitable property would:

• be a sheep or cattle grazing property

• be without permanent natural surface water

• have a mixture of landscape types

• be available to be visited by the researcher for field work

• have a land manager or other knowledgeable person available for occasional discussions with the researcher during field work;

• and ideally have a land manager or other knowledgeable person available to provide some background information to the researcher at the project start.

The outputs are some maps, a GIS dataset, report to SAAL-NRM, report to the landholder, and an industry newsletter report. The outcome is information contributing to management plans and land condition monitoring on the property. The project was completed by mid-2009.

Study Area: Moonaree Station Moonaree Station is in the Gawler Ranges, occupying the whole land area between Lakes Acraman and Gairdner (Fig. 1), an area managed in pre-European times by the Kokata and Wirangu peoples (Tindale 1974). The station is reached by graded unsealed public road from Iron Knob. The public road continues north to Kingoonya on the Trans-Australian Railway, but there are no other public roads on the property. The nearest fuel is at Mt. Ive station, an hour’s drive south of the Moonaree boundary. Moonaree is currently held by Alistair and Catherine McTaggart. The property grazes sheep and a few cattle, and mustering is mostly by motorbike. Paddock names and major creek names are shown in Map 1, and watering points (dam and bore) and station tracks are shown in Map 2 (Landholders’ version only).

Early European exploration in the Moonaree region included Steven Hack in 1857 and Major Warburton in 1858, and the Moonaree Pastoral Lease was taken up in about 1862 (Blisset, 1985). The McTaggart family settled at Nonning Station in the Gawler Ranges in the late 1800s, and the Nonning Pastoral Company acquired Moonaree sometime before 1923 (Anderson 1995). The Nonning Pastoral Co was at one time Australia’s largest pastoral company. In the 1980s company broke up and Moonaree passed to Mr. McTaggart. Moonaree Station lies on a road extending from Iron Knob in the south to the Trans-Australian railway line to the north (near Coondambo Station, Fig. 1), but there are not known to have been inter-property stock routes through the station.

Moonaree Station is 2,466 square km in area, its perimeter is 426km, and it extends a maximum of 57km east to west and 65km north to south. It is bounded to the east and west by playa lakes, and to the north and south by belts of sandhills. Current homesteads include Old Moonaree to the south of the property and Kangaroo Well closer to the northern boundary. There are also a homestead paddock at Mt Harper Well and a crutching shed at Belt Hill/Yeltabinna. According to Mr. McTaggart, all the station shearing used to be at Old Moonaree, so stock were droved there from the distant parts of the station along the main road (from Mt Cooper Dam to Beviss Well and into the old homestead area). Beviss, Ram and Willigenda paddocks, part of the shearing area, have had hard usage, as have Kallinta, Kallinta South, and Waurea paddocks. Kallinta, Kallinta South, and Ram are management focus paddocks and have had very low stocking levels for the last 20 years (A. McTaggart, pers. comm., 2009). Some range rehabilitation (contour ripping) has been done by Mr McTaggart in overgrazed areas along the main road in Beviss and Kallinta paddocks.

Moonaree Station is watered by surface dams and by wells, some of which supply distant watering points. Some of the wells were created by drilling, but a number of historical wells were dug by hand through surface alluvium and into rock. The McTaggarts report that the character of the wells and dams is quite variable. Some wells had good water for a while that then went salty or reduced in quantity; dams may be brackish or fresh, clear or muddy. Most water is fairly brackish. Finding water on the property is quite a challenge as it is not clear where to drill to get good quality water.

Fig. 1 Location, Moonaree Station, showing public roads and neighbouring properties. Playa lakes in blue, study area (Moonaree Station) in yellow.

Moonaree is not overly troubled by feral pests. Foxes, rabbits, and feral plants (African boxthorn, and Wards Weed) were reduced during drought, though rabbits and foxes are now slowly increasing in numbers. There are a few dingos or wild dogs getting through the local dog fence. There are no pigs or wild horses. There are a few (300-400) goats; not many, in comparison to the rest of the Gawler Ranges. Mrs. McTaggart says the goats don’t like to come over the sandhills along the northern and southern property boundaries, and only a few come in by way of the salt lake islands.

Moonaree weather history (A. & C. McTaggart pers. comm. 2009) includes drought during the 1980s, and big floods in 1997 when the lakes filled after several hundred mm of rain (12 inches of rain in 4 hours). The creeks ran hard, washing away fences and windmills, and Old Moonaree was completely flooded. There was a lot of damage, especially in the south of the property. Yeltabinna Creek ran 15 feet deep, and Garden Well and Jolly Creeks ran very high. A lot of erosion gullies were created. A new cohort of Acacia trees has germinated in the creek lines; there weren’t so many before the floods.

Other station issues include erosion along some old tracks, and the McTaggarts’ curiosity as to the causes of differences between paddocks on the property: Little Gorge is one of the most productive, Belt Hill has a lot of stock deaths, and a whole belt along the west has lots of flies, and the country is soft, good in a good year but bad in a bad year (Dingo Hill, Kulgulya, Waurea, Bond Hill, Kallinta and South Kallinta).

Methods The project hinges around the collection of field information, with a pre-field component of stakeholder engagement and desktop information studies, and post- field data analysis and report compilation.

Project aims and objectives (above) had been defined prior to project beginning. The SAAL-NRM Water Projects Officer indicated a region of interest (the Gawler Ranges),and the SAAL-NRM field officer identified two potentially suitable properties. The properties were contacted, and Moonaree Station indicated willingness to be involved. Preliminary discussions were held regarding the nature of the Station and suitable time for fieldwork. As part of stakeholder engagement, phone discussions were held with the field officer, and materials were prepared for presentation at the Gawler Ranges Landcare meeting and the SAAL-NRM Arid Lands Muster.

Desktop information studies commenced with locating and obtaining paper copies of the 1:250,000 scale topographic and geological maps (GAIRDNER SH53-15 and YARDEA SI53-3), and obtaining paper or digital copies of relevant references (Anderson 1995, Blissett 1985, Dickinson 1942, Williams & Gostin 2005, etc.).

The most important part of pre-field studies is landform mapping. Digital Elevation Model data was obtained from the NASA SRTM (Shuttle Radar Topography Mission) web site, and modelled using Global Mapper. Digital topographic data sets were obtained from South Australian and Commonwealth departments, and a GIS database collated (including resolution of datum issues: the SA government uses a slightly unusual datum). Aerial photography was identified at the South Australian Department of Lands. The original intention was to purchase the relevant aerial photographs for examination with the high-quality in-house stereoscope, however South Australia no longer possesses a photographic laboratory. An arrangement was made to visit Mapland in Adelaide and examine photographs from their library in their office. Preliminary landform units were identified and traced on transparent air photo overlays. This information was then transferred to a base map. A number of base map options were examined, including Quickbird, LANDSAT, ALOS, SPOT, purchased orthorectified aerial photography, in-house orthorectified aerial photography, and topographic dataset. Ultimately a combined GIS dataset, comprising pan-sharpened LANDSAT, SRTM DEM, and topographic data was used.

The preliminary landform units (Fig. 2), with their provisional process assessment, were compiled into maps and printed for use in the field. Fieldwork took place for two weeks in early June. During fieldwork, the researcher visited sites identified from the air-photo mapping (Fig. 2). Some sites were single-location examinations, and some were traverses of up to 4 km. Geology, geomorphology, and some aspects of vegetation were recorded descriptively and by photograph, and sites were located by autonomous GPS. (Note that this study did not attempt to identify vegetation types, however the general assessment was made of species diversity, population density, and individual plant health.)

During fieldwork, stakeholder engagement took place: seeking information from the local landholders about their property and what information might be useful to them, discussion with neighbouring landholders, and a close-of-visit presentation to the local landholders of information derived from mapping.

Figure 2. Landform-process units identified by air-photo and satellite image interpretation prior to field work (areas of colour), and sites examined in the field (red circles). The scope of the project did not permit sufficient field work to turn this interpretive map into an actual map, so no description of the units is presented here. This map was used to guide field work and is not meant to be a representation of the results.

Post-field analysis comprised recompilation of the GIS dataset, including field site locations, correlation between field observations and map elements, and data management; revision of the geomorphology map and process assessment in light of the field data; and documentation of the observed geomorphology and its implications for land management.

This report consists of three parts: 1, the introductory material and (because this is a pilot study) an assessment of the methods and resources; 2, analysis of the field area; and 3, arising from this study and at the request of SAAL-NRM, a discussion on surface water regulation in SA. The landholders’ report consists of the introductory material plus Part 2 (pages 1-13, 25-58, and all four maps).

Results: Assessment of Methods and Resources

Constraints on Timing This pilot study was estimated to take five months to achieve a usable result, and this was an appropriate amount of time. It was not enough time to produce a sufficiently reliable map of landform process, particularly in view of the area’s unexpectedly complex geology (see Part 2). Soils and vegetation specialists contributing to fieldwork would have also been a useful expansion of the work. Approximately six weeks was taken up exploring different options for landform visualisation, and in a future project these questions would not need to be readdressed.

The time allocated in the field was two weeks (not counting travel time), and this was enough to understand many of the processes, ground-truth the preliminary map, and consult with and present to the landholders. The time in the field was hampered by very rainy weather, which lost two days (bad access). Time allotted to field work does not include the days spent in physical preparation before and after the field (gear & vehicle overhaul, etc); this should be borne in mind when allocating project time. More field time would have been better. The ideal field time for this 5-month project would be a week at the beginning of the project to get an overview of the area and consult with the landholder, and three weeks in the middle of the project for data collection and information presentation.

There were conflicting time requirements in this project: the landholders’ work timetable, and the firm dates for the project’s start and finish. The project began early in 2009 and was required to be acquitted by the close of the financial year (30 June 2009). The landholders could not have the researcher on-property until after shearing in late May; the field work happened in early June. Even with a two-week extension for the draft report, the post-field analysis time was short.

Landholder engagement with this type of project will not happen if researchers come on to the property at a time when they are not welcome; fieldwork must fit in around shearing, crutching, lambing, and so on. However, fieldwork should not be scheduled in the summer: hot-weather remote-area work involving foot traverses by workers not deeply familiar with the country is very poor OH&S. Ideally, fieldwork should happen sometime between autumn and early spring, at a time that suits the landholders –

however that’s the end of the financial year. This report recommends that a way be found to untie project scheduling from ending with the financial year.

Landholder Feedback At the end of the fieldwork the Mr. McTaggart was taken to selected sites and the results of the mapping discussed with him. Some labelled rock samples were left for his information. He seemed quite interested and pleased but at the close of this project I will have had no further discussion with him. Ideally, a project of this type would have time built in between draft and final reports for the landholder to comment on the project outcomes. I would be especially interested in hearing what new information was most valuable for land management. The time built in would be non-project time: a month (say) would elapse but no billable work would take place. To put it another way, the project timeline would increase by (say) a month, but the project budget would not increase by a month’s costs.

Aerial Photography and Satellite Imagery The existing published material covering the field area is limited to the 1:250,000 geological maps and their explanatory notes, a water resources report dating back to the middle of last century, and a Historical Society paper on the family history (Dickinson 1942, Anderson 1995). This small amount of information is all that can be expected to be available for a remote rural area in Australia: research and publicly- funded services concentrate around population centres. It is for this reason that field studies are crucial to a project of this kind, as background data are scarce. Other relevant information included references on the Acraman Impact Structure and on South Australia’s palaeochannels (Keeling & Self 1996, Williams & Gostin 2005).

Geomorphological mapping on the regional scale, such as this, requires a method of observing a broad swathe of country which also allows specific narrow-scale features to be identified on the ground. In this way, landforms can be grouped into categories and their distribution plotted on the map, and representative examples of the landform can be selected for ground-truthing and detailed examination. For example, a belt of sand dunes 50 kilometres wide can be mapped, and a specific sand dune can be selected from the map, and identified in the field. Key features that identify landforms and indicate landscape processes are shape, colour, and vegetation. The medium of observation must therefore be accurate in its display of elevation, true-

colour, and sufficiently high-resolution; it must also be affordable. Aerial photography (of the kind that once was routinely flown by State Departments of Lands, which gave overlapping photographs) is the ideal medium for this type of work. The resolution is very high (limited only by the magnification that the photograph is viewed under), the colours reflect what can be seen on the ground (Fig. 3A, 3E), existing photography covers the entire state, and with the use of a quality stereoscope the landform shapes are accurately revealed.

More modern remote sensing techniques are often assumed to be superior to air- photo interpretation, but this is not the case. Very few remote sensing techniques accurately depict elevation, and without elevation landform cannot be determined. Accurate elevation can be obtained with the use of Quickbird or LIDAR, but these techniques are beyond the budget of a land-management project. Free SRTM digital elevation data is available from NASA, however its elevation is only accurate to ± 2- 9 m. At this degree of accuracy, it is useful for regional overviews but not at all usable for landform mapping (Fig. 3D). Similarly, very few remote sensing techniques show true colour. Remote sensing which records within certain wavelength bands, and in which bands are combined to simulate true colour in a single image, do not really reflect the true colour of the soil or the vegetation (Fig. 3C). Finally, much remote sensing has a very large pixel size, suitable for regional overviews but inappropriate for detailed landform mapping. Some remote sensing has quite small pixels (for example, Quickbird,) but again these techniques are quite expensive. The more easily available and affordable LANDSAT has a 25 m pixel in most bands, and a 12.5 m pixel in the black and white panchromatic band.

Landforms were therefore mapped on to clear overlay sheets, using stereo pairs of 1:80,000 colour aerial photographs (film number 4874, flown in November 1994, 67 photographs in all). Sourcing these aerial photographs was unexpectedly difficult. Aerial photography has routinely been flown over all of South Australia, and negatives and existing prints are held by Mapland, at the Department of Environment and Heritage. I had expected to be able to purchase prints, but South Australia no longer has a photographic laboratory. Mapland will do an inkjet print from a scan of the negative or of the existing photo, however an inkjet print is not of sufficiently high quality to do geomorphological mapping. In addition, Mapland have not scanned the negatives from these kinds of rural areas, deeming them to be too remote to be of

A B

C D

Figure 3 Comparison of a small uplands valley as seen by: A, 800 dpi scanned aerial photograph; B, photomosaic made from the same scan; C, pan-sharpened LANDSAT bands 741; D, a DEM from SRTM data has a 90m pixel. Field of view 2.4 km. E, the landscape view. Real-world colours and shapes correspond better to air-photos than to LANDSAT. Better resolution of the air-photo allows individual hillslopes and vegetation groups to be identified.

general interest. Scans of the negatives would therefore be charged for (an additional $130 for each of the 67 photos over this study area). After some discussion, Mapland staff generously allowed the researcher to visit their office for a week to examine their existing photographs on-site.

Different options for obtaining air photos are examined in Table 1, and compared with the cost of purchasing air photos in a state which still has a photographic lab. Of the available options, examining Mapland’s photos on-site is the only economically feasible one. However, it is a concern that Mapland is not set up to be a library, and they might choose not to let people have access. A technically feasible solution is purchasing Mapland negative scans and having them printed photographically interstate, however the combined cost of negative scanning and interstate photographic printing may be prohibitive for NRM projects. Finally, the Mapland office is said to be downsizing, and the prints and negatives may be archived somewhere, possibly with limited access. The combined cost structure, low priority assigned to photos of remote rural areas, and lack of access to existing prints may remove this valuable information from the orbit of land managers. There are no plans for systematic acquisition of new digital aerial photography. Rather, new photography is commissioned and paid for by government agencies, and organised through DEH.

This report recommends that discussions between the NRM boards and the Department of Environment and Heritage make some provision for controlled access to the existing aerial photograph prints. If Mapland can host the occasional visitor that would work well; otherwise it is suggested that the existing airphoto prints be archived at an appropriate venue such as the geoscience library at PIRSA.

This report also recommends that consideration be given to the issue of access to aerial photography and orthophotography. Range managers are increasingly required to manage for whole-country goals (such as biodiversity or feral animal control), and certainly popular opinion can be very judgemental towards graziers and pastoralists who are not seen to be looking after the country. Nonetheless here is a primary tool for landscape interpretation which is no longer being supplied by government for the remote areas. Strong representations should be made to State and Federal policymakers that population density should not be the only determinant of information supply.

Figure 4 Comparison of the central Moonaree paddocks as seen by different visualisations at basemap- scale. A, Paddock map (superimposed over the budget photomosaic, see Fig. 4); B, aerial photograph photomosaic; C, pan-sharpened LANDSAT bands 1,2,3; D, pan-sharpened LANDSAT bands 7,4,1; E, LANDSAT panchromatic; F, DEM; and G, ALOS. Field of view 29 x 17 km.

The output of aerial photograph examination is 33 clear overlay sheets, marked with unit boundaries and areas of interest, and a list of geomorphic units, described and with preliminary process interpretations. The next stage is to compile the overlay sheets by plotting the researcher-defined unit boundaries onto a base map. The base map should cover the study area seamlessly, be georeferenced (usable in GIS and able to give real-world coordinates for points on the map), and contain enough information that matches may be made between base map and air photos.

The ideal base map would be orthorectified air-photo mosaics purchased from Mapland, with the added advantage that this would be a valuable resource for the landholders. However, these are not available over the South Australian Arid Lands. Considering the issues described above about resolution, colour, and cost of satellite imagery, it was decided to use LANDSAT bands 1,2,3 (attempting to produce something like true colour) with the resolution sharpened by combination with the panchromatic band. The LANDSAT was incorporated into a GIS database, and used in combination with vector topographic data and the digital elevation model as a multi-layered base map.

The LANDSAT was partially successful. The image was sufficiently clear that overlay information could be transferred to it, and its ease-of-use in the GIS made it very suitable for locating points of interest, plotting waypoints, etc. However, its colours

Fig. 5 Comparison of trial photomosaics over the central Moonaree paddocks (see Fig. 4A) created in- house during this project. A) budget run photomosaic, B) best run photomosaic. The budget run used scans of every other photograph (nos. 30 and 32) at ~25% overlap; the colours are uneven and piece boundary ghosting is visible but the image would be suitable for basemap production. The best run used scans of every photograph (nos. 30, 31, 32, 33). The colours are well-balanced to produce a nearly seamless image, although at detail level it has a little less contrast. In both cases the photomosaic process reduces image resolution, c.f. the original scans.

were not true colours (Fig. 4), making it difficult to relate LANDSAT features to units defined on aerial photographs (the transfer from overlay to base map taking place when the aerial photographs were no longer available). The end result was a map

(Fig. 2) partly defined on spectral characteristics not visible in the field, leading to misinterpretation of the map features.

In light of the mismatch between LANDSAT and field observations, after field work two trial versions of rectified photomosaic of part of the study area were created in- house, one (best run: photos 30, 31, 32) with more overlap than the other (budget run: photos 30, 32) (Table 1). These were successful and their geolocation properties were good. They had a little ghosting at piece edges, and the less overlap version had imperfect colour balance, but both were acceptable for basemap creation. Their costs are shown in Table 1 and comparison is shown in Figures 4 and 5. Photomosaicing decreases image resolution, so detailed work on airphoto scans is best done on unaltered images (Fig. 3A, 3B).

For further projects of this type, this report recommends the use of panchromatic LANDSAT, DEM, and vector topographic data as base map layers if unit boundaries can be drawn while air photos are still available for comparison. Rectified and georeferenced photo mosaics created in-house will be better than the LANDSAT, and will also produce an image the landholder can use (for example, for an EMS study); however it is more expensive. ALOS, a new type of satellite image, may also be useful: although the example used during this project (250cm_2007_RGB_LCC) was incomplete the false-colour was a better approach to real colour than other satellite data (Fig. 2).

Table 1 Comparison of cost and utility of images for geomorphic mapping. Costs based on 67 photos, or equivalent satellite coverage, at prices mid-2009 *On-site photo examination costs include transport to Adelaide, one week’s accommodation and food. ** Note that a best-run photomosaic would call for scans of all photos covering the study area, reducing the cost of the “print photos interstate” option to $1,675.

Image Type And Source Sufficient True Sufficient Georefe Cost Reso- Colour Scale & -renced lution for Detail for Per Whole Landform Base Map Photo Project Mapping $ $ For Landform Mapping - Images of Parts of the Study Area: for comparison, photos, existing lab, NSW Y Y N N 26 1,748 inkjet prints of existing scans (SA Mapland) Y Y N N 78 5,226 photos, scan negative, print interstate Y Y N N 155 10,385 photos, scan photo, print photo interstate ** Y (barely) Y N N 66 4.422 examine photos on-site SADEH* Y Y N N 1,680 new aerial photography SA DEH (for an Y Y N N 25,000 area equivalent to a 1:100,000 map sheet) -35,000

For Basemap and GIS dataset - Images of All the Study Area: LANDSAT panchromatic (black & white) N N partly Y 727 LANDSAT pan-sharpened bands 247 N N partly Y 1,680 Quickbird (stereo) Y Y Y 50,000 SPOT with 3-D Y N Y Y 9,688 SPOT N N Y Y 3,600 ALOS (stereo) Y N Y Y 1,320 SRTM digital elevation model N N partly Y 0 orthophotos from SA Mapland not available n/a n/a rectified and best run N Y Y Y 10,834 georeferenced photomosaics made within budget run N Y Y Y 5,489 the project, incl. scans ** Topographic GIS (vector) N N partly Y 0 Google Earth N N partly N 0

LANDSCAPE PROCESSES OF MOONAREE STATION: A PILOT STUDY

Part 2: Report to Landholders

Geology

This section describes the geological origin of Moonaree’s rocks and sediments. Land management implications are shown in italics, and summarised in the land management section of this report.

Gawler Range Volcanics > 1,600 million years old

During the Proterozoic geological age, the Gawler Range Volcanics were laid down. The Volcanics comprise a number of fairly similar silica-rich rock types, each with a different name. Most of Moonaree’s hills are the Yardea Dacite, but there is a belt of different rock types extending across Waurea, Kulgulya, Dingo Hill, 10 Mile, and part of Charpatta Paddocks. All of the Gawler Range Volcanics are dense, hard rocks, which are very fine-grained. A fresh broken surface of these rocks is a rich reddish- brown colour, studded with small (1-3mm) pale or dark crystals of feldspar or epidote (Fig.6). The outer surface is usually weathered to a more pale orange-brown.

Fig. 6 Chunks of the Yardea Dacite, the most common rock type on Moonaree Station.

Management implication – Groundwater: The rocks are dense and very fine-grained. They are not porous and the only way these rocks can hold water is in fractures and cracks.

Regional Fracturing and the Acraman Impact Structure, >580 million years old

During the Proterozoic geological age, earth movements created extensive fracturing in the Gawler Range Volcanics (Fig. 7). This is expressed in the landscape as valleys (along the fractures) and hilltops (where the rock is less fractured, it is more resistant to erosion). These structures trend towards 0º and 35º (north and north-northeast). Another, larger fracture system trends 320º (northwest). These fractures predate the Acraman impact.

Figure 7. The influence of geological structure on Moonaree landscape is strong, as shown by the trends of the fracture-valleys (dark lines) and the hilltops (gray lines).

Management implication – Groundwater: These fractures permit water to be stored in, and flow through, the volcanic rock.

580 million years ago, a meteorite or comet struck the earth west of the present Gawler Ranges. At that time the land surface was approximately three kilometres above its present level. The impact shattered the Gawler Range Volcanics most strongly in the central crater zone, forming what is now Lake Acraman and its surrounding flat plain, and the outer crater rim, forming what is now the Yardea Corridor (Fig. 8). In these places, the broken rock has been more easily weathered and removed, so the landscape is less hilly. The impact also added some curved fractures elsewhere. Fist-sized debris landed as far away as the Flinders Ranges, and the catastrophic effect of the impact is thought to have influenced the course of evolution of Earth’s early life (Williams & Gostin 2005).

Figure 8 Digital Elevation Model (DEM) showing a regional view 1 of the Gawler Ranges and its fracture pattern. 1, Lake Gairdner; 2, the Acraman Impact Structure; 3, the Yardea Corridor (black dashed 2 line); 4, like the Yardea Corridor, this curved feature (black dashed line) lacks hills. Picture is ~150 km wide. Lowest 3 elevation (lake level) is purple to deep blue, with 4 rising elevation climbing through the colours green-yellow-red-purple, to highest hilltops at dark blue to light blue.

Management implication – Gnamma Waterholes and Hilltop Domes: The same fracture systems that are expressed on the landscape scale are also visible in a much smaller area (Fig. 9A). Where fractures occur close together and the rock is broken into small pieces, it weathers more rapidly. Rock fragments are transported away from the fractured area, leaving a small depression in which soil collects and vegetation grows (Fig. 9B). The soil acidity and moisture speeds up the weathering process, and the small depression becomes bigger and deeper. In this way a wide and relatively shallow hole (a gnamma) is created, in some cases becoming a waterhole: a reliable water source for human and animal use.

Where fractures are widely spaced (tens to hundreds of meters apart) the rock is resistant to weathering and can form a broad unvegetated dome on the hilltop or hillside. (Sometimes rock domes weather by spalling or onion-skin weathering, where 5-20 cm thick sheets break free from the underlying rock.) Very little vegetation grows on a rock dome, and it will be a high-runoff area during rain. Vegetation on hillslopes beneath such domes would benefit from the increased runoff, though there is also an increased risk of erosion during heavy storms.

A B

Figure 9 A) Closely-spaced fractures on a hilltop in Crown Paddock, oriented north-northeast, northwest, and east. B) Rock rubble has weathered away from closely-spaced fractures, leaving a soil-filled hole which can eventually form a gnamma.

Tertiary sediments: white claystone (65 million years to 2 million years ago)

At Moonaree a number of outcrops of white claystone were seen (Fig. 10), the best being on the banks of Gorge Creek, 100m from the Outside fence in Gorge Paddock Fig. 11), and beneath the gypsum layers in the gypsum field at Parkers Outside Paddock. The claystone is crumbly-looking, bright white, slightly sandy, and cut by a network of very fine ?quartz veins. The white claystone is normally very poorly exposed and hard to see, and also hard to distinguish from the weathering profile (see below).

Figure 10 Observed occurrences of white claystone (dark pink triangles) and weathering profiles (pale orange triangles), plotted on a Digital Elevation Model which assumes a lake level at an elevation of 200 metres above sea level (dark grey). The current Lake Gardener shoreline is at 130 metres ASL.

During the Tertiary geological age, earth movements and variations in the local climate caused sea level to rise and fall many times. In the Gawler Ranges area,

large creek systems extended from the ocean back as far as Glendambo. These old creek systems (or “palaeochannels”) are now marked by chains of salt lakes. Elsewhere in southern South Australia, these palaeochannels are known to contain dolomitic (magnesium-limestone) shales grading in colour from dark grey to bright white (Keeling & Self 1996). With wetter climates and higher sea levels, lake levels would have been higher than they are today. If the Moonaree white claystone is similar to the channel sediments found elsewhere in South Australia, it is likely to have been deposited during these high lake-level times. Locations of the observed outcrops are consistent with being deposited in water-filled lakes (Fig. 10).

(Note: during fieldwork it was thought this white claystone might have been related to the Great Artesian Basin, but further investigation indicates this is unlikely.)

Figure 11 A 4m high bank of white sandy claystone, overprinted by brown, red, and dark purple-red of the weathering profile. Location is 100 m from the station track, along Gorge Creek, near the northeastern fence of Gorge Paddock.

Management implication – Gilgai Soils: The palaeochannels are strongly associated with swelling clays (smectite and palygorskite) (Keeling& Self 1996). These types of clay crystal are much bigger wet than dry, and are an important component of shrink- swell soils like gilgai. Since gilgai soils are prominent on Moonaree Station, the distribution of these swelling clays (Fig. 16) will influence paddock productivity.

Tertiary weathering profile, ironstone (buckshot) gravel, and silcrete (65 million years to 2 million years ago)

At some time during the Tertiary geological age, there have been episodes of strong weathering. Long exposure, warm climates, and plenty of groundwater will change the chemistry of the rocks close to the earth’s surface, leaching out some substances (like quartz) and concentrating others (like iron). The result is a weathering profile: bleached white clays, in places heavily mottled with iron oxides. This can be best seen at the claystone outcrop (Fig. 11), but there are places where the dacite is so heavily weathered that it also looks like white claystone (for example Jolly Creek in garden well paddock).

Management implication – Groundwater: The weathering profile will penetrate most deeply where the rock is most fractured. The chemical changes taking place during weathering are likely to partially fill the fractures, decreasing fracture permeability. In this way locations that look promising for putting down wells (like the intersection of two fracture-valleys) might have surprisingly poor results. The extent of the weathering profile may strongly influence groundwater availability and quality, however further investigation is required to answer this question.

In some places the iron oxides are so concentrated in the weathering profile that they form bands of very hard, very dark red to almost black, siliceous ironstone (Fig. 12). This ironstone breaks into small shiny dark pebbles (sometimes called “buckshot gravel”) which are found in patches on the surface of some paddocks, especially Little Gorge and Parkers.

Under certain conditions of groundwater chemistry, silica from quartz dissolved out of the weathering profile gathers together and recrystallises, cementing together or even completely replacing the sediment it surrounds. The result is a silcrete: a very hard fine-grained pale rock often used for making stone tools in pre-European Aboriginal technology (Fig. 13D). No outcrops of silcrete were observed during this fieldwork but some is known to exist at Crown and Woolly Paddocks, and there is a large silcrete outcrop at the lake margin on Mount Ive Station (Fig. 13C).

C A

B D

Figure 12 Tertiary rocks and weathering profile. A, white claystone mottled with dark red iron oxides of the weathering profile; B, heavily weathered dacite; C, heavily weathered rock that might be claystone or might be dacite; D, buckshot gravel cemented by iron oxides; E, well-rounded silcrete pebbles with a characteristic highly-polished surface.

Tertiary sediments (65 million years to 2 million years ago) Ironstone and silcrete fragments, eroded from their original places in the weathering profiles, can be transported in creek systems, becoming smooth, water-worn pebbles over time. Sometimes these old creek sediments can become re-cemented into new rocks. Highly-polished silcrete pebbles and re-cemented buckshot gravel (Fig. 12 D, E) were found in a few places.

Management implication –Groundwater: The sediments indicate that the sub-surface geology of Moonaree is likely be complex. This would contribute to the difficulty in predicting the location of good-quality groundwater.

Tertiary to modern calcrete (65 million years ago up to the present day)

Calcrete is a rock similar to limestone. Like the silcrete, it precipitates from groundwater and cements or even replaces the sediments that the groundwater is surrounding. It requires different conditions of groundwater chemistry, but the

groundwater must be very high in dissolved calcium and carbonate. Calcrete is hard and usually white (but less bright than the claystone) to pale pink or tan. The outer surface of calcrete nodules and rocks has a slightly grainy, matte appearance.

B C

A D

Figure 13 A, calcrete; B, dirt-covered calcrete pisoliths surrounding something that is now weathered away (possibly dacite rubble), leaving a mass of hollow spheres; C, an unusual silcrete from Mt. Ive, in the form of hollow spheres, possibly silica replacing a calcrete rock similar to B; D, a stone tool made from fine-grained silcrete.

Calcrete occurs all over Moonaree, from the hilltops to the valley bottoms. It has no relationship to higher lake levels (Fig. 14). It occurs in a variety of forms (Fig. 13): as sheets cementing rocks or dacite rubble, as rubble at the break of slope between bare rock and rocky soils on hillslopes, as layers of nodules close to the ground surface (best visible as white patches on station tracks), and as chips or veins through the soil profile (“calcareous earths”). The silty soils associated with calcrete are a slightly pale orange and are hard underfoot.

Calcrete and silcrete can form over the top of each other, and the complex relationships at Moonaree indicate that there have been several episodes of calcrete and silcrete formation.

Figure 14 Widespread distribution of calcrete outcrop (blue crossed boxes) or calcareous soils (blue wavy lines) has no relationship to previous higher lake levels. (Digital Elevation Model shows the lake level at 200 metres above sea level (dark grey); the current Lake Gardener shoreline is 130 m ASL).

Pleistocene to modern gypsum (2 million years ago up to the present- day)

Gypsum (calcium sulphate) is widespread on Moonaree. Its most common occurrence is as high-salinity groundwater in some wells and dams, and a few creek channels (notably in Crown Paddock, near the Yeltabinna fence, where a hyperconcentrated brine was precipitating gypsum at the sediment-water interface). Its next most common occurrence is as kopi (also known as “seed gypsum”), fine (< 1mm) lens-shaped crystals. Typically these are pale tan-coloured accumulations, very light and soft underfoot, described by the landholder as “something your motorbike would get bogged in”. Kopi forms above the groundwater surface, in or above the soil profile. Older deposits of kopi can consolidate and harden (gypcrete), such as the bench of old kopi a few hundred metres from the creek channel in Crown Paddock. Gypsum that precipitates over a long period of time below the groundwater’s surface forms narrow, long (>10 cm) semi-transparent bladed crystals

with swallow-tail twinning. These can be found either deep in the soil (exposed when a dam is deepened, such as Mount Cooper dam), or in the wet sediment of the lake (Fig. 15E). ”Desert rose” is a special kind of gypsum crystal that probably forms in a narrow zone of fluctuating groundwater level at the lake’s edge. Its crystals are large (0.5-5 cm) but lens-shaped and may gather in semi-circular flower-like structures. A layer of desert rose crystals (like those at the Parkers Outside inlet) indicates a previous lake shoreline.

There have probably been many lake-full episodes during the Ice Ages. At the end of each, while the lake was gradually drying up, water salinity would have increased and gypsum precipitated within the lake and around the lake edges. The distribution of gypsum at Moonaree is related to higher lake levels (Fig. 16).

Figure 15 A, modern soil C derived from dacite; B, slightly silty quartz sand; C, old consolidated kopi D (gypcrete); D, older gypsum blade crystals; E modern A B gypsum blade crystals dug from the lake’s edge. E

Management Implications – Groundwater: The presence of gypsum crystals in an area, or its position in the old lake (Fig. 16) may indicate decreased chance of good- quality groundwater. However, it is also possible that high-salinity brine underlies the whole station. Hydrological investigation would be useful to answer this question.

Pleistocene to modern soil and sand (2 million years ago up to the present-day)

Much of the soil at Moonaree is derived from the local dacite rock. It is (like the rock) coloured a deep red-brown (Fig. 15A), and is fairly coarse in its grain-size. Pure dacite soil is firm but not hard underfoot. It is easy to drive over (not bulldusty, not soft, only a little greasy when wet). Dacite soils support a rich variety of chenopod vegetation.

Figure 16 Previous high lake levels may control the distribution of gypsum (surface occurrence, white blazes; subsurface occurrence, white crossed boxes) and soils containing swelling clay (gilgai, green circled diamond; soft crunchy soil, green wavy lines; hard-water dams, green boxes). (Digital Elevation Model shows the lake level at 200 metres above sea level (dark grey); the current Lake Gardener shoreline is at 130 metres ASL).

Management Implication – Plants: The type of soil, and depth of soil, has a clear influence on plant type. Rocky, thin, poor soils of the hilltops support spinifex but rarely chenopods, and the boundary between the two plant groups is usually sharp. Hillslope soils may support chenopods but they will be small and sparse if the underlying rock is close to the surface, or if the soil is full of dacite rubble. Soils with more fine silty material and less rubble have thicker stands of bigger bushes. There were several different types of chenopod assemblages on different chenopod plains. A survey with a plant specialist and a soils specialist would be informative.

Pleistocene to modern sand (2 million years ago up to the present-day)

A wide belt of sand extends east-west across the southern paddocks of Moonaree, visible in the LANDSAT as blue-and-white colour. It is mostly medium-fine quartz

sand which has been blown hundreds of kilometres from the west. It is softly sandy underfoot, fine-grained enough to bog vehicles. It supports hardy specialist plants like spinifex and mallee, and makes poor grazing country. The belt of sand is in the form of a flat sand sheet, with high narrow dunes trending east-west, covering the dacite except for the highest hill crests. Sand is thickest to the east of dacite hills, where it is sheltered from wind. The sand sheet is shallow in places, but piled high along its northern margin, where it faces into Moonaree (Fig. 17). The sand is generally stable although there are some patches of erosion near dune crests and on the steeper dune slopes near the northern margin. Closer to Lake Acraman, in the west of Mulcaree, the sand has a greater proportion of dacite. This is visible in the LANDSAT as brown-and-white colour. It is much more firm underfoot than the quartz sand.

Management Implication – Feral Animal Control: According to the landholders, sheep and goats are disinclined to travel over quartz sand, and they attribute the relatively goat-free status of Moonaree to the sand belts north and south of the station. The dacite sand however is firm under foot, and is not such a barrier. Where dacite sand extends almost through the quartz sand (for example, west of Mulcaree, or far east of Morinippi), an opportunity may exist for goats to enter the station. The LANDSAT image (Map 3), which shows the difference between quartz and dacite sand, can be used to suggest where such a pathways exist.

Management Implication – Erosion Control: the northern edge of the sand belt has the potential to become susceptible to erosion. At the moment, it is held in place by the combination of protecting hills and anchoring vegetation. Where vegetation is depleted, it shows signs of mobility (erosion), and in places the sand is creeping northwards over the chenopod plains. If the sand sheet was to encroach northwards, paddock productivity would decline. Preservation of sand dune vegetation should be considered, possibly including fire management (controlled, small, cool, patch burns in the highly flammable spinifex + mallee country).

Figure 17 (next page) The Moonaree southern sand sheet. A) Sand elevation profile along the track south to north through Mulcaree; yellow dots match landmarks noted at the bottom of the profile. B) LANDSAT view of southwestern Beviss; green lines are fences and the yellow line is the path of the sand profile shown in C. The sand sheet is a blue and white colour, the dacite hills are dark and the chenopod plains are brown. Note the dune crests (look like ripples) near the sand sheet edge. C) Sand profile southwest to northeast through the southwestern corner of Beviss.

Kittles high gate track Sprft Dam dacite hills gate dogleg Dam

fence sand profile edge dogleg

Geomorphology and Landscape Process

This section describes the landscape processes currently operating on Moonaree’s landforms. Land management implications are shown in italics, and summarised in the land management section of this report.

Figure 18 Landscape elements (Gorge Paddock). rocky hilltop upper lower chenopod

hillslope hillslope plain

The Moonaree landscape consists of the following elements (Figs.18, 20):

• rocky hilltops, forming more or less high domed hills, where bare rock is exposed; characteristically the vegetation includes spinifex and sometimes small trees • upper hillslopes, where not much rock is exposed but where the soil is dominated by dacite sand and abundant large chunks of dacite rubble; rock is probably only shallowly buried; vegetation is thinly-spaced chenopods and sometimes small trees • low rubble hills, without exposed rock; the soil is dominated by dacite sand and abundant rubble; rock is probably only shallowly buried; vegetation is thinly-spaced chenopods • lower hillslopes, with some rubble in the dacite-sand soil; vegetation is more closely-spaced chenopods; grades into chenopod plains

• chenopod plains, wide flat spaces, more or less densely occupied by chenopods; soil is silt and dacite sand with little or no rubble; the central drainage lines are typically discontinuous small creeks, or unchannelled swales • sand sheet with dunes, near the lake edges, and in a belt along the southern station • narrow rocky valleys between closely-spaced hilltops; created by weathering along the regional fractures (see Geology above); drainage lines are small, discontinuous gutters flanked by closely-spaced small trees (Fig. 19) • the larger creeks (especially the creeks that flooded in 1997: Jolly, Yeltabinna, Garden Well, Old Homestead) are arroyos with steep sides, typically 1-4 meters deep and 5-20 metres wide

Figure 19 In a small rocky fracture-valley (Dingo Hill Paddock), the central drainage is only sometimes contained within a channel: this small creek is discontinuous (alternates between channelised flow and unchannelled sheetflow). Arrow shows end of channel; flow towards camera along arrow path.

Salt Lakes

When the lakes are dry, gypsum crystallises on the lake floor and is sometimes blown into the eastern shore, forming shoreline dunes of very pale fluffy gypsum sand. When the lakes are full, the prevailing westerly wind creates waves and currents along the eastern shoreline. This is why the western shorelines of Lakes

Acraman and Gairdner have a rough outline on the map, whereas the eastern shore lines have been smoothed by wave action.

Fig 20 Chenopod plains. A) Belt Hill Paddock: Lower hillslope with rubble and chenopods merging down to chenopod plain (see Fig. 21) in distance (small white dot below hills is white ute for scale). B) Chenopod plain in Bond Paddock has a surface that is very often wet: the cryptogam crust is very thick and dark, and there are mosses and liverworts. (White ute in the distance for scale.)

Chenopod-covered Lower Hillslopes, and Chenopod Plains: “Sponge Country”

On air-photos and satellite images, Moonaree looks like it’s mostly rocky hills, but on the ground the landscape is dominated by wide, low-gradient vegetated land: lower hillslopes and chenopod plains. Red soil partly covered by closely-spaced chenopods rises gently towards the surrounding hills (Fig. 20A). The hillslopes and plains have soil dominated by dacite sand and silt, with a variety of differences and supporting different plant communities. Pure dacite soil is firm but not hard underfoot; calcareous soil is hard underfoot; kopi (gypsum sand) is very soft and fluffy; and soil containing swelling-clays is soft underfoot and sometimes has a slightly crunchy crust. Some chenopod plains had evidence of being frequently wet at the surface: the cryptogam crust was very thick, dark, and wrinkled, and there were moisture-loving plants (mosses and liverworts) growing between the shrubs (Fig. 20B).

In many chenopod plains where you’d expect to find a creek in the valley centre, there is no visible drainage network (Fig. 21).

Figure 21 Looking across the drainage axis of a Mt Cooper Paddock chenopod plain, there is no visible creek. The dip in the road (right of the photo) is damp and tiny salt crystals (?gypsum) are forming in the wheel tracks.

Management Implication – Paddock Productivity: The lack of organised drainage is a key factor in paddock productivity. Rainwater and hillslope runoff stays around instead of heading off down the channel – water has the best possible chance of infiltrating the soil. The relatively closely-spaced chenopod vegetation, and the shape of that vegetation (bushy right down to the ground level) also promotes water infiltration, by placing barriers across downslope flow. The lower hillslopes and chenopod plains act like giant sponges, and as a result there is good growth of vegetation. Where vegetation is completely lost, there is no barrier to water runoff. Water is not retained, vegetation does not germinate, and sheet erosion is likely (for example along the stock route in Kallinta Paddock, Fig. 22).

Furrowing rehabilitation works were undertaken by the landholders who report successful plant establishment after ~20 years. The works were not able to be laid out along surveyed contour lines. Contour furrowing in the Western Catchment of New South Wales has been shown to be most successful where contour elevations are most accurate. It is likely that the Moonaree works would have had more success sooner had surveying been available to assist the landholders.

Figure 22 The stock route in Kallinta Paddock has bare patches along it, and sheet erosion is demonstrated by the remnant soil pedestal (isolated plants, lower right). Furrowing rehabilitation works have been undertaken (where Adrian is standing), and the landholders report some vegetation growth after ~20 years.

Discontinuous Ephemeral Creeks

In some hillslopes and plains, the drainage axis can be seen as a line of scattered Acacia trees around a poorly-defined swale (a swale is a depression that holds water, but which is not as clearly-marked in the landscape as a creek channel or a gully) (Fig. 23A).

Figure 23A) A swale with trees is the only sign of central drainage in the chenopod plain of Belt Hill Paddock. Backpack and pink notebook for scale, swale extends the full photo width at the backpack location, there are no clear banks. B) Sketch of a line of discontinuous creeks down a drainage line: channels are separated by unchannelised sheetflow zones (light blue arrows). Looking from above, arrows show flow direction, orange shapes are the banks of the channel. As water flows over the chenopod plain, it is sometimes it is gathered into a channel. At the downstream end of the channel it spreads out as sheetflow again.

In other places, the drainage is a little more organised, forming short discontinuous creeks, sometimes singly, sometimes one after the other in a line separated by channel-free sheetflow areas. In each of these discontinuous creeks there is an

Fig. 24 A discontinuous creek in Mt Cooper Paddock, near the road crossing at Yeltabinna Creek..

A Looking upstream along the

beginning of a discontinuous creek. Water flowing down the centre of a chenopod plain gathers enough energy to trigger erosion.

The hammer (for scale) is between two plunge-pools, and small creek channels pass to right and left of the camera.

B Looking downstream along the

central creek channel. It is shallow and the banks are not sharply defined. This channel is close to the station track at the north of Mt Cooper Paddock.

C Looking downstream along the end of a discontinuous creek. Water flowing down the channel

from B (above) disperses over the chenopod plain as the channel gets more and more shallow and eventually disappears. The shrubs here have very water-rich leaves.

upstream eroding end, a central channel, and a downstream distributary end (Fig. 23B, 24). In other landscapes the downstream end of a discontinuous creek is a floodout (a richly-vegetated “sweet spot”) but at Moonaree the floodouts are not strongly visible, probably because the downvalley slope of the drainage lines is so strongly influenced by non-fluvial processes.

Management Implications – Channels & Decreases in Paddock Productivity:. Discontinuous creeks are common across Moonaree and there is no reason to think they aren’t a natural part of the landscape. (For example, few are found in high- erosion areas: stock routes, piospheres.) However, in other parts of the world, where discontinuous creeks become eroded and the channel either deepens or lengthens, water that was once retained in the floodplain goes straight into the channel and paddock productivity decreases (Figs.25, 29, 32).

This kind of erosion can happen when a valley-floor is scoured by extreme rainfall events. It can also happen under more normal rainfall if the valley-floor resistance to erosion is weakened by vegetation loss through overgrazing or drought. Keeping the valley-floors well-vegetated will help to keep the discontinuous creeks from joining up – which will help keep the valley-floors well-watered.

Fig. 25 Yeltabinna Creek in Ten-Mile Paddock. After channel incision (possibly during the 1997 floods) water has less chance to accumulate in the chenopod plain, and trees along the drainage line may find it difficult to survive.

Large Creeks: big flows, channel incision, channel relocation

The longest creeks, or creeks with the biggest catchment areas, are most likely to gather enough water to carry large flows. Other factors include 1) the path of a storm’s most intense fall will cross some catchments and not others, and 2) rain that falls on rocky (high-runoff) ground will quickly find its way to a channel, but rain that falls on “sponge” ground will not travel far. In the 1997 floods, the landholders report that Garden Well, Jolly, Old Homestead, and Yeltabinna Creeks ran very high. Two are the longest creeks with the biggest catchments, and two are shorter with small catchments but collect water from high-runoff rocky areas (Figure 26, Maps 1, 3).

Moonaree’s main creeks are not well-connected. Instead of the channels being continuous from hills to lakeshore, they may be discontinuous even along the main drainage axis (for example, Crown Paddock). Some channels are deeply incised with clear banks (for example, Jolly Creek), others show evidence of recent channel shifts (Boolatta Creek in Little Gorge). Moonaree’s creeks are not stable landscape features. This makes Moonaree a little unusual: most landscapes are governed by river landform development, but Moonaree is more influenced by hillslope processes.

Figure 26 Catchment areas of the Moonaree creeks. 1, Yeltabinna; 2, Garden Well; 3, Jolly; 4, Old Homestead; creeks that were flooded in 1997.

Gilgai

Gilgai soils have clay chemistry resulting in shrink-swell characteristics (see the Geology section above). When the ground is wet, gilgai soaks water like a sponge. When it is dry, it shrinks and cracks. It can be very rich and biologically productive. There are many different kinds of gilgai. On Moonaree there are gilgai pits in Little Gorge, Gorge, Kallana, Kallana Outside and White Well Outside Paddocks (and probably others too). In Gorge, North Kallana and Kallana Outside there is stony gilgai as well as gilgai pits.

Gilgai pits are roughly circular depressions in the ground (ranging from less than a metre to 100s of meters in diameter, Figs. 27, 28), centred around crabholes (very deep cracks that open up after the wet ground dries out). Stony gilgai is a step-like arrangement of rocky steeper ground and flattish muddy cracked ground. The cracks or crabholes are the key characteristic of the gilgai. The shape of the ground – the pits or the steps – helps to trap the first flush of rainwater, which drains into the open crack. The water deeply penetrates and soaks the ground, which swells and heaves. The heave creates the pit or the stony steps. In this way the gilgai are self-sustaining landforms. The water-retaining nature of the gilgai supports dense vegetation. At the time of this study, the gilgai pits were distinguished by a rich crown of bright grasses (Fig. 27). The chenopods were often larger, or more closely spaced, or had more water-soft leaves in the gilgai pits.

Distribution of gilgai around the property probably relates to previous high-lake levels (Fig. 16). It is likely that the swelling-clay soil elements are blown to nearby hilltops and washed downhill by rain, so isolated occurrences of gilgai may be found elsewhere also. Other influences on the size of gilgai structures include soil depth to bedrock, and mixture of coarser elements (such as dacite sand).

Management implications – Paddock Productivity: The landholders report Little Gorge Paddock as being the one of station’s best paddocks, and it is also a paddock very rich in gilgai pits. Observations during this study indicated gilgai pits were highly productive ( more biomass, greater species diversity, higher number of desirable species; Fig. 27). It is likely that paddock productivity is high where many gilgai are

A Fig. 27Gilgai pits. A, the inside of a small gilgai pit in Gorge Paddock, with pale grasses around the edge and a deep open

crabhole below the hammer. Moisture-loving vegetation (possibly nardoo) is now dried-out and dead (brown-grey colour) between the

grass and the bottom of the photo. B, A moderately large gilgai pit in White Well (ringed by bright grass, Adrian for scale). Heavy local rainfall has eroded sediment from the upper hillslopes (rocky foreground,

and bare hillslope in the distance). The sediment has washed into a shallow valley and a gilgai pit has formed in this local area

of deeper soil. C, The hillside across the valley from this fence is dotted with large gilgai pits, shown by the patches of pale grass (White Well Paddock).

Figure 28 Along the fenceline in Gorge Paddock, one of the

biggest gilgai pits on Moonaree Station is hundreds of metres in

diameter (bright grass). Trees and very large crabholes are in the pit’s centre, where the track dips. On the other side of

the fence, near the watering point, there are few signs of gilgai

function as well as little vegetation.

present. Future management plans may benefit from noting gilgai distribution about the property. Future vegetation monitoring sites should include some gilgai country.

Management Implications – Maintenance of Paddock Productivity: Some rangeland managers consider that vegetation plays a key role in maintaining gilgai processes. Certainly there are places in western New South Wales where loss of vegetation along stock routes has been accompanied by a decrease in the visible signs of gilgai processes (fewer crabholes, less ground “heave”). Similar situations may exist on Moonaree, in Gorge and Spearfelt Paddocks (Figs. 28, 29). Preservation of a minimum level of vegetation on gilgai landforms is likely to be important for preserving their function and productivity. At present, there are no accepted techniques for rehabilitating gilgai-pit country (although contour furrowing has been shown to be useful in some circumstances for stony gilgai country in western New South Wales).

Management implications – Station Tracks: Very large gilgai (such as Fig. 28) are likely to be uncomfortable driving in all weathers. Sloppy and boggy when wet, when dry they will be uneven and thick vegetation will conceal sump-busting holes. If erosion starts along the track (see below) and drains into the central crabhole, the result will be a two-headed gully extending away from the crabhole on either side. No matter how much material erodes from the track, the crabhole will not fill up (they are incredibly big). For the same reason, dumping road base into the hole may not be effective either. It’s probably easier just to drive around them.

Heavy Rain and Local Hillslope Erosion: Moonaree’s Dominant Landscape Process Today

In most properties, region-scale slopes and rivers are the dominant influences, so modern-day landscape change usually relates to pre-existing river landforms. At Moonaree, the hillslopes are all local (originating in the dacite fracture pattern), and the valley slopes have largely been controlled by local hillslope sediment deposition and low-gradient old lake sediments. River landforms are a relatively minor part of the landscape, and the creeks are generally only poorly-connected so flow events tend to have only local affects.

Figure 29 Gilgai at Cornish Well in A Spearfelt paddock. A, Hammer handle sticking out of a deep crabhole. B, Although there are

many crabholes, there are no gilgai pits in this heavily grazed area.

Hillslope processes dominate modern day landscape change in Moonaree, and the active agent for geomorphic change on hillslopes is heavy rain.

Hilltops:

Intense cloudbursts falling on bare rocky hilltops can give rapid runoff. In many parts of Moonaree the rocky hilltops have a belt of dense vegetation at their base, where the plants have received the benefit of the extra runoff. In other parts of Moonaree, the break of slope between rocky hilltop and upper hillslope (see Fig. 18) is marked by common outcrops of calcrete, which also indicates a flush of groundwater.

Dacite soils, and clays and silt blown onto the hills during dry seasons, are washed downslope by rain storms. If the rain is particularly strong, or if vegetation has been reduced (for example, by lightning-strike fire in the spinifex), the rocky hilltop can exhibit moderately severe local erosion. In some hilltops the spinifex grows in contour-parallel bands and traps some of the downward-moving sediment (Fig. 30). Sometimes, the downslope movement overwhelms the hillside vegetation and a little patch of sediment forms lower down (Fig. 30). If the clay chemistry is right a gilgai pit will form in this patch.

Figure 30 Contour parallel bands

of spinifex (arrowed) trap sediment in a convergent valley in Parkers Paddock. Arrow shows direction of runoff flow. A wedge of sediment (*) has been washed from higher in the hillslope at some previous time.

Upper Hillslopes:

The upper hillslopes, with their thin vegetation and rocky soils, are not likely to absorb much rain. If a lot of water is shed during heavy rain, local patches of erosion may be created and sediment deposition will take place on the lower hillslopes and chenopod plains (for example, at White Well Paddock, Fig. 27 B.). Short sections of discontinuous creeks along hill bases may be activated.

Lower Hillslopes and Chenopod Plains: Chenopod plains will absorb most rain and runoff, as long as the vegetation and the gilgai are in good condition. However, creek channels running through the chenopod plains will tend to capture water and funnel it away.

Management Implication – Keeping Hillslope Processes Dominant: A little excess runoff from place to place, or from time to time, giving localised patches of erosion and sediment deposition or short discontinuous creeks, is a natural part of this landscape. Moderately large channels at the downstream end of the largest creek systems are also to be expected in this landscape. The important point for Moonaree is to avoid creating conditions in the lower slopes and chenopod plains where repeated runoff may trigger erosion, creating channels or joining discontinuous creeks to create longer, deeper channels.

Factors which may promote erosion and channel formation are loss of vegetation (Fig. 22), stock pads (Fig. 31), and station tracks (see below). Vulnerable locations include stock routes, watering points, and convergent slopes (where several hillsides face towards each other and the valley between receives extra runoff, Fig. 30, 31). A combination of factors can make an area especially vulnerable, such as at Cornish Well, where a substantial gully system (Fig. 32) has formed in an area where a track down a convergent slope leads to a poorly- vegetated watering point.

A really extreme rainfall event will also promote erosion and channel formation, regardless of range condition, however good range condition will mitigate against erosion in most weather conditions.

Fig. 31 A stock pad down the centre of a convergent slope may become a focus for erosion. Hammer for scale.

Figure 32 A deep and actively- expanding gully system can be traced upslope to an old track near Cornish Well, Spearfelt Paddock (and see Fig. 29).

Management Implications – Gullying Along Station Tracks: Graded and ungraded station tracks are naturally predisposed towards gullying because barriers to flow (like vegetation) have been removed, and hillslope water is concentrated between track edges. In addition tracks are sometimes located where they intercept hillslope runoff, gathering large amounts of runoff (and incidentally starving the downslope vegetation of water).

Gully erosion along tracks in Moonaree is moderately common (noted in Parkers Outside, White Well Outside, Charpatta, Beviss, Bond Hill, Belt Hill Paddocks). Most gullies erode upslope from some low point such as a creek crossing or the central low point of some plain (Fig. 33A, B). Some in gullies erode in two directions into a central point within the gully itself, like someone pulling themselves up by their own bootstraps. This is probably related to deep cracking in the valley sediments (see Gilgai above).

Most of the gullies are not large, but nonetheless they use station resources in grading for track maintenance. Some gullies are very large and represent a real risk

A B

C Figure 33 Gullying along station tracks. A) Charpatta Paddock, gully draining towards centre of chenopod plain; B) gully draining towards creek, Kumburta Paddock (hat for scale); C) Beviss Paddock, a double-headed gully draining in towards its own

central point. One gully head is in photo foreground, the other is shown by white arrow.

to vehicles, or station tracks have had to be relocated (for example White Well Outside). The biggest concern is if these gullies become actively-eroding large networks, such as the one in Spearfelt Paddock (Fig. 32). Such gully networks can considerably reduce or destroy paddock productivity, and are extremely difficult to rehabilitate. Examples elsewhere are known to be kilometres in length.

Effective techniques for track placement, design and maintenance are not within the scope of this report, however should be an important part of the SAAL-NRM information resources made available to land managers.

Implications for Land Management: Summary

The statements in this report do not imply any criticism of past or present managers of Moonaree Station or SAAL-NRM.

Groundwater is most likely to be carried in fractures through the volcanic rocks, and in the younger sediments of the fracture-valleys. However, weathering and sediment deposition during higher lake-levels in the geological past has created a complex groundwater situation in which quality and supply are variable. Map 4 and Fig. 16 may indicate areas affected by heavy weathering and gypsum deposits, however it should be noted that this project’s mapping component was intended to provide an overview only. Groundwater flows down from hills to lake, and flow is more rapid where the ground surface is steeper (Dickinson 1942). Local variations in slope will also therefore add complication to the groundwater picture. Local aquifers in wide flat areas are more likely to be slow-flowing and stagnant. Those closer to steeper hillslopes may be more quickly freshened by rain on the hillslopes. Finally, landholders observe that fresh water is usually above saline water – this is likely to be a result of the greater density of saline water. Fresh water isn’t weighted down with dissolved gypsum, so it floats on top of the saline water.

Gilgai soils are an important part of Moonaree’s productivity. Map 4 and Fig. 16 may also indicate where gilgai soils are most likely be found.

The principal soil components are dacite sand, silt, clays (including swelling clays), quartz sand, gypsum and calcrete. Vegetation is strongly related to soil types, from the relatively poorly productive spinifex communities (in quartz sand and in rocky hilltop areas) to the highly productive chenopod plains in the wide, low- gradient valleys. Although superficially similar to each other, many of the chenopod plains have different soils and different plant communities.

The wide belt of sand across the south of Moonaree is not very productive country but may play a role in keeping feral goats off-station. The sand is currently stabilised by underlying hills, and by vegetation; loss of vegetation may lead to sand encroachment across more productive country.

The chenopod plains are the most important landform on Moonaree Station. Their lack of connected drainage makes them into giant “sponges”, which is the heart of their productivity. Short discontinuous creeks are common throughout Moonaree, but long continuous channels are not common. Valley-floor erosion, creating new channels or joining together short shallow creeks into long deep channels, will decrease or destroy productivity in the chenopod plains.

Vegetation is an integral part of the landscape processes that maintain gilgai landforms and protect chenopod plains. It is not only important to preserve vegetation for its own sake, but also because the loss of vegetation means the loss of landscape processes which allow the vegetation to grow.

Gully erosion down tracks is present in many places in Moonaree. While not currently a serious problem (except in Spearfelt), such erosion has the potential to seriously degrade landscapes. Information on effective track creation and maintenance should be available to all landholders.

The loss of vegetation along stock routes and near watering points is a common problem throughout the rangelands. Because the loss of vegetation may change the landscape processes, the land may not effectively recover without intervention. Rangeland rehabilitation processes such as contour furrowing are most likely to be successful and economically viable if information on successful techniques is available to the landholders, and if the landholders have access to accurate surveying when installing the rehabilitation works.

LANDSCAPE PROCESSES OF MOONAREE STATION: A PILOT STUDY

Part 3: Controls on Surface Water in SA Arid Lands

The Great Artesian Basin is not the only important water resource in South Australia: surface waters in drylands rivers are also a resource worth regulating and conserving. Though subordinate to waters from the Great Artesian Basin in amount, surface waters nonetheless supply a substantial part of the South Australian Arid Land area. The grazing industry relies on small ephemeral creeks to water stock (either directly via earth tanks and dams, or indirectly as recharge for bores drilled into alluvial sediments). The tourist industry relies on surface waters for visitor services and to underpin the beauties of the landscape which the visitors come to see. Aboriginal homelands need surface waters for human services and cultural values. Most significantly, all over South Australia, surface waters are the key element supporting ecosystems.

Legislation regulating the use of surface waters generally aims to ensure all stakeholders (including the environment) have reasonable access to the resource. At present there is little legislative framework governing the use of surface waters in the SA Arid Lands. In Australia, much existing legislation focuses on the extraction of irrigation water from large perennial river systems. Regulation of ephemeral creeks tends to be by simple rules of thumb. In New South Wales, landholders are allowed to build earth tanks (dams) on first- and second-order stream channels. In the Northern Territory, the owner/occupier of the property can extract water from a watercourse for stock without a license; permits are required for earthworks that halt or divert water, but rural dams <3 m high or with catchment <5 km2 are exempt. Western Australia appears to be in the process of formulating new water management plans.

Low Priority Given to Understanding Drylands Catchments Regulation of arid land surface waters is hampered by a generally poor understanding of drylands rivers. There is a general assumption that drylands rivers work in the same way as “normal” (by which is meant: temperate-zone, perennial) rivers, except they don’t work most of the time. The very words we use show that our understanding is grounded in our European heritage: “waterway”, for example, or the definition of “river” in Australia’s own Macquarie Dictionary as “a considerable natural stream of water flowing in a ... channel”.

Figure 34 The degree of accuracy given to understanding catchment in remote SA is much less than that given to populous places. A) Detailed capture of drainage divides in the Fleureau Peninsula: catchments (pink) and subcatchments (blue). B) Basins for all of SA; the area shown in A) is included (pink and blue). The water basins defined for the less populated parts of South Australia are not hydrologically correct. For example, the Gairdner basin groups together many smaller drainage networks, including some offshore ones, and the Lake Frome basin (orange) lumps together Lakes Frome, Callabonna, Blanche, and Eyre.

A B

It is understandable that population centres and large water resources have the lion’s share of organisational attention, however the result is that organisations (such as SAAL-NRM) which have to manage the largest area of land in the State have to do it with the least information. For example, online maps of South Australia’s surface water basins have catchments and drainage divides in great detail for e.g. the Fleurieu Peninsula (Fig. 34A), whereas the basins for the arid lands are not even correct (Fig. 34B). Similarly, aerial photography can be purchased from existing scans of negatives from areas close to Adelaide, whereas there are no resources available to scan negatives of remote lands, thus doubling the cost. SA inkjet prints of remote-area air-photos are six times more costly than actual photo prints from states with existing facilities.

South Africa, a nation equally concerned with managing arid lands, has recently begun to upgrade its surface water regulation. Their water board has begun by attempting to characterise the rivers. However, they have been told to use the basins that were defined (based on unknown criteria) some decades ago. These basins don’t correspond with the hydrological reality, so the task has been difficult and the results are unlikely to be worth the resources spent to achieve them.

Different Features of Drylands Rivers Drylands rivers are qualitatively different from temperate-zone rivers, with processes and landforms that may not be recognisable in terms of “normal” rivers. They are working when they are dry: the vegetation:landform relationships continue to affect landform even when the water is gone. The markers for waterway health are different.

Overland Flow, Not Flow Through Soil In a temperate river, rain falling on hillslopes infiltrates and makes its way to the river by horizontal flow through the soil. A channel downstream from a dam will receive, through the soil, some of the water that fell uphill from the dam. In the rangelands, the soils often seal after only a little rain, and overland flow – surface runoff – plays a much larger part in delivery of water to the river network. A channel downstream from a dam will probably receive nothing of the rain that fell uphill from it. A regulation that allows damming of every second-order channel in a river will not trap most of the

water in a perennial setting, but in an arid-zone river it could sequester most of the catchment’s water.

Arid Rivers Often Discontinuous Australian rangeland rivers are often discontinuous. What seems to be a string of small insignificant creeks in flat country can actually be a “discontinuous ephemeral stream” (Pickup 1995, Bull 1997, Wakelin-King & Webb 2007) – where the river landforms are repeating sequences of gullies, channels, and then unchannelled floodouts. The idea that a flat bit of ground without a channel can be a river reach is not commonly accepted in catchment management circles, yet discontinuous streams, separated by floodouts, are a very common landform in some parts of arid Australia. They are the dominant fluvial landform in Moonaree.

The NSW Water Act allows dams to be placed across first- or second-order channels. The channels are defined according to features displayed in listed 1:100,000 topographic map sheets. The problem with this is that the cartographers don’t recognise the discontinuous nature of the creeks, so drawing a single drainage network as two. Dams thrown across what are effectively third-order channels are threatening the viability of a downstream terminal wetland. There is the potential for water-rights disputes in the future.

Arid Rivers Depend on Episodic Flow Drylands rivers and their ecologies are designed to cope with flow variability. The occasional large floods that push high volumes of water far downstream or across the floodplains are just as valuable (sometimes more valuable) as smaller flows. Retaining floodwaters (for example, to irrigate crops) is not “good use of water otherwise wasted”, it deprives downstream ecosystems of their most important flow events. Dams which release excess water as a small trickle over a long time favour local ecosystems at the expense of distant ones. This is an ecological issue as well as a industry water rights issue.

Floodouts Floodouts are functional equivalents to riparian zones: they mediate erosion, store nutrients, are valuable ecological niches, and play an important role in maintaining

river health (Wakelin-King 2006, Wakelin-King & Webb 2007). However, lacking channels, they aren’t well-known in the professional sphere as fluvial landforms. They may not be recognised as deserving the same protection from erosion or the same access to Landcare funding.

Banded Vegetation Banded vegetation is a natural runoff-runon landform-ecology link that is common in some parts of arid Australia. Some examples include the mulga grove country in Central Australia, or the hillslopes bordering the Finke River (Wakelin-King 1999). Where there is banded vegetation, water flows as sheetflow across the slope. This landform has no place in the list of known fluvial landforms, but it transmits water and has relevance to groundwater recharge, erosion management, road alignment placement, and land use planning.

Erosion Can Propagate Downstream Generally, where erosion is triggered by some human activity, the erosion propagates upstream. Stream rehabilitation manuals are written around this sequence of events. In western NSW, there appear to be clear cases where an erosion trigger causes channel incision to propagate downstream instead. For example, at a creek crossing, a river-level dip in the road was replaced by a raised causeway with culverts transferring flow under the road. The culvert sizes were based on a temperate-zone expectation of the likely flow, and the result was flow concentration and erosion. The channel deepening proceeded downstream where it has killed the riparian vegetation and triggered lateral floodplain gullying. This downstream direction of action is not an accepted feature of stream rehabilitation, but it is consistent with the processes which create discontinuous ephemeral streams (see above).

Monitoring Waterway Health Standards for monitoring waterway health often use indicators such as water salinity and turbidity, or the heath and species diversity of fish or aquatic invertebrates. These are not relevant for rivers and creeks over an enormous part of Australia. Other indicators need to be developed for dry creeks, and for creeks that are dry along most of their length but which have semipermanent waterholes. The non-wet bits need to be managed also – this is an incredibly important yet almost totally

overlooked land management issue. Potential indicator factors could include riparian and floodout vegetation, herpetofauna, and bird life (though obviously for all these allowance for local drought history must be made). Other factors such as landform stability may also be relevant, as long as some understanding of the river’s natural processes is a starting point.

Human Activities that Affect River Flow

Dams impound water. Most dams have a way to get rid of excess water: some have pipe outlets which release a little water slowly, others have bywash channels. Both systems have advantages and disadvantages. Dams that are thrown across flat country, collecting in wide shallow basins, will lose a lot of water through evaporation. Water-capture earthworks, for example low ridges (bunds), divert floodplain water into the channel, usually upstream of a dam. Waterspreading and waterponding divert floodplain water from its course, either spreading it across another section of floodplain, or collecting it in shallow ponds. In either case, the aim is to encourage water to infiltrate the floodplain, rather than drain to the channel and be conducted away. Contour furrowing re-establishes vegetation on overgrazed areas. Where successful, it reduces runoff, which may reduce water entering channels and could be a useful control for erosion. Success of the furrowing technique needs accurate surveying and tailoring of the program for local conditions.

The above activities can be viewed in two ways. On the one hand they are valuable rangeland management techniques which increase productivity and in some cases are rehabilitation works which may restore rivers to something like their pre- European state. On the other hand they may deprive downstream users of water that might otherwise have come to them. There seems to be little attention given at the moment to the water-rights implications of some rangelands works. Resolving the issue may be complex: what is the logical starting point? The downstream user’s customary flow pattern, as of some recent date? In that case an upstream user may be prevented from repairing pre-existing damage. (For example, if an upstream reach had incised due to vegetation removal from overstocking, the downstream reach might get more water. If the upstream reach floodplain is rehabilitated by waterponding, the downstream reach gets less water yet the river as a whole becomes more healthy.) Knowing the pre-European state of a river would provide a

good starting point. Historical data can be ambiguous or hard to find; understanding river processes is likely approach. However, more research is needed.

Road construction and creek crossings are often triggers for downstream erosion, through flow concentration and gully initiation. In NSW the Road Transport Authority was only required to manage erosion along its 50m easement: that is, anything outside that easement is not their business. This is clearly not an ideal regulatory environment, as it encourages the displacement of erosion problems to the easement edge, rather than resolution of the problem. All over the Australian rangelands, station tracks also commonly concentrate flow and evolve into creeks (Figure 35) or initiate gully networks. These are hard to rehabilitate once formed. Continued investigation into, and support for, effective track construction and maintenance techniques should take place.

Figure 35 Aerial photograph of a long straight gully (dark line) developed from a station track in NSW. There is no natural surface drainage in this paddock (a mud gilgai plain, mottled orange and brown ground surface). Red scale bar = 1.6 km.

Recommendations for Surface Water Management

Small creeks and ephemeral rivers should be recognised as important contributors to the economy and amenity of South Australia’s rural sector, and as critical underpinnings of rangeland biodiversity. Regulation might be required to protect the interests of all users.

Management organisations should understand that dryland rivers and drainages do not necessarily operate in the same way as perennial rivers, nor necessarily look like them. Dryland fluvial landforms include discontinuous ephemeral streams, floodouts, and banded vegetation. Management policies should be developed from understanding South Australia's creeks, rather than derived from policies developed for temperate-zone rivers. As a first step, basic data should be collated about the types of river, and the true catchment boundaries around them. Historical information on stock route locations, pastoral ownership, and regional stocking rates would also be useful.

As a preliminary measure, the regulations limiting size of dam capture that are used in NSW (dams of 1st or 2nd order channels) or the NT (dam <3m high or across catchment <5 km2) may be useful. However, they should be applied with some judgement. The stream order rule shouldn’t be tied to a cartographer’s judgement as to what constitutes a single drainage; the <5 km2 rule shouldn’t apply to a tiny stream. Neither rule should apply to every single small lower-order catchment of a larger river, as that will cut off most of the water.

The authority that builds roads should have some incentive to care about erosion outside the road easement.

Measures to rehabilitate gullied station tracks, and to build and maintain erosion-free tracks, should be sought.

An awareness of downstream water rights, and their relationships with rehabilitation techniques such as waterponding and waterspreading should be part of discussions on surface water regulations. This is a complex issue and not one that will be resolved in a hurry.

Water rights should include support for ecology and biodiversity as well as human needs.

References

Anderson, Ruth. 1995 The McTaggarts; stories of a pastoral dynasty. Online extract , SA History Conference ( 2005) www.history.sa.gov.au, from The McTaggart Story : five generations of pastoralists. Openbook Adelaide.

Blissett, A.H., 1985. Gairdner, South Australia : Sheet SH/53-15. 1:250,000 geological series - explanatory notes. Geological Survey of South Australia, Adelaide.

Bull, W.B., 1997. Discontinuous ephemeral streams. Geomorphology 19: 227-276.

Dickinson, S.L., 1942. The Moonaree Station saline ground waters and the origin of the saline material. Transactions of the Royal Society of South Australia 66: 32-45.

Keeling, J.L., & Self, P.G., 1996. Garford Palaeochannel palygorskite. MESA Journal 1: 20-23 (South Australian Dept. Mines & Energy, Adelaide)

Pickup, G., 1985. The erosion cell - a geomorphic approach to landscape classification in range assessment. Australian Rangeland Journal 7: 114-121.

Wakelin-King, G.A. 1999. Banded mosaic ("tiger bush") and sheetflow plains: a regional mapping approach. Australian Journal of Earth Sciences 46: 53-60.

Wakelin-King, G.A. 2006. Landscape history controls vegetation ecology: formation of mid-creek floodouts in western NSW (abstract, poster). Australian Rangelands Conference, Renmark, SA, 3-7 Sept. 2006.

Wakelin-King, G.A. & Webb, J.A., 2007. Threshold-dominated fluvial styles in an arid-zone mud- aggregate river: Fowlers Creek, Australia. Geomorphology 85: 114-127.

Williams, G.E., & Gostin, V.A., 2005. Acraman – Bunyeroo impact event (Ediacaran), South Australia, and environmental consequences: twentyfive years on. Australian Journal of Earth Sciences 52: 607 – 620.

Map 1: Moonaree Paddocks and Major Creeks

Moonaree Paddocks and their names (in black) and the major creek lines and their names (in blue)

Map 3: Moonaree LANDSAT view

Lake Acraman

Lake Gairdner

Moonaree station viewed in LANDSAT 721pan, with paddock boundaries in green. Dacite hilltops are very dark brown, or purplish- brown. Dacite sand and soil is mid-brown in colour. The southern sand belt is pale blue (quartz sand), or brownish and bluish (dacite sand, near Lake Acraman), with dark hilltops emerging from the sand. Speckled dark brown and pale blue in Morinippi and Morinippi Outside shown a thin veneer of quartz sand over dacite rock.

Map 4: Moonaree as if the lakes were full to 200m ASL

Digital Elevation Model of Moonaree station viewed as if the lake levels were at 200m ASL (current shoreline is at 130m ASL). The grey area would have been under water, and only the main hilltops and higher central valleys would have been dry land. The areas with observed gilgai or swelling-clay sols are shown in green stars. Lake-full episodes in the wetter climates of previous geological ages are likely to have governed rock weathering and sediment deposition, influencing today’s water quality and paddock productivity.