 |
 |
Case studies of desertification assessment at two scales: paddock and regional
One of the difficulties in studying arid areas is the problem of extreme scales. How do
you accurately extend data from a small scale such as 160 km2 for a paddock to a large scale 38 000
km2 for a region?
This was addressed by:
1) Detailed studies at two locations 200 km apart (Kunoth Paddock and Knob Paddock), and
2) Desertification assessment of broad areas at the regional scale (an area of a path
covered by the Landsat satellite which is referred to as Region 102).
|
|
Techniques referred to as grazing gradient methods, which detect changes in vegetation cover over time and space, were used in this project to assess desertification. These methods used calibrated satellite data over a period of time (usually Landsat MSS or TM) to detect patterns of cover change with increasing distance from watering points (see diagram below). The satellite data used were acquired before and after large rainfall events when the vegetation had a considerable opportunity to recover from the effects of grazing. Grazing gradient analyses use several Geographic Information System (GIS) data layers: vegetation cover, distance from water and landscape types. [Detailed descriptions of these techniques using remotely sensed data can be found in the Details of the Research section on this CD-ROM.]
 |
|
|
Low cover close to water |
High cover far from water |
Vegetation cover
Indices derived from high resolution satellite data provide an ideal way of studying
vegetation dynamics through time. The PD54 vegetation index (Pickup et al., 1993)
is used in this project as it provides good discrimination between bare soil, dry
vegetation and green vegetation in arid areas. The index is calculated from
satellite data in the visible green and visible red bands. Below are images of the
PD54 index of vegetation cover of Kunoth Paddock calculated from Landsat TM data.
The colour scheme is from red (low cover) through yellow, green, pale blue to dark blue
(highest cover). [PD54 images from SPOT XS and JERS-1 OPS are
found in the satellite methods section of this CD-ROM]
| December 1994 - 'dry' | March 1995 -'wet' after very heavy rainfall |
 |
 |
 |
Below are PD54 images for the Region 102 acquired before and after a moderate rainfall event.
Spatial data layers
Distance from water
In arid areas, cattle grazing is centred on artificial sources of water such as bores, wells and dams. In addition, fences control where animals graze. Large hills and mountain ranges also act as barriers to cattle movement. Thus, if the sites of watering points are linked with fences and natural barriers, the distance of all locations from the nearest source of water can be determined. Below is a windmill used to pump water for cattle to drink at a 'water point'. The areas marked as red circles (in the distance from water images below) are watering points. Click on the images to see full-size.
 |
|
|
| Windmill at cattle watering point | Region 102 |
Kunoth Paddock |
Landscape types
As cattle selectively graze where the vegetation is most palatable, we need to be able
to analyse landscape types separately. The different landscape types within a region are
mapped and included in the GIS so that grazing gradient analyses can be made on each
landscape type separately. Landscape types in central Australia at their broadest
level include floodplains, ranges, hills, calcareous plains, timbered plains and
sandplains. [Photographs of each landscape type and
descriptions are found in the GIS layers section of the CD-ROM.]
|
|
| Region102 | Kunoth Paddock |
Results
1) Paddock scale
Kunoth Paddock was used as the first test site because it is an area where CSIRO has collected a large amount of information about the different landscapes over the past 25 years. Of particular interest has been how the vegetation present influences where cattle graze and, in turn, what effect this grazing has on the vegetation. Landscape types present in Kunoth Paddock range in area from 1.8 km2 to 40 km2. [Detailed explanations are given in the satellite data section of the CD-ROM.]. Grazing gradient techniques tested in this paddock were subsequently extended to Region 102 (see below) where grazing gradients were derived for selected landscape types.
Another way of assessing the effects of cattle grazing is to use the Resilience Method. This method produces a map showing how pasture growth following good rains varies across a paddock. [The example for Knob Paddock explains how the results might be used to help pastoralists in their land management. This is located in the "results using satellite data" section on the CD-ROM ]
We appreciate the ready access that the owners of both Hamilton Downs (Kunoth Paddock) and Erldunda (Knob Paddock) have allowed CSIRO and the Japanese collaborators during this GRNS project.
2) Regional scale - Region 102
Grazing gradients were calculated on 26 landsystems within the Region 102 ranging in area from 67 km2 to 3,137 km2. A land system is a way of grouping areas containing similar soil and vegetation. Each land system was analysed separately and two examples of the results obtained are presented here. [Details are in the satellite data section of the CD-ROM.] The distribution of each land system is shown in red on each map. Each graph shows average cover levels present at increasing distance from water before (red line) and after (green line) a moderate rainfall event. The Y axis is scaled in units of PD54 cover between 0 and 254: values below 60 equate with bare ground while values above 240 represent 100% cover. The horizontal blue (or black) line indicates the expected cover level if vegetation close to water had fully recovered from the effects of grazing following the rainfall received.
Todd Land System (alluvial plain)
 |
 |
Todd land system describes alluvial plains and floodouts of the Todd River and other major streams immediately south of the MacDonnell Ranges. The main landscape type is an open woodland and the land system thus has a moderate cover compared with more timbered land systems. Todd land system grows palatable forage and is a highly productive grazing area for cattle. The graph shows that there is a consistent trend of increasing cover to 7 km from water in the dry period. Cover levels increase after rain but the grazing gradient persists indicating that the vegetation shows minimal recovery from the effects of grazing.
We have used ground-based techniques to measure the different components of total vegetation cover at increasing distance from one bore in Todd land system. At this location, there was more bare ground close to the bore (see pair of photos above adjacent to the grazing gradient diagram). Annual grasses and forbs made up a higher proportion of the lower cover closer to water while significant amounts of palatable perennial grass were only found beyond about 5 km from water. Management techniques which encourage the regeneration of palatable forage species, including perennial grasses, would seem necessary in some paddocks to allow vegetation closer to water to respond to rainfall in a similar manner to that on more distant areas. Such management includes lighter stocking, periodic resting from grazing and construction of ponding banks to revegetate eroded areas (see Purvis, 1986 for a description of how these techniques have been successfully used to improve the productivity of one central Australian cattle station).
Singleton Land System (sandplain)
 |
 |
Large areas of spinifex sandplain have been mapped into Singleton land system. Hummocks of spinifex and scattered shrubs mean that the land system has a moderate cover although cover is temporarily reduced following either wildfire or a controlled burn. Most of the vegetation present is unpalatable to cattle and grazing has a negligible effect on average cover at increasing distance from water. Average cover levels increase over the first km from water in both the dry and wet periods and then increase more gradually to 12 kms. Cover increase beyond 1 km is due to natural variation in the country analysed rather than due to grazing.
The infertile sands mean that there is little that pastoralists can do to improve the grazing value of Singleton land system. Controlled burns temporarily remove the harsh spinifex and encourage the growth of shorter-lived colonizing species, some of which are marginally palatable to cattle. However, the spinifex sandplains do support very large numbers of termites and have a diverse lizard population and thus have an important role in biodiversity.
Conclusion: The grazing gradient method developed by CSIRO and tested under GRNS is able to measure the effects of grazing on different landscape types. The technique is thus useful for monitoring the condition of very large areas of grazed rangeland.
