Wind Erosion Trends From Meteorological Records

Australia: State of the Environment Technical Paper Series (Land), Series 2
Grant McTainsh, John Leys and Kenn Tews
Department of the Environment and Heritage, 2001
ISBN 0 642 54779 3


  1. Project Aims
  2. Description
  3. Data and Methods
  4. Results and Discussion
  5. References

1. Project Aims

This project aims to quantify spatial and temporal trends in wind erosion in Australia during 1996-1999. It aims to differentiate accelerated wind erosion caused by agricultural and pastoral activities from "naturally occurring" wind erosion. This methodology is a refinement of the approach taken for the National Collaborative Project on Indicators for Sustainable Agriculture report (NCPISA).

2 Description

Broadscale wind erosion rates can be approximated using meteorological data which reports the frequency of occurrence of dust events (dust storms and local dust events) weighted according to their intensity as a Dust Storm Index (DSI) (McTainsh, 1998). Changes in wind erosion rates are measured in both space and time using the DSI and the impacts of land use are approximated by comparing measured DSI values with the rates predicted by a wind erosion model (Ew model) (McTainsh et al. 1990).

The Ew model is an indicator of wind erosion under natural conditions, without the impact of land use activities. The model predicts the level of wind erosion by utilising both wind erosivity (the power of the wind to erode) and land type erodibility (the susceptibility of the land to erosion). Wind erosivity is measured using wind run data, and land type erodibility is measured using precipitation and evaporation parameters to estimate effective soil moisture, which when averaged annually, approximates vegetation cover. Soil erodibility is not separately described in the model.

The Ew model predicts the wind erosion activity for a location, and when the measured wind erosion (DSI) exceeds the predicted value, wind erosion is evidently occurring at a faster rate than would be expected under the climate and vegetation conditions of the area. In this instance accelerated wind erosion is assumed to have occurred, and is quantified by the ratio of the predicted and measured values, called the Accelerated Erosion Index (AEI). As soil erodibility is not described by the model, a component of AEI may be due to differences in soil erodibility. Spatial variations in AEI may therefore be partly influenced by soil erodibility, but at this generalised level of analysis it is unlikely that temporal changes in AEI are influenced by changing soil erodibility at a location.

3. Data and Methods

3.1 Data sources

Data on dust event type, wind speed, temperature and rainfall were obtained from the Bureau of Meteorology (BoM) in a pre-processed format. The dust event data used were the present and past BoM weather records for: dust storms (codes 09, 30-35), thunderstorms with dust storms (code 98) and local blowing dust (BoM codes 07 and 08). Quality control flags, which indicated the reliability of that record, accompanied each record and only the reliable data were used. High quality data were available for 109 stations (Fig. 1). At locations, where there have been more than one weather station operating over the period of the study, a composite record was made up for the location. Only data up to 1999 can be used here as, in 2000 the Bureau of Meteorology started converting stations to automatic monitoring which means that routine observations of phenomena such as dust storms are not made.

Wind data were received as the mean daily wind speed (m/s) per station. The height at which wind speed is recorded differed between stations. To standardise the wind speed height of measurement, all wind speeds were converted to the 10 metre height equivalent using the 1/7 power equation of Elliot (1979).

The temperature data were received as the mean maximum and minimum temperature per month for each station for the period 1960 to 1999. Rainfall data were received as monthly rainfall per station for the period 1960 to 1999. Missing values were replaced by the overall mean value for the month, if there was more than 4 consecutive missing values the station was rejected.

3.2 Analysis and interpretation

Data processing was completed using the statistical program SAS (v6.12) and maps were produced using MapInfo (v6.0) with Vertical Mapper (v2.5). Interpolated values between stations were derived using a geometric estimation technique.

3.3 A comparison of methodologies used in the NCPISA and SoE-2001 surveys

Although the basic methodology used here is the same as for the National Collaborative Project on Indicators for Sustainable Agriculture report (NCPISA), there have been significant improvements in the quality of data received from the Bureau of Meteorology (BoM), and an improved climatic model of wind erosion is used. For these reasons the data output from the present analysis will be slightly different from those of NCPISA. To demonstrate the effects of these improvements, new DSI and AEI maps are provided for the NCPISA reporting period (1986-1996).

3.3.1 Dust Storm Index (DSI) (Fig.2)

The general pattern of DSI is similar to NCPISA (McTainsh, 1998, Fig.5.7), but there are a number of significant differences which are explicable in terms of: errors in the BoM data, and addition or removal of stations. The apparent hot spots in the NCPISA map at Bencubbin (SW WA) and Croydon (NW Qld) are in fact artefacts of the BoM data provided to us. BoM has confirmed the problems with the Bencubbin data, and Croydon had the same problem. The new DSI record (Fig.2) shows a hot spot on Eyre Peninsula, which results from the addition of data from Kyancutta (a station not available for the NCPISA analysis).

3.3.2 Accelerated Erosion Index (AEI) (Fig.3)

There are differences between the new AEI map (Fig.3) and the equivalent NCPISA map (McTainsh, 1998)., Fig.5.9) This is most likely due to differences in the BoM data and/or the improved model (Ew model) used here. The Ew model measures effective soil moisture (as a surrogate for vegetation) and wind erosivity, whereas the Em model, of Burgess et al. (1989), used in the NCPISA report does not measure wind erosivity. Therefore, some of the variance in the relationship between measured wind erosion and that predicted by the Em model for a location, could be due to such inadequacies of the Em model, rather than land use effects. The differences at: Croydon in north Queensland, SW WA, and Eyre Peninsula, are the result of the BoM data issues described earlier, whereas the more extensive area of accelerated erosion in the Lake Eyre Basin, the accentuated hot spot at Thargomindah, and the more pronounced area of accelerated erosion in the Balranald-Hay area are likely to result from the improved data and model.

4. Results and Discussion

4.1 Spatial patterns of wind erosion 1996-1999 (Fig.4)

The main region of wind erosion during 1996-1999 is bounded by the 500mm rainfall isohyet (1975-1999) (Fig.5), which includes most of the Lake Eyre Basin and the lower Murray-Darling Basin. Western Australia has less active wind erosion (Fig.5). The areas with higher probable accelerated erosion tend to be found within the semi-arid zone (500 to 300mm rainfall) (Fig.6), rather than in the arid heart (<300mm rainfall) of the Lake Eyre Basin. The hotspots are around: Thargomindah (in SW Queensland) Tibooburra and White Cliffs (in NW NSW), the Hay-Balranald-Mildura, region in SW NSW, Eyre Peninsula in SA, and Southern Cross in the SW and Marble Bar in the NW of WA. These areas have a combination of erodible soils and land uses that accelerate wind erosion. Cattle grazing is the main land use in SW Queensland and NW NSW and NW WA, whereas all other areas have mixed farming enterprises (wheat / sheep grazing). The high AEI values in SW Queensland and north west NSW may also, to a certain extent, reflect the active wind erosion occurring in the Strzelecki and Simpson Desert dunefields to the west, because most dust storms in that region originate from the western sector.


4.2 Temporal changes in wind erosion (1994 and 1999)

As the 1990's progressed, climatic conditions ameliorated, thus providing an opportunity to demonstrate the interacting effects of changes in climate as well as land management practices on wind erosion. A comparison between 1994 and 1999 is included to emphasise the temporal nature of wind erosion in Australia. 1994 was chosen because drought conditions were widespread in eastern and southern Australia at this time (Fig.7) and the mean annual SOI was strongly negative (-11.9).

In 1994 there was very active wind erosion in the Lake Eyre Basin region, the Mallee region, Eyre Peninsula and SW WA (Fig.8). In response to the drought conditions in eastern and southern Australia, wind erosion spread east into central Queensland and New South Wales, and south into Victoria. From 1994 to 1999, overall climatic conditions ameliorated, with persistent drought only in southern Victoria. Wind erosion activity decreased during this period and contracted to the south (Fig.8 and Fig.9), largely in response to changes in rainfall, leaving the arid and semi arid western region of the continent more stable. In contrast, the semi arid regions of SA east of Port Pirie and Marree had a slight increase in AEI. Persistent drought in Victoria, however appears to have maintained erosion activity in the Mallee area and southern NSW.


The areas experiencing accelerated wind erosion during 1994 were quite localised, with hot spots around Thargomindah (SW Queensland), Balranald and Hay - with an outlier at Parkes (NSW), Eyre Peninsula (SA) and SW WA (Fig.10). By 1999 the northern locii of accelerated wind erosion shifted east and south to White Cliffs (Fig.11). The hot spot in SW WA is dramatically reduced, but the rate of erosion increased in the Mallee of Victoria with a new locus appearing around Ouyen in 1999.


4.3 Possible land management effects

4.3.1 Rabbits

One of the major management-induced changes in semi arid regions in the late 1990s resulted from the release of the Rabbit Calisi Virus (accidentally in 1995 and planned in late 1996), which reduced rabbit populations to as low as 15-20% in 1998. This effect was most pronounced in the Strzelecki and Simpson Deserts regions (Neave, 1999) which witnessed widespread vegetation recovery, aided by increased rainfall. While the southward shift in the locus of wind erosion (Fig.8 and Fig.9) is very likely to be in response to rainfall-induced improvements in vegetation cover, that the accelerated erosion index also moved south (Fig.10 and Fig.11) indicates that wind erosion rates were reduced by more than can be explained by climatic conditions. This evidence lends support to the interpretation that rabbit depopulation may have resulted in measurable reduction in accelerated wind erosion in the semi-arid regions of SW Queensland, northern SA and far western NSW.

4.3.2 Conservation-oriented agricultural methods

During the late 1990's expansion of conservation-oriented agricultural methods continued in cereal cropping regions. The reduction in wind erosion in SW WA since 1994 (Fig.8 and Fig.9) is accompanied by increases in rainfall, but as the accelerated erosion index in this region also decreases (Fig.9 and Fig.10) erosion rates have evidently been reduced by more than would be expected from climatic conditions alone, which may indicate that conservation-oriented land management strategies are reducing wind erosion. Both total wind erosion and the accelerated component decrease in the Eyre Peninsula of South Australia, suggesting a more subdued land management effect, whereas in the Mallee area and further east in central NSW both total wind erosion rates and the accelerated component have remained reasonably static with the exception of parts of western Victoria.


Elliot, D.L. (1979) "Adjustments and analysis of data for regional wind energy assessments". In Proc. Workshop on Wind Climate, Asheville. N.C. 12-13 November.

McTainsh, G.H., Lynch, A.W. and Burgess, R.C. (1990). "Wind erosion in Eastern Australia". Australian Journal of Soil Research, 28(2), 323-339.

McTainsh, G.H. (1998). "Dust Storm Index". In: Sustainable Agriculture: Assessing

Australia's Recent Performance. A report of the National Collaborative Project on Indicators for Sustainable Agriculture. SCARM Technical Report 70, 65-72.

Neave, H.M. (1999) "Overview of effects on Australian wild rabbit populations and implications for agriculture and biodiversity." Report 1: Rabbit Calicivirus Disease Program, Bureau of Rural Sciences, Canberra.