Australia Australia

Description of cropping systems, climate, and soils

Wheat production in Australia accounts for 55% of the total cropland, averaging 12.6 million ha over the 14 seasons from 1998-9 to 2011-12 (ABARES, 2012). Over the same period, the wheat crop area has increased (at the expense of pastures and other crops) by an average 1.5% per year, but production has increased by 1.0% per year. Globally, Australian wheat averaged approximately 3.5% of world production over the same period, and 12.1% of global wheat exports (2005-2012).

Wheat crops are commonly grown in rotation with break crops such as pulses, canola, lupins, barley, oats and sorghum. Wheat is also grown in crop-pasture rotations and a small but increasing number of wheat crops are grazed during their vegetative phase. A survey of farm practices with 1312 grain producers found that in 2011, the average farm size was 3810 ha. On average 61% was cropped, 33% was maintained as pasture and 5% was reported as native vegetation. Of the cropped area, 58% was planted to wheat, 14.5% to barley, 4.3% to other winter cereals and 3.6% to summer cereals. 9.6% of the cropped area was planted to winter oilseeds and 6.8% to pulses. More than 75% of the cropped area was planted where a crop was present in the preceding year. The percentage of the crop planted using zero or no-tillage was 60%. The use of controlled traffic was 20%. A similar area has been yield mapped but only 10% of the area used variable rate technologies (GRDC 2012).

Compared to Midwestern America and Europe, only one tenth to half the amount of fertilizer and agrichemicals are used per hectare of cropland (Price 2006). While it required about 5.9 ha to feed one person, one unit of agricultural labour (counting farmers plus their employees) annually feeds about 109 people. A defining characteristic of Australian agriculture is that it is predominantly a low input, low output system, where labour efficiency is extremely high (ABS 2010; Hochman et al., 2013).

Table 1. Australian wheat harvested area, average yield and total production over 14 seasons (1998-9 to 2011-2). Source: (ABARES, 2012).

Crop
season

Area
('000 ha)

Yield (t/ha)

Production (Mt)

1998–99

11543

1.86

21.5

1999–2000

12168

2.03

24.8

2000–01

12141

1.82

22.1

2001–02

11529

2.11

24.3

2002–03

11170

0.91

10.1

2003–04

13067

2.00

26.1

2004–05

13399

1.63

21.9

2005–06

12443

2.02

25.1

2006–07

11798

0.92

10.8

2007–08

12578

1.08

13.6

2008–09

13530

1.58

21.4

2009–10

13881

1.57

21.8

2010–11

13502

2.03

27.4

2011–12

13963

2.14

29.9

Average

12622

1.69

21.5

 

Australia is known for a variable climate and highly variable rainfall which results in substantial variation in wheat yields. For example average yield nationally in 2002-3 was 0.91 t/ha, whereas in 2011-12 it was more than double that at 2.14 t/ha (Table 1). In those years, total national wheat production was 10.1 Mt and 29.9 Mt.

Average annual rainfall in Australia ranges widely, with some areas receiving less than 200 mm while other areas average over 3200 mm. Rainfall is generally higher towards the coast, and is influenced by elevation, with mountain areas in northeastern Queensland, southeastern Australia and western Tasmania receiving higher rainfall totals. In the northern areas of Australia, most rainfall is associated with monsoonal patterns and occurs in the summer months. Wheat is primarily a winter crop and is mainly grown in southern Australia where most rain falls during the winter months. The majority of wheat cropping is located in areas receiving between 300 mm and 600 mm annually, with 46% receiving 300–400 mm, 24% receiving 400–500 mm and 17% receiving 500-600 mm. Only 4% of wheat cropping occurs in drier areas, and 9% in wetter areas (600-800 mm) (ABARE-BRS, 2010; BoM, 2009).

The major soils in the wheat cropping areas are shown in Table 2. Australian Soil Classifications (ACLEP, 2012) were used as the basis for crop yield simulations, and Soil Taxonomy Order equivalents are provided for comparison.

Table 2: Australian Soil Classification and Soil Taxonomy Order (Isbell, 1996) of major soils in the wheat cropping areas.

Australian Soil Classification (ACLEP, 2012)

Soil Taxonomy Order
(Isbell 1996)

% of area

Vertosols

Vertisols

23%

Sodosols

Alfisols, Aridisols

20%

Calcarosols

Aridisols, Alfisols

16%

Chromosols

Alfisols, some Aridisols

15%

Kandosols

Alfisols, Ultisols, Aridisols

12%

 

Data Sources and Assumptions (following GYGA protocols)

Yield gap calculations are based on 15 years of data. This period is long enough to account for climate variability but short enough not to be substantially affected by technology change and climate change.

Harvested area and actual yields

The Australian Bureau of Statistics (ABS) collates national agricultural data at the level of statistical division (SD) annually, and at the finer scale of statistical local area (SLA) every five years when a census is carried out. SDs are relatively uniform regions that cover the whole of Australia, and are characterized using socioeconomic criteria. SLAs are subdivisions of SDs that are based on local government areas, but also cover unincorporated areas which are not within local government areas. There are over 30 SLAs per SD on average across the wheat/sheep zone. Data on annual crop harvested area and average yields for the years 1996 to 2010 were sourced from ABS (2012) at SLA level for census years (1996, 2001, 2006) and at SD level for intervening years. For years where only SD level data were available, crop yields (t/ha) were downscaled from SD to SLA level using linear regressions fitted to SLA level data from 17 past census years.

Spatial datasets showing the location of specific crop types (such as wheat) are not available at a national scale. The most suitable dataset is the ABARE-BRS (2010) National Land Use of Australia version 4 (2005-6) available at http://www.daff.gov.au/abares/aclump/pages/land-use/data-download.aspx. This provides a ‘cereals' land use class of 20,812,366 ha which was used to indicate the distribution of wheat cropping. Making a simplifying assumption that wheat comprised a fixed proportion (59.79%) of cereal cropping over the entire mapped cereals area, those areas were adjusted to the 2005/6 wheat area of 12,443,000 ha (ABARES, 2012).

Weather data and reference weather stations

The Australian Bureau of Meteorology (BoM) manages a network of meteorological stations for which daily data are available. Within the wheat zone, 19 BoM stations were available which have maintained an ongoing rainfall, temperature and evaporation data for at least the previous 20 years and which we used as the primary meteorological stations. In addition data for 2562 Patched Point Dataset (PPD) sites were used as secondary meteorological stations. PPD stations are those at which only limited variables are recorded (e.g. only rainfall) or provide incomplete temporal coverage. The PPD station data comprise of actual recorded observations where available, and are in-filled with interpolated data (Jeffrey et al., 2001) to cover missing periods and attributes. Daily solar radiation is recorded at relatively few stations, and was sourced as interpolated data for both primary and secondary stations.

Following the GYGA protocol (van Wart et al., 2013a), we selected six Global Yield Atlas Extrapolation Domain (GYGA-ED; van Wart et al., 2013b) agro-climatic zones (CZs) each of which contain at least 5% of the national cereals cropping area (Table 3), and which together cover 78.8% of the total cropping area. Within these six CZs, 22 reference weather stations (RWS) were selected using a Python geoprocessing script run in ArcGIS (Environmental Systems Research Institute, 2010). The script iteratively selected RWS from lists of candidate meteorological stations as follows. The primary meteorological stations were used as the first set of candidates, then the list of secondary stations:

  1. Each candidate meteorological station was allocated a ‘buffer zone' defined by a 100 km radius, and which was clipped to the boundary of the CZ in which each station is located.
  2. The station with the largest area of cereal cropping within its buffer zone was added to the list of selected RWS.
  3. Eliminate any candidate stations if their buffer overlaps those of selected RWS by more than 25%.
  4. Repeat steps 1 to 3 until no more candidate stations are available.

The combined cropped areas within the 22 RWS buffer zones covered 50% of the national cereals area. Actual yields (Ya) were allocated to each RWS based on annual yield data from SLAs covered by the RWS buffer zones. Each RWS buffer zone partially or wholly covers more than one SLA. The contribution of each SLA to the Ya of a buffer zone was therefore weighted by its ‘cereal' land use area within that zone.

Table 3. Proportion of national cereals area covered by the six selected GYGA-ED zones.

GYGA-ED Zone (CZ)

% of cereals area

5102

15.0%

5202

5.6%

6002

25.8%

6102

17.4%

6202

6.7%

7102

8.4%

Total

78.8%

 

Soil data

For water-limited yield potential simulation we relied on characterized soils from the APSoil database (http://www.apsim.info/Wiki/APSoil.ashx; Dalgliesh et al., 2009). In order to select suitable characterized soils for cropping simulations, ArcGIS analysis tools were used to summarize the area of cropped soils within each RWS buffer zone using national soil grids (ACLEP, 2012). Based on these summaries, three soil types were selected within or as close as possible to each RWS buffer zone which matched, as best as possible, Australian Soil Classification (ASC) order, texture, plant available water capacity (PAWC) and bulk density of the three main soils within the RWS buffer zone.

Crop system and management information for crop simulations

Water-limited yield potential was simulated using the APSIM (http://www.apsim.info/Wiki/Downloads.ashx; Keating et al., 2003) modeling framework. APSIM is a daily time-step cropping systems simulator that incorporates meteorological data, crop growth and soil water and nutrient models. APSIM is well validated for Australian wheat crops (e.g. Hochman et al., 2009; Carberry et al., 2013) and has been partially ground tested for yield gap assessment (e.g. see Hochman et al., 2012). A large number of wheat varieties have been parameterized for APSIM, and the model allows sowing and fertilizer rules to be flexibly specified.

Since starting soil moisture conditions at each site were unknown, values were arbitrarily set for 1981 and continuous wheat- summer fallow- wheat simulations were run from 1981 to 2010. Cycling through a sequence of crops experiencing wetter and drier seasons and fallows, causes soil moisture to self correct over a few seasons. This cautious approach enabled the simulated soil moisture conditions to stabilize over 14 years of simulations from 1981 to 1995.

Wheat varieties

Five wheat varieties (Mace, Scout, Derrimut, Endure and Bolac) representing progressively slower maturity types, were simulated for each soil type and RWS to examine which are best suited to each RWS. After sowing and fertilizer application rules were applied, the simulated yields of the variety which produced the highest average yields for each site over the 15 year period was chosen to represent Yw.

Sowing rule

Two different sowing rules based on local expertise were used depending on the RWS region. For northern sites (n=9) the rule was set as follows:

Sow if at least 15 mm or more of rainfall over 3 days and PAW>=30 mm from 26 April-15 July

For western and southern sites (remaining sites)

Sow if at least 15 mm or more of rainfall over 3 days regardless of soil moisture from 26 April-15 July

For all site simulations the crop was sown on 15th July if the above criteria were not met during the sowing window. Sowing density was 150 plants m-², row spacing was 0.25 m and sowing depth was 3 cm.

Fertilizer rule

Unlike with some other crop models, it is not possible to turn the nitrogen module off in APSIM in order to simulate a nutrient unlimited yield. This is because there are many dependencies between the soil N and soil water and surface OM modules. Without the N module residue breakdown will be affected which in turn would affect the water balance. Instead we use fertilizer application rules (see rules below) that ensure that there is at least 50 kg/ha N in the top 60 cm up to the ‘first awns visible' growth stage. This means that soil N rarely limits crop yield. You can still have N stress but it is due to lack of soil water where the soil nitrogen is plentiful. This rule is sensitive to the problem of ‘haying off' - a situation where excessive vegetative growth leads to lower yields when water stress occurs during grain-filling. After experimenting with variations to the fertilizer rules with higher rates of N, we are confident that these rules result in the best estimate of water limited and N unlimited wheat yields.

If average yield for last 15 years (based on simulations 1996-2010) < 2 t/ha then:

At sowing, add 70 kg/ha NO3 minus soil nitrate in top 60 cm of soil on April 25th

Check top 60 cm soil daily, if NO3<50 kg/ha and PAW>30 mm and Zadok growth stage < 49 then add 50 kg N/ha

If average yield for last 15 years (1996-2010) >= 2 t/ha then

At sowing, add 100 kg/ha NO3 minus soil nitrate in top 60 cm of soil on April 25th

Check top 60 cm soil daily, if NO3<70 kg/ha and PAW>30 mm and Zadok growth stage < 49 then add 70kg N/ha

Potential yield (Yp) estimation

To estimate potential yield in the absence of moisture constraints, the APSIM simulations used for Yw (as described above) were modified to reset soil water to drained upper limit at the start of each day of the simulations. Additionally, to ensure that abundant soil nitrate was available, soil N levels were reset at the start of each day. In this way, crops were simulated as though fully irrigated.

Review of RWS results

On review of the final APSIM outputs, twenty of the RWS zones showed Y% values in the range 36.9% - 67.3% while RWS results for Kyancutta RWS and White house Farm RWS  showed anomalously high Y% values. For Kyancutta, where average Yw values were much lower than the Ya values, we re-calculated Yw by replacing the Kyancutta meteorological data with data from Warramboo, about 14 km away and well within the RWS buffer zone. The re-calculated Yw values were substantially higher in all years. Although the Kyancutta met station used originally was one with good long term temperature and rainfall data, it was apparent that it was not as representative of the cropping region in the RWS buffer as the rainfall-only Warramboo met station. The Warramboo station data were therefore utilized for the final Kyancutta RWS buffer Yw predictions instead.

The White House Farm RWS results gave a Ya value that was 88% of the Yw value. For this RWS we identified that the soil we had selected from the APSoil database  to represent the dominant soil in the RWS buffer area had severe subsoil constraints which are more extreme than common for this soil type. APSIM simulations were re-run, after replacing this soil with a generic soil of the same soil type, which was derived from the median values of all characterized soils of this soil type. The re-calculated Yw estimates increased substantially, and those revised simulation results were utilized for the final analysis.

Aggregating simulation outputs

The outputs of each RWS x soil combination were weighted by the proportion of soils estimated for each RWS buffer zone. Final estimates of Ya, Yw, Yg and Y% were derived by aggregating the 15 year average Ya and Yw for each RWS within the CZs, weighted by the cereals area of each RWS within the CZ. Then the Ya and Yw for each CZ were aggregated to a national figure, weighted by the proportion of national cereals area within each CZ.

References:

ABARE-BRS (2010). Land Use of Australia, Version 4, 2005-06.  Australian Bureau of Agricultural and Resource Economics - Bureau of Rural Sciences (ABARE-BRS). Australian Collaborative Land Use and Management Program (ACLUMP).

ABARES (2012). Agricultural commodity statistics 2012. Canberra: Australian Bureau of Agricultural and Resource Economics and Sciences.

ABS, 2010. Agricultural Commodities, 2008-09, Australia Australian Bureau of Statistics Catalogue No. 7121.0, Canberra.ABS (2012). Agricultural Census: Value of Agricultural Commodities. Australian Bureau of Statistics.

ACLEP (2012). Australian soil classification (ESRI Grid).  Australian Collaborative Land Evaluation Program (ACLEP) endorsed through the National Committee on Soil and Terrain (NCST).

BoM (2009). Mean annual rainfall data (base climatological data sets).  Commonwealth of Australia (Bureau of Meteorology).

Carberry, P. S., Liang, W.-l., Twomlow, S., Holzworth, D. P., Dimes, J. P., McClelland, T., Huth, N. I., Chen, F., Hochman, Z. & Keating, B. A. (2013). Scope for improved eco-efficiency varies among diverse cropping systems. Proceedings of the National Academy of Sciences 110(21): 8381-8386.

Dalgliesh, N. P., Foale, M. A. & McCown, R. L. (2009). Re-inventing model-based decision support with Australian dryland farmers. 2. Pragmatic provision of soil information for paddock-specific simulation and farmer decision making. Crop and Pasture Science 60(11): 1031-1043.

ESRI (2010). ArcGIS 10.0. Redlands, CA: Environmental Systems Research Institute.

Hochman Z, PS Carberry, MJ Robertson, DS Gaydon, LW Bell, PC McIntosh. 2013. Prospects for ecological intensification of Australian agriculture. European Journal of Agronomy, 44, 109-123.

Hochman, Z., Gobbett, D., Holzworth, D., McClelland, T., van Rees, H., Marinoni, O., Navarro Garcia, J. & Horan, H. (2012). Quantifying yield gaps in rainfed cropping systems: an Australian case study Field Crops Research 136: 85-96.

Hochman, Z., Holzworth, D. & Hunt, J. R. (2009). Potential to improve on-farm wheat yield and WUE in Australia. Crop and Pasture Science 60(8): 708-716.

Isbell, R. F. (1996). Australian Soil and Land Survey Handbook: The Australian Soil Classification. CSIRO Australia.

Jeffrey, S. J., Carter, J. O., Moodie, K. B. & Beswick, A. R. (2001). Using spatial interpolation to construct a comprehensive archive of Australian climate data. Environmental Modelling & Software 16(4): 309-330.

Keating, B. A., Carberry, P. S., Hammer, G. L., Probert, M. E., Robertson, M. J., Holzworth, D., Huth, N. I., Hargreaves, J. N. G., Meinke, H., Hochman, Z., McLean, G., Verburg, K., Snow, V., Dimes, J. P., Silburn, M., Wang, E., Brown, S., Bristow, K. L., Asseng, S., Chapman, S., McCown, R. L., Freebairn, D. M. & Smith, C. J. (2003). An overview of APSIM, a model designed for farming systems simulation. European Journal of Agronomy 18(3–4): 267-288.

Price, G.H., 2006. Plant nutrients in the Environment (Table 10.1, Chapter 10). In: Australian Soil Fertility Manual, 3rd ed. CSIRO. Melbourne, 176 pp.

van Wart, J., Kersebaum, K. C., Peng, S., Milner, M. & Cassman, K. G. (2013a). Estimating crop yield potential at regional to national scales. Field Crops Research 143(0): 34-43.

van Wart, J., van Bussel, L. G. J., Wolf, J., Licker, R., Grassini, P., Nelson, A., Boogaard, H., Gerber, J., Mueller, N. D., Claessens, L., van Ittersum, M. K. & Cassman, K. G. (2013b). Use of agro-climatic zones to upscale simulated crop yield potential. Field Crops Research 143: 44-55.

 

Go to the Atlas

Get access to the Atlas for advanced users

 

Download GYGA results

All downloads are available under the Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported (CC BY-NC-SA 3.0) license.

Data from the Global Yield Gap Atlas is available for commercial use. We have different commercial license types available, either for a single use of the data within a project context or multiple use and commercialization, suiting your needs. Please email Patricio Grassini (UNL) or Martin van Ittersum (WUR) to discuss options.

Country agronomists Country agronomists


Zvi Hochman

Project leader, Agronomist

http://www.researcherid.com/rid/E-8993-2010

 


David Gobbett

GIS and data analysis

http://www.researcherid.com/rid/F-8910-2010

 


Heidi Horan

Cropping systems simulation

 


Di Prestwidge

Data management and analysis

http://www.researcherid.com/rid/D-3392-2011

 


Javier Navarro Garcia

Systems modelling and Data analysis