Variation in irrigation requirements of forages in Northern Victoria

The dairy industry in northern Victoria relies on irrigation water to grow a large proportion of its feed inputs. But, as a result of drought conditions in the last 15 years, annual irrigation allocations have been substantially lower and more variable than the 20 to 30 years before that. This has caused dairy farmers to change their feedbase and has made it difficult for them to plan their forage mix over the following years.

In this environment of low and variable water availability, it is essential that dairy farmers have accurate estimates of the amount of irrigation water required to grow a range of forage types. The irrigation water requirements of a range of perennial forages, winter growing annual pastures and summer forage crops has been measured in experimental situations in northern Victoria. But plant irrigation water requirements can vary markedly from year to year.

Fortunately, there are models such as 'FAO-56'¹, which can use climatic data to predict the water requirements of irrigated forages. This model has been used in northern Victorian and there has been good agreement between the measured and modelled water use for most forages³^,⁴. So the FAO-56 model can be confidently used to predict the total and irrigation water requirements of forages typically grown in northern Victoria using historic weather data.

This information presents the probability distribution of annual irrigation water requirements, using historical weather data, for five irrigated forages, namely:

• perennial pasture
• lucerne
• short season, winter-growing annual pastures (irrigated from mid-March to mid-October)
• long season, winter-growing annual pastures, (irrigated from mid-February to mid-November)
• maize.

These probability distributions were determined for 3 locations in northern Victoria:

• Kerang
• Kyabram
• Yarrawonga.

The data on irrigation water use can be used by dairy farmers when planning their management responses to limited irrigation water.

This information also presents the probability distributions for irrigation water requirements for these forages at these locations for a range of climate change scenarios.

Methodology

An internationally recognised model¹ was used to estimate the irrigation water requirements of the five forage types at the three locations.

These estimates were then compared to measured irrigation water use from experiments based near Kyabram³^,⁵, with the model parameters then being adjusted if necessary to ensure consistency of modelled and measured water requirements. Historical climatic data for the three sites was sourced from SILO for the period 1935 to 2005. The three climate change scenarios used 70 years of data and assumed a steady-state climate in 2070.

Results

Historical irrigation water use

Analysis of the 70 years of historical climate data showed that rainfall was lowest at Kerang and highest at Yarrawonga (Table 1). In contrast, the 'reference crop evapotranspiration' (ETo) was highest at Kerang and lowest at Kyabram.

ETo is a description of the evaporative power of the atmosphere for a standard crop (typically a well water and well managed grass sward) — refer to Appendix 1 for how ETo was used to calculate irrigation water use. This resulted in the difference between average annual ETo and rainfall being higher at Kerang (920mm) than at either at Kyabram or Yarrawonga (740mm).

Table 1. Average climatic data (1935 to 2005) and modelled annual irrigation water use (ML/ha) for sites in northern Victoria.

These geographic differences in rainfall and ETo were also evident in modelled irrigation water use, with irrigation water use for all crop types being higher at Kerang than at either Kyabram or Yarrawonga (Table 1).

Modelled annual irrigation water use was highest when averaged for the perennial pasture (9.1ML per hectare) and Lucerne (8.7ML per hectare), intermediate for the maize (6.0ML per hectare) and long season annual (5.3ML per hectare) and lowest for the short season annual (3.2ML per hectare), when averaged over the three locations.

The large range in irrigation water use means that a simple average (and range) is a poor descriptor of the amount of irrigation water that is required for a given crop in either a wet, average or dry year. So the probability of exceedance values were calculated for each forage at each location using the historical climatic data.

A simple summary of these probability of exceedance values are presented in Table 2 while the complete data sets are presented in the probability of exceedance curves in Appendix 2.

Some interesting features of these probability of exceedance values in Table 2 are:

• For all forage types there was little difference in irrigation water use between Kyabram and Yarrawonga, as both rainfall and ETo are slightly higher at Yarrawonga than at Kyabram.
• The lucerne required a little less irrigation water than the perennial pasture due to the earlier finish to its irrigation season (mid-April rather than mid-May).
• The curves for lucerne and maize are very 'stepwise' compared to those for both the annual and perennial pastures — this is a result of them having a longer irrigation interval, and hence more water applied at each irrigation, compared to the pastures.
• The range in water use for maize at each location (difference between dry and wet years) is smaller than that of the other forage types.

Understanding probability of exceedance

The probability of exceedance curve for the irrigation water use of a perennial pasture at Kyabram is shown in Figure 1.

The graph showing irrigation water intake from 0 to 14 megalitres per hectare measured against probability of exceedance from 0 to 100 per cent.  The curve shows the proportion of years (vertical axis) for which the irrigation water use will exceed a certain level (horizontal axis). For example, annual irrigation water use will exceed approximately 5 megalitres per hectare for 99 per cent of years, 6.9 megalitres per hectare for 90 per cent of years, 8.6 megalitres per hectare for 50 per cent of years and 10.3 megalitres per hectare for 10 per cent of years.

The curve shows the proportion of years (vertical axis) for which the irrigation water use will exceed a certain level (horizontal axis). For example, annual irrigation water use will exceed approximately:

• 5ML per hectare for 99 per cent of years
• 6.9ML per hectare for 90 per cent of years
• 8.6ML per hectare for 50 per cent of years
• 10.3ML per hectare for 10 per cent of years.

So annual irrigation water use for perennial pastures at Kyabram is between 6.9 and 10.3ML per hectare for 80 per cent of years, with water use falling below this range for 10 per cent of years and above this range for 10 per cent of years.

Table 2. The probability of exceedance values for annual irrigation water requirements (ML/ha) for 5 forage types for Kerang, Kyabram, and Yarrawonga for wet, average and dry years as calculated using historical (1935-2005) climate data.

Seasonal conditions are defined as:

• Wet — irrigation water requirements were exceeded in 9 out of 10 years
• Average — irrigation water requirements were exceeded in 5 out of 10 years
• Dry — irrigation water requirements were exceeded in 1 out of 10 years

Climate change impacts on irrigation water use

Predicted changes to average climate conditions for the three climate change scenarios are shown in Table 3. These predicted changes involve increases in maximum and minimum temperatures and ETo and a decrease in annual rainfall. The changes were of a similar magnitude at all 3 locations (data not shown).

The predicted reduction in annual rainfall in the catchment areas is also likely to result in reduced runoff, inflows to storages and availability of irrigation water. These issues are not considered in this information.

With the high climate change scenario, rainfall declined by 40 to 50mm and ETo increased by 170 to 190mm for each of the three locations, compared to the historic data (Table 4). The resultant modelled increase in annual irrigation water use under the high change scenario was around:

• 2.2ML per hectare for the perennial forages
• 0.9ML per hectare for maize
• 1.3ML per hectare for the long season annual
• 0.8ML per hectare for the short season annual.

The increase in water use for all forages was higher at Yarrawonga than at either Kerang or Kyabram.

Probability of exceedance water curves for perennial pasture at Kyabram under historic and 3 climate change scenarios are shown in Figure 2. The curves show that the modelled annual water use at Kyabram is expected to increase by around 2 ML/ha for perennial pastures under the high climate change scenario.

The graph is showing data from 1935 to 2005 and predictions for climate change as of 2070 for low (B1), medium (A1) and high (A1F1) high climate change scenarios.

One noteable feature is that there is no effective change in the shape of the curves (they are reasonably parallel) meaning that the modelled increase in water use in the wettest years is likely to be similar to that in the driest years. This feature also occurred for all of the forage types at all three locations (data not shown).

Table 3. Predicted changes to annual average conditions using B1 (low), A1 (medium) and A1F1 (high) climate change modelling scenarios by 2070.

Table 4. Average climatic data and modelled annual irrigation water use (ML/ha) for three sites in northern Victoria using historic data (1935-2005) and predictions for climate change as of 2070 for the high (A1F1) climate change scenario.

Conclusions

Modelled annual irrigation water use using historical climate data was highest for the perennial pasture (9.1ML per hectare) and lucerne (8.7ML per hectare), intermediate for the maize (6.0ML per hectare) and long season annual (5.3ML per hectare) and lowest for the short season annual (3.2ML per hectare), when averaged over three locations in northern Victoria.

However, there was a large range in irrigation water use at each location, and large difference between locations. This means that probability of exceedance values will give a better description of how much irrigation water is likely to be required for a given forage at a given location, than a simple long term average.

Predicted increases in annual irrigation water use in 2070 under the high climate change scenario was around 2.2ML per hectare for the perennial forages, 0.9ML per hectare for maize 1.3ML per hectare for the long season annual and 0.8ML per hectare for the short season annual, when averaged over the 3 locations.

This data on irrigation water use and its predicted increases with climate change need to be used by dairy farmers and their advisors when planning their use of, and requirements for, irrigation water.

Appendix 1: Methodology for estimating irrigation water requirements

Simplification of validated, dual crop coefficient model

The FAO-56 single crop coefficient model¹ was used for this long-term modelling project as the actual dates of irrigation and grazing or harvesting are not known. The FAO-56 process uses climatic data (ETo) and crop characteristics (Kc) to calculate the water use of a crop. These terms are defined as:

• ETo is the 'reference crop evapotranspiration' and is a description of the evaporative power of the atmosphere for a standard crop (typically a well-watered and well-managed grass sward).
• Kc is the 'crop coefficient' and is a description of how much water a given crop will use compared to the standard crop.

The 'crop water use' (ETc) of a well water and well managed crop is given by:

ETc = ETo × Kc

Selection of climatic data

Historical climatic data for Kyabram, Kerang and Yarrawonga was sourced from SILO for the period 1935 to 2005. SILO data was used as it is readily accessible and is a relatively clean and comprehensive data set. The period 1935 to 2005 was selected as it is the same period of data used to generate climate change forecasts by the Intergovernmental Panel on Climate Change (IPCC).

The data for three climate change scenarios, based on the IPCC low, medium and high climatic change predictions⁶, were also used to model irrigation water use. The future climatic data used 70 years of data and assumed a steady-state climate in 2070.

Modelling rules

All forages were irrigated using border-check irrigation. It was assumed that all run-off from irrigation was captured and reused on the farm. However, no allowance was made for the capture of run-off from rainfall.

The irrigation interval for each forage was defined in terms of cumulative ETo-R since the last runoff event, where:

• ETo (as defined)
• R is effective rainfall (rainfall less any runoff resulting from that rainfall).

Forages modelled

The 5 forages modeled and their irrigation intervals (given in brackets) were:

• perennial pasture (ETo-R greater than 45mm)
• lucerne (ETo-R greater than 75mm)
• maize (pre-irrigated 15 November, sown 22 November, 110 day growing season, last irrigation 22 March), (ETo-R greater than 60mm)
• short season annual pasture (ETo-R greater than 45mm), irrigated from 15 March to before 20 October
• long season annual pasture (ETo-R greater than 45mm), irrigated from 15 February to before 20 November.

None of the forages were irrigated between 15 May and 15 August, with the last irrigation for the lucerne being prior to 15 April.

Deep drainage

An allowance was made for deep drainage during times when the model indicated surface ponding. The estimates from using this approach were consistent with the findings of Bethune², who estimated the discharge to the deep aquifer in the Goulburn Valley at 30 to 40mm per year.

Water intake at initial irrigation of winter annuals and maize

The FAO-56 methodology cannot predict water intake at the initial irrigation of annual forages. So the method derived by Lawson, based upon a relationship between cumulative E-R since 1 December and intake at the initial irrigation, was used to predict intake at the first irrigation for the long and short season annuals. This method sets an upper limit of 1.5ML per hectare of water at the initial intake, which may underestimate intake for some grey cracking soils that are often found around Kerang.

For the initial water intake of maize it was assumed that the site was irrigated in spring prior to the establishment of the maize — so water intake at the initial irrigation was set at 80mm.

Validation of modelling outputs

Two water use experiments conducted near Kyabram measured the water use of three maize crops and seven forage systems over a three year period. The modelled water use values were compared to the measured water use values to ensure consistency. Where necessary, the model was adjusted to ensure that the modelled water use agreed with the measured water use.

Appendix 2: Probability of exceedance curves using historical (1935-2005) climatic data

Modelled annual irrigation water requirements for perennial pasture (PP), lucerne, maize, long season annual pasture (LSAP) and short season annual pasture (SSAP) for Kerang, Kyabram, and Yarrawonga. See Figure 1 and the accompanying text for an explanation of the curves.

References

¹Allen RG, Pereira LS, Raes D, Smith M (1998) 'Crop Evapotranspiration: Guidelines for Computing Crop Water Requirements.' FAO Irrigation and Drainage Paper No. 56 (FAO: Rome)

²Bethune M (2004) Towards effective control of deep drainage under border-check irrigated pasture in the Murray-Darling Basin: a review. Australian Journal of Agricultural Research 55, 485-494.

³Greenwood KL, Mundy GN, Kelly KB (2008) On-farm measurement of the water use and productivity of maize. Australian Journal of Experimental Agriculture 48, 274-284.

⁴Greenwood KL, Lawson AR, Kelly KB (2009) The water balance of irrigated forages in northern Victoria, Australia. Agricultural Water Management 96, 847-858.

⁵Lawson AR, Greenwood KL, Kelly KB (2009) Water productivity of winter-growing annuals is higher than perennial forages in northern Victoria. Crop and Pasture Science 60, 407-419.

⁶Weeks A, Christy B, O'Leary G (2010) Generating daily future climate scenarios for crop simulation. In 'Food Security from Sustainable Agriculture' edited by H Dove and RA Culvenor. Proceedings of 15th Agronomy Conference 2010, 15-18 November 2010, Lincoln, New Zealand.