- Email: [email protected]
Selection of flow rate and irrigation duration for high performance bay irrigation
Agricultural Water Management 228 (2020) 105850
Contents lists available at ScienceDirect
Agricultural Water Management journal homepage: www.elsevier.com/locate/agwat
Selection of flow rate and irrigation duration for high performance bay irrigation
T
RJ Smitha, MJ Uddinb,* a b
Centre for Agricultural Engineering, University of Southern Queensland, Toowoomba, Qld, Australia Trangie Agricultural Research Centre, NSW Department of Primary Industries, Trangie, NSW, Australia
A R T I C LE I N FO
A B S T R A C T
Keywords: Surface irrigation Border-check irrigation Infiltration Simulation Application efficiency Optimization Time to cut-off
The maximum efficiency attainable by surface irrigation in any particular situation is determined largely by the soil infiltration characteristic and the flow rate onto the field. Performance evaluations have suggested that higher flow rates than those traditionally recommended can lead to increases of about 20% in the application efficiency of bay irrigation across the dairy regions of southern Australia. However, substantially reduced irrigation durations are required to realise these efficiency gains and greater precision is required in the selection and management of these shorter durations. In this paper, infiltration characteristic curves representative of the predominant (cracking and non-cracking) soils of the region are used in the surface irrigation simulation model SISCO, for a range of bay lengths and soil moisture deficits, to determine: (i) the flow rates required to achieve the maximum efficiency on the bay irrigated soils of southern Australia; and (ii) the means for real-time estimation of optimum irrigation durations. The aim is provision of guidance to irrigators seeking higher efficiency through the use of higher flow rates. Real-time estimation of optimum time to cut-off must recognise the substantial variations in infiltration within a soil type and with time due to changes in antecedent moisture content, and hence must be adaptive to the conditions prevailing at the time of any irrigation. A method for estimating time to cut-off (previously developed for furrow irrigation), is extended to the hydraulically more complex case of bay irrigation. The method is based on simple linear relationships between the advance time to a nominated point part way down the field and the time to cut-off. Its application to the management of irrigations is demonstrated using data from multiple irrigations of a single bay under varying soil moisture conditions.
1. Introduction Despite the evidence that well designed and managed surface irrigation systems have the potential to deliver high application efficiencies, most commercial systems still appear to be operating with significantly lower and highly variable efficiencies. The reason for this can be traced to the use of lower than optimal flow rates and more particularly the use of inappropriate times to cut-off. For example, in the Australian cotton industry, Smith et al. (2005) have shown that increasing the furrow inflow rates to 6 L/s and reducing the time to cut-off (Tco) would potentially raise the average application efficiency from the previous low of 50% to an improved 75%. More recent evaluations of furrow irrigation performance in that industry (Montgomery and Wigginton, 2008) have confirmed that adoption of higher flow rates has led to average application efficiencies in excess of 70%. As a result, the cotton industry is estimated to have
⁎
saved 28.5 G L of water per annum and achieved an industry improvement in WUE of 10% (BDA Group, 2007). Selection by farmers of appropriate times to cut-off remains problematic and the simulation work by Smith et al. (2005) suggested that raising efficiency further could only come about by selecting an individual Tco for each irrigation that gives optimum performance for the prevailing conditions. Substantial work toward real time selection of Tco has been undertaken with some success by Khatri and Smith (2006); Koech et al. (2014) and Smith et al. (2018). A similar situation prevails in the bay (border check) irrigated dairy and fodder cropping regions in southern Australia. A study by the Cooperative Research Centre for Irrigation Futures (Smith et al., 2009 and Gillies et al., 2010) demonstrated that significant gains (at least 20%) in application efficiency were possible in the bay irrigation of pasture and fodder crops, again by the simple expedient of doubling flow rates (from 0.07 to 0.22 L/s/m width to 0.24 to 0.44 L/s/m) and
Corresponding author. E-mail address: [email protected] (M. Uddin).
https://doi.org/10.1016/j.agwat.2019.105850 Received 22 February 2019; Received in revised form 4 October 2019; Accepted 6 October 2019 0378-3774/ © 2019 Elsevier B.V. All rights reserved.
Agricultural Water Management 228 (2020) 105850
R. Smith and M. Uddin
simulation model used in this study the infiltration data were converted to the form of the three parameter modified Kostiakov equation:
reducing irrigation durations. For growers seeking Federal Government funding for on-farm efficiency improvements, a minimum flow rate of 0.2 ML/d/m width is now required (GBCMA et al., 2012), based on the above work of Smith et al. (2009) and Gillies et al. (2010). No guidance was provided on selection of appropriate Tco. At these higher flow rates the very much reduced times to cut-off required mean that the ability to achieve ‘optimal’ performance is particularly sensitive to the cut-off time. For farmers to manage irrigations with these higher flow rates and shorter run times they will require, as a minimum, very clear and precise guidelines and it typically will necessitate real-time estimation of Tco. More recently, Smith et al. (2016) evaluated the irrigation performance achieved on nine farms employing automation of their irrigated bays, all of which used flow rates in excess of the recommended 0.2 ML/d/m. This study showed that with careful selection of cut-off times, application efficiencies in excess of 90% could be obtained routinely. Unfortunately some of the farmers were less diligent in their management of cut-off times and did not receive the full efficiency gains that are possible through automation. In other parts of the world researchers and farmers are wrestling with the same issues relating to management of bay irrigations. Efforts to reduce water use in the bay irrigation of lucerne (alfalfa) in California (for example, Saha et al., 2011 and Arnold et al., 2014) have tended to ignore flow rate (presumably because of infrastructure limitations) and have focussed on reducing tail-water by attempting to improve selection of time to cut-off. The purpose of the present study is to use simulation along with published infiltration and irrigation evaluation data to: (i) determine the flow rates that will provide best performance over the range of soil types across the dairy regions of northern Victoria and southern New South Wales, and (ii) assess methodologies that will allow adaptive real-time estimation of optimal times to cut-off.
I = kt a + fo τ
(1)
where I is the cumulative depth of infiltration (mm), a and k are fitted parameters, fo is the final or steady infiltration rate (mm/min) and τ is the infiltration time (min). The feature most evident from the infiltration data is the very substantial variation in infiltration within a field, within a soil type and between soil types. This is supported by the results of Selle et al. (2011) who showed that the within field and within soil type variation in final infiltration rate were not dissimilar in magnitude. Averaging infiltration over an irrigated bay (as is done by any measurement that uses the bay as the infiltrometer) masks some of the within field variation. However, the within soil type variation means that infiltration cannot be estimated from the soil descriptions provided by the available soil surveys. This poses a real challenge for the development of irrigation management guidelines which, as a consequence, need to be based on objective hydraulic criteria rather than on the basis of soil type. The data also suggest that the soils in the region are substantially more permeable than indicated by reports such as those by Mehta and Wang (2005) and Selle et al. (2011), based on point measurements of the soil hydraulic properties. This has significant implications for irrigation management, the potential magnitude of deep percolation losses and accessions to groundwater resulting from over irrigation. The other feature evident from the data is that the soils fall clearly into two key groups, cracking and non-cracking although it should be noted that different occurrences of a particular soil might fall into either group. However, soil cracking is a readily identifiable characteristic and farmers should have no difficulty in identifying into which group their soils fall. In the cracking group of soils the parameter a in the modified Kostiakov equation is zero or very small and the k parameter represents the instantaneous initial infiltration or crack fill. This in effect reduces the equation to the two parameter linear equation as used in the AIM model of Austin and Prendergast (1997):
2. Study Area This study pertains to the southern Murray Darling Basin region of northern Victoria and southern New South Wales (Fig. 1). The landform is an alluvial plain. Soils have been formed by floodplain deposition from the Murray River and tributary streams that have meandered over the floodplain for millennia. Soils range from coarse sands deposited adjacent to streams, to fine textured, cracking clays resulting from deposition of fine material remote from stream channels (Butler, 1950). Soil types across the region are as described in Skene (1963); Skene and Freedman (1944); Skene and Harford (1964), and Skene and Poutsma (1962) Border irrigation of perennial and annual pastures and fodder crops uses approximately 1600 G L of irrigation water annually on over 400,000 ha of irrigated land (ABS, 2008).
I = ZCR + fo τ
(2)
where ZCR is an instantaneous crack fill (mm). The crack fill is a function of the soil moisture deficit (Zreq) and for Lemnos loam is typically taken to be (Robertson et al., 2004):
ZCR = 0.75Zreq
(3)
The steady infiltration rate for these soils ranges from 0 to about 5 mm/h but is thought to be strongly skewed toward the lower values (Mehta and Wang, 2005). It is often simply taken to be 1 mm/h (e.g. Robertson et al., 2004). The published infiltration data compiled in this study are biased toward the cracking group of soils, partly because it contains some of the more prevalent soils of the region such as Lemnos loam and Kerang clay. Four infiltration curves (Fig. 2) were selected to represent these cracking soils: C1 – final infiltration rate 1 mm/h, deficit 40 mm (a 0.08, k 0.020, fo 0.0000167 m/min) C2 – final infiltration rate 1 mm/h, deficit 40 mm (a 0.08, k 0.020, fo 0.00005 m/min) C3 – final infiltration rate 3 mm/h, deficit 60 mm (a 0.08, k 0.032, fo 0.0000167 m/min) C4 – final infiltration rate 3 mm/h, deficit 60 mm (a 0.08, k 0.032, fo 0.00005 m/min) In the case of the non-cracking soils there is insufficient data to give any understanding of how the infiltration characteristic for a particular soil might vary (if at all) with deficit. Hence the strategy used here was to select three infiltration characteristics (Fig. 3) representative of the range of characteristics encountered, each of which is based on the published data for a named soil. The soils are:
3. Infiltration Data A detailed search of the published literature drew 51 infiltration characteristics covering 15 named soil types plus a further 8 characteristics where the soil type was not specified other than in general terms of soil texture. These data were originally gathered for a variety of purposes using of various measurement methods ranging from point measurements using ring infiltrometers (Tisdall, 1950; Robertson et al., 2004) and recirculating infiltrometers (Turral and Malano, 1996) to field averages where the irrigation bay was the infiltrometer. In this latter group, methods included volume balance approaches such as the two-point method (Maheshwari et al., 1988; Hume, 1993) to least squares fitting using hydrodynamic models (Turral, 1996; Austin and Prendergast, 1997; Smith et al., 2009; Gillies et al., 2010). Only in a very few instances was the soil moisture deficit at the time of the measurement specified in the source publication. The source data were reported in a variety of formats. To suit the 2
Agricultural Water Management 228 (2020) 105850
R. Smith and M. Uddin
Fig. 1. Location of study area (source Department of Primary Industries, Victoria).
SISCO hydraulic simulation model (Gillies and Smith, 2015) for a range of flow rates and for three bay lengths. For each combination of length and flow rate, the irrigation was optimised to determine the duration of inflow or time to cut-off (Tco) that would give maximum application efficiency (Ea). Run-off from the end of the bay was set to 5% to ensure that the irrigation always reached the downstream end of the bay. This is a small margin and may at times be insufficient compensate for errors in the estimation of parameters such as the Manning n. The implications of this and the control requirements have been discussed in later sections of the paper. For many (mostly cracking) soil types in the region the deficit is replenished rapidly (via the process of crack fill) hence the 5% runoff condition also typically ensures that the irrigation is
N1– low permeability - based on Wanalta loam or similar (a 0.4, k 0.0045, fo 0.0, approximately equivalent to a USDA 0.15 Light Clay). N2 – moderate permeability - Cobram loam (a 0.4, k 0.0055, fo 0.0, USDA 0.2 Clay Loam). N3 – higher permeability - a more permeable occurrence of Cobram loam (a 0.4, k 0.0066, fo 0.0, USDA 0.25 Clay Loam).
4. Simulations 4.1. Simulation methodology and optimisation objective For each of the case study soils, irrigations were simulated using the
Fig. 2. Representative infiltration characteristics for the cracking soils. 3
Agricultural Water Management 228 (2020) 105850
R. Smith and M. Uddin
Fig. 3. Representative infiltration characteristics for the non-cracking soils.
than those recommended in the earlier extension guidelines of Lavis et al. (2006). The exceptions are for longer bays on the more permeable soils with small moisture deficits where maximum application efficiencies reduce to below 80%. It must be stressed that these are maximum possible application efficiencies and assume perfect selection and control of irrigation duration (Tco). Few farmers would be achieving maximum efficiencies at their current flow rates and without assistance with selection of Tco would certainly not achieve them if they adopted higher flow rates. The simulations suggest preferred flow rates for cracking soils of 6 ML/d/ha for irrigation of pasture with soil moisture deficits up to about 50 mm. This should increase to 8 ML/d/ha for irrigation of deep rooted fodder crops with deficits in the order 80–100 mm. For pasture on non-cracking soils the recommended flow rates depend on the soil permeability with 6, 10 and 14 ML/d/ha recommended for lowly, moderately and more highly permeable soils, respectively. For a 400 m long 50 m wide bay, the above flow rates correspond to unit rates of 0.24, 0.4, and 0.56 ML/d/m width or 2.8, 4.6 and 6.7 L/s/m width, respectively. Again these rates should be increased by 2 ML/d/ha for higher deficits. In some instances the recommended flow rates might exceed the supply rate available, in which case growers will have to balance the potential small loss of application efficiency with the cost of reducing bay widths to increase the flow rate per unit width. At these higher flow rates the optimised times to cut-off are much shorter than those required for the lower flow rates traditionally used. It also means that cut-off will occur well before the advance has reach the halfway point down the field posing a significant challenge to any procedure employed to estimate the time to cut-off. One factor, not accounted for explicitly in the above simulations, that might limit application efficiencies on some sites is residual water stored in the micro-topography of the bay after the irrigation is finished. This factor is most significant in bays which are on heavier soils, on low slopes, imprecisely graded, with dense pasture and where traffic by stock has generated substantial micro-relief (Campbell, 1988). An early attempt to quantify this residual water is contained in the report by Campbell (1988) who measured residual depths of from 1.1 to 12.5 mm on two bays with slopes of 1:750 and 1:1500 respectively, with the larger values occurring on the flatter bay. In a detailed water balance of a particularly long duration and inefficient irrigation on a bay with heavy clay soil, Gilfedder et al. (2000) showed that on the particular site, 20 mm of water remained on the surface of the bay 20 h after the irrigation. For modern steeper and well graded bays and where high flow rates and much shorter irrigation durations are employed this term should be small. In cases where the infiltration characteristic has been determined from irrigation advance data this depth of residual
adequate in meeting the requirements of the crop. The other usual performance parameters of requirement efficiency (Er) and uniformity were also calculated for each irrigation, along with an estimate of the deep percolation loss. The optimisation objective used in studies such as this is always subjective. Individual researchers and growers place different values on the key performance parameters depending on their particular circumstances. For example, it is not uncommon for the requirement efficiency to be set at 100%, meaning complete replenishment of the soil moisture deficit (e.g. Feyen and Zerihun, 1999). Such a strategy inevitably results in substantial runoff and deep drainage volumes that reduce application efficiency significantly, and arguably is not a valid strategy on the more permeable soils or when water is in short supply. As well as reduced application efficiency it also leads to reduced water use efficiency (WUE). This was well illustrated by Uniwater (2008) who showed substantial increases in WUE on irrigated lucerne by reducing cut-off times below those required for 100% Er. An alternative strategy sometimes used is to maximise uniformity (e.g. Finger and Morris, 2012). This is only valid if Er is substantially less than 100%. If Er is 100% then the uniformity of moisture stored in the root zone is by definition 100% and this maximisation objective is non-sensical.
4.2. Simulation results The full set of results covers a very large number of combinations of soil type, moisture deficit, field length, inflow rate, slope and crop type (Manning n). Typical results for the cracking and non-cracking soil groups are provided in Figs. 4 and 5 respectively, as plots of the key performance measures Ea and Er versus flow rate. The optimum time to cut-off (Tco) and the distance reached by the irrigation advance (Xco) at the point of cut-off are also shown on these plots. In these plots, as in most cases simulated, the application efficiencies initially show a rising trend with increasing flow rate until reaching a plateau at some particular flow rate. The magnitude of this flow rate is a function of the soil type, crop density, bay length and soil moisture deficit, being higher for the more permeable soils, denser crops, longer bays and higher deficits. At or just before the point of maximum Ea, the requirement efficiency Er was generally adequate (> 95%) thus defining the preferred flow rate. At this flow rate the irrigation durations (Tco) were typically about 2–3 hours except for the very long bays and high deficits. The exceptions to the above were the cracking soil types with a short bay and low deficit, where the application efficiency was essentially constant at all flow rates. These data show that application efficiencies in excess of 90% are possible on many soils in the region through the use of flow rates higher 4
Agricultural Water Management 228 (2020) 105850
R. Smith and M. Uddin
Fig. 4. Example results for summer pasture (n = 0.26) in a 400 m long bay slope 1:750 on cracking soil (C1) and 40 mm moisture deficit.
pasture or fodder crop can be anywhere in the range 0.1 to 0.36 whereas n for a bare furrow is usually taken to be 0.04. This means that in a typical bay at times at or near cut-off there is a far greater volume of water contained in temporary storage in the advancing irrigation flow. This is illustrated in Fig. 6 for an efficient irrigation on a hypothetical irrigation bay employing a high flow rate and short time to cut-off. At the point of cut-off (2 h) between 50 and 60% of the water applied to the field is still on the field. The problem is compounded by the substantial difficulty in accurately estimating the volume of this surface ponding. Finally, because of the much shorter times to cut-off, any sensing and calculations required to estimate Tco and the subsequent decision to turn the flow off have to be made much earlier in the irrigation (typically before the advance has reached 40% of the length) and more quickly than in furrow irrigation. To be effective, methods for selection of Tco need to be simple to apply, involve the minimum of measurement and calculation on the part of farmers, but at the same time be adaptive to changes in the key variables of flow rate, soil moisture deficit, and crop density. The rate of advance of the wetting front down the bay integrates the effects of most of the relevant variables, including those above. It is also something that irrigators can measure relatively simply. It is also assumed that irrigators will at least also know the inflow rate into their bays. Some simplification of the problem can result if farmers irrigate each time at
water is most likely already embedded in the estimated infiltration characteristic, most probably in the initial infiltration term k, and has thus been accounted for in the efficiency calculations. There is no evidence in the literature of significant residual water occurring on the non-cracking soil types.
5. Toward a method for REAL-TIME selection of time to CUT-OFF 5.1. The challenge Recent work on furrow irrigation in the Australian cotton industry by Khatri and Smith (2006); Koech et al. (2014) and Smith et al. (2018) has led to two equally effective methods for estimation Tco in real time. They differ in complexity but both only require one measurement of the irrigation advance taken relatively late in the irrigation. However it is not known how these methods might translate to bay irrigation. In Australia bay irrigation differs from furrow in a number of significant ways. Firstly, with lengths ranging from about 200 to 600 m bays are much shorter than the 600 to 1500 m furrows used in the cotton industry. Secondly, the hydraulic resistance to the irrigation flow provided by the crop or soil surface (as represented by the parameter n in the Manning equation) is typically much greater and more variable than in a clean furrow. For example the Manning n for a summer
Fig. 5. Example results for summer pasture (n = 0.26) on a permeable non-cracking soil N3, length 400 m, slope 1:750 and 60 mm deficit. 5
Agricultural Water Management 228 (2020) 105850
R. Smith and M. Uddin
Fig. 6. Schematic of the water balance of an efficient bay irrigation event employing a high flow rate and short time to cut-off.
Saha et al. (2011) and Arnold et al. (2014) proposed the use of a simple volume balance approach and applied it to bay irrigated Lucerne in southern California with mixed results. The method utilises measurement of advance at two points as well as an estimate of the average depth of surface ponding. Two issues that limit the application of this method are immediately evident. First is the use of a very much simplified infiltration equation that consisted only of the initial rapid infiltration term k, and which neglected any continuing infiltration described by fo. For many of the soils of the region which display a nonnegligible continuing infiltration the method will underestimate the depth or volume of infiltration and consequently will underestimate the Tco thus introducing the real possibility that the advance will not reach the end of the field. There is no obvious way to introduce the continuing infiltration into the volume balance analysis. The second issue is the fact that the calculation of the infiltration term and hence Tco is very sensitive to the (estimated) value of the average depth of water on the field. Estimating this term from a limited number (only two) of depth measurements is highly problematic. For furrow irrigation on the heavy clay soils in the Australian cotton growing regions, Smith et al. (2018) showed that for a particular field there exists a simple linear relationship between Tco and the time of the irrigation advance to a nominated point at least 50% of the distance down the furrow. This relationship was shown to be independent of the flow rate and the soil moisture deficit, knowledge of which was not required. The method was shown to provide robust estimates of Tco that resulted in high application efficiencies. The remainder of the present paper will focus on the possible application of this approach to bay irrigated pasture and fodder crops.
the same soil moisture deficit thus elimination problems caused by variation of the infiltration characteristic of the bay, and if they use the same and constant flow rate each time they irrigate. This is now possible in the region under consideration following full automation of the supply system. 5.2. The options From an examination of the literature, three methods appeared worthy of consideration, viz: 1 The use of a hydrodynamic model in a manner similar to the Autofurrow model of Khatri and Smith (2006) and Koech et al. (2014), 2 A volume balance approach following on from the work of Saha et al. (2011) and Arnold et al. (2014), and 3 Simple unique empirical relationships for each bay linking Tco with the rate of advance to some nominated point down the bay, similar to that used in furrow irrigation by Smith et al. (2018). The Autofurrow method for the real-time selection of Tco in furrows (Khatri and Smith, 2006; Koech et al., 2014) requires measurements only of the inflow rate and advance to a single point. The main features of this optimization process are: the use of a model infiltration curve and a scaling process to describe the current soil infiltration characteristic; measurement of the inflow rate to the furrows; measurement of the water advance at a point at least midway down the furrow; an assumed constant Manning n; and a hydraulic simulation program based on the full hydrodynamic model to predict the optimum time to cut-off. To apply this method to bay irrigation requires additional measurements (for example, additional advance points and/or flow depths) to enable estimation of the Manning n parameter for the crop or pasture growing in the bay. Even so, a major difficulty arises because n and the infiltration parameter k both influence the rate of advance in a similar manner, that is, an increase in either parameter causes the advance to slow. However they have opposing effects on the volume of surface ponding. Hence multiple measurements of flow depth are required to isolate the values of these two parameters. It is easily shown through simulation that the number and location (at or near the downstream end of the field) of sensors and measurements required to identify the correct values of n and k make this approach prohibitive as a practical approach for farmers to employ. This issue with the sensor locations was first intimated by Walker (2005) and further demonstrated by Gillies and Smith (2015) during the initial testing of the SISCO model.
5.3. Relationship between Tco and advance time for irrigation bays The simulations used in the previous section to establish preferred flow rates involved optimisation to give maximum Ea and 5% runoff for a wide range of the key variables. Here these same data are used to explore the relationship between advance time and optimum Tco. For this purpose the advance down the bay was represented by the time taken to reach a point 3/8th of the distance down the bay (Tadv). This distance used for the advance time was selected arbitrarily but in all cases the advance time was less than the Tco hence would give the farmer time to make the control decision. An example of the results, in this case for each of the cracking and non-cracking soil types (C1, C2, C3, C4, N1, N2 & N3) for summer pasture (n = 0.26) on a hypothetical bay 400 m long, and with an inflow rate of 2.31 L/s/m width are presented in Fig. 7. The two lines representing the cracking soils (C1 & C3, C2 & C4) were each calculated for various values of the infiltration 6
Agricultural Water Management 228 (2020) 105850
R. Smith and M. Uddin
Fig. 7. Optimum Tco vs advance time to 150 m (3/8th distance) for summer pasture in a 400 m long bay for various soil types and moisture deficits and an inflow rate of 2.31 L/s/m width of bay.
the results for a range of soil moisture deficits ranging from 30 to 60 mm. The cause of the difference between the results for bay and furrow irrigation appears to lie in the higher value of the Manning n for pasture and the very short advance distances necessarily used in this analysis. The matter is further complicated by the fact that a different set of curves also results for different crop densities, that is for different values of the Manning n (Fig. 9). In this figure curves are presented for a single flow rate and for bare soil (n = 0.05), sparse annual pasture or fodder crop (n = 0.1), summer pasture (n = 0.26), and a very thick summer pasture or crop (n = 0.3). The difference caused by the Manning n can be explained by recourse to a simple volume balance. For a given deficit and the same flow rate, the same Tco will be required in each case to apply the correct volume of water to satisfy the deficit. The higher the Manning n means greater flow depths, greater temporary storage of water on the surface of the bay and hence slower advance. A similar situation applies to the non-cracking soils. While not presented here, the relationship between Tadv and Tco is also dependent on the length of the bay.
parameter k. For the non-cracking soils (N1, N2 and N3) a single value of the soil moisture deficit was used. For each case the results show linear relationships between Tco and Tadv. It should be noted here that changing the values of the key parameters will result in different relationships. Changing the point used for the advance measurement simply results in another family of straight lines, as does altering the inflow rate or Manning n. Similarly changing the target runoff from 5% to reflect a different risk or uncertainty also results in another family of straight lines. For the cracking soil, the soil infiltration characteristic, and in particular the rapid initial infiltration, are known to be a function of the soil moisture deficit (Eq. (3) and associated text). The majority of the infiltration occurs in the first few minutes that water is present on the surface with only relatively small depths infiltrating after that time. This tends to result in a relatively uniform rate of advance. Because the two lines for the cracking soils are each for various values of the infiltration parameter k and they therefore represent varying deficits, suggesting that the relationships for these soils are independent of the soil moisture deficit. This is consistent with the similar result for furrow irrigation on cracking soils (Smith et al., 2018). The difference between them is the effect of the different rates of continuing infiltration represented by different values of the parameter fo (1 mm/h for C1 & C3 and 3 mm/h for C2 & C4). The non-cracking soils vary from the cracking soils in that the relationship between the infiltration characteristic and the soil moisture deficit is not well known. The three non-cracking soils (N1, N2 & N3) are in order of increasing infiltration (increasing k) the results for which fall on a single straight line (Fig. 7) for the given flow rate, deficit and crop condition. This difference in infiltration between the three soils was primarily intended to show how the advance vs Tco relationship might change with increasing soil permeability. However it also might show the effect of increasing soil moisture deficit. Assuming the infiltration curve varies with deficit by varying the parameter k (as with the cracking soils) then the relationship for these soils might also be largely independent of the soil moisture deficit. (Note that while the selection of the optimum Tco might be independent of the deficit the resulting application efficiency is very much dependent on it.) The values of Tco for the non-cracking soils are typically higher than for the cracking soils at the same flow rate. This reflects the very different shape of the infiltration curves (and the greater ongoing infiltration that occurs at later times). The greater permeability also results in lower maximum application efficiencies. The previous furrow irrigation work by Smith et al. (2018) showed that for furrows the relationship between advance and Tco was also independent of inflow rate. This not the case for bay irrigation, as illustrated in Fig. 8 which shows the relationships for a cracking (C1) soil, each line corresponding to a particular flow rate, and represents
5.4. Sensitivity analysis The relationship for the cracking soil (C1) presented in Fig. 7 is: Tco = 3.19 Tadv −235
(4)
Using the same data as used to develop this relationship the sensitivity of the irrigation performance to changes or errors in the soil moisture deficit, flow rate and Manning n were assessed by varying each of those parameters by ± 10% (Table 1). As expected a change in the deficit of 10% causes only a minor change in performance with the application efficiency changing by 2%. However performance is a little more sensitive to changes in the inflow rate and Manning n. A 10% change in these parameters causing about a 5% change in performance and possible instances of failure to reach the end of the field. For irrigators in the Goulburn Murray Irrigation District of northern Victoria, full automation of the supply system has provided them with constant and precisely known flow rates, removing the likelihood of errors or uncertainties in flow rates. The uncertainty involved in estimating the Manning n is more problematic. However this can be countered effectively by either erring toward higher values of n or by aiming for a slightly higher runoff volume when establishing the Tco/Tadv relationship. Either approach provides an additional margin of safety. Irrigating with a consistent flow rate at a consistent soil moisture deficit will help remove some uncertainty.
7
Agricultural Water Management 228 (2020) 105850
R. Smith and M. Uddin
Fig. 8. Optimum Tco vs advance time to 150 m for summer pasture in a 400 m long bay on a cracking soil at various inflow rates.
5.5. Management implications
Table 1 Results of sensitivity analysis of irrigation performance with the variation of soil moisture deficit, flow rate and Manning n/.
To illustrate application of the method to real on-farm management of irrigations, data from five sequential irrigations in a single bay planted to maize being grown for fodder were used. These data were obtained in an evaluation of automated bay irrigation undertaken in 2014 by Smith et al. (2018) and reported in that paper as farm D. The bay is 400 m long and 50 m wide with an average slope of 1:916. Soil type is a heavy clay loam. Each of the five irrigations were monitored and evaluated, with continuous measurement of the inflow hydrograph and the flow depths at three points down the field. Infiltration parameters and the Manning n were determined by a calibration process using the simulation model SISCO (Gillies and Smith, 2015) and the performance measures (Ea, Er, runoff and deep drainage) estimated again using the SISCO model. Inflow rate was the same for each irrigation at 14 ML/d but the soil moisture deficit prior to each irrigation varied wildly (Table 2) resulting in infiltration characteristics of very different magnitude (Fig. 10). The Manning n increased from 0.06 to 0.09 throughout the sequence (Table 2). In addition to the actual performance achieved by the farmer, each irrigation was optimised using SISCO with the results
Control Deficit +10% Deficit -10% Inflow +10% Inflow-10% n +10% n -10% a
Tadv
Tco
Ea
Er
112.0 118.0 105.5 104.5 119.0 116.0 107.0
122.3 141.4 101.5 98.4 144.6 135.0 106.3
92.8 89.5 94.8 99.1 87.5 84.5 99.6
98.2 99.7 92.6 92.9a 98.7 98.8 91.7a
Advance stopped 5% short of the end of field.
also given in Table 2. To develop the relationship between Tco and the time of advance to the 3/8th distance for a particular bay at least one irrigation at the site must be evaluated as described above. For farm D the relationship was developed from the data from the first irrigation (29 Jan 14) by varying the deficit and hence the infiltration parameter k and optimising as above with the result given by the trend line in Fig. 11. The relationship
Fig. 9. Optimum Tco vs advance time to 150 m for various crop conditions (different Manning n values) in a 400 m long bay on a cracking soil and inflow rate of 3.09 L/s/m width. 8
Agricultural Water Management 228 (2020) 105850
R. Smith and M. Uddin
Table 2 Comparison of farmer, optimum and real-time managed irrigation performance – farm D. Farmer
Optimum
Managed (equation 5)
Date
Deficit (mm)
n
Tco (min)
Ea %
Tco (min)
Ea %
Tco (min)
Ea %
Runoff %
29 Jan 14 4 Feb 14 11 Feb 14 23 Feb 14 8 Mar 14
29 45 57 31 43
0.060 0.060 0.070 0.083 0.090
110 110 110 115 110
55 70 77 48 67
43.0 67.6 82.1 33.5 64.0
95 95 95 95 95
43.0 67.4 92.0 48.7 73.2
95.0 95.0 87.4 76.4 87.7
5.0 5.0 12.6 23.6 12.3
bay the efficiencies obtained would have been very much improved from those obtained by the farmer (Table 2). The equation accommodates the varying soil moisture deficits and varying infiltration but does not account for the increases in the Manning n with time as the crop develops. Hence the efficiencies that would have been obtained in the latter part of the season are slightly sub-optimal. While one data set is not sufficient to allow firm conclusions to be drawn it does illustrate the potential of the simple approach for estimating optimal Tco based on a measured time of advance to a nominated point. It has the advantage of requiring only a single sensor in each bay but does require full monitoring and evaluation of one irrigation to enable the relationship between advance time and optimal Tco to be developed. High flow rate bay irrigation with its attendant early cut-off is a radical departure from the traditional practice in northern Victoria. It does carry some risk but offers high rewards in terms of high application efficiencies, minimal deep percolation losses, reduced waterlogging of surface soils and reduced accessions to the water-table. Those irrigators who have adopted the practice have not all achieved these rewards (Smith et al., 2018) and those who have did so by trial and error. The method described in this paper offers improved performance to all adopters.
is: Tco = 2.89 Tadv – 65.5
(5)
Tco for all five optimised irrigations are also plotted against the corresponding advance time to the 3/8th distance in Fig. 11. Finally, the irrigation performance that would have been achieved had the remaining irrigations been controlled by equation 5 are presented in the final columns of Table 2. The significant issues with the management of irrigations on Farm D are:
• a lowly permeable soil with an unusual infiltration characteristic, • the uncertain and varying deficit, and • the variation in Manning n. In his attempt to satisfy the soil moisture deficit, the farmer used excessively long irrigation durations (Tco) the consequences of which were low application efficiencies and large volumes of runoff (Table 2). Runoff in fact was the sole loss at each irrigation, there were no deep percolation losses. The optimisations of the individual irrigations, undertaken with the benefit of hindsight, show that very much higher application efficiencies could have been obtained through substantial reductions in irrigation duration (which would vary in accordance with the deficit). However in all cases the optimised irrigations failed to fully satisfy the deficit. Because the farm is fully automated this under-irrigation is not a major issue, it simply means that the next irrigation will be required a little sooner. It also might lead to some yield improvement through a reduction in waterlogging of these heavy soils. It is worth noting at this point that the deficits were measured using an uncalibrated capacitance probe and that while the relativity between irrigations would likely be correct the magnitudes of the deficits may not be. Had the irrigations been controlled by equation 5 in conjunction with a single advance measurement located 150 m from the head of the
6. Conclusions Published data on soil infiltration properties across the Murray Dairy region of southern Australia were used in a simulation study of bay irrigation performance. It has confirmed that flow rates higher than those recommended by previous guidelines have the potential to deliver higher application efficiencies for bay irrigated pasture and fodder crops across this region. Preferred flow rates for the major soil groups are recommended. These higher flow rates result in much shorter irrigation durations (or times to cut-off) and consequently greater accuracy and precision in the selection and management of time to cut-off is
Fig. 10. Infiltration characteristics for farm D (from Smith et al., 2017). 9
Agricultural Water Management 228 (2020) 105850
R. Smith and M. Uddin
Fig. 11. Optimum Tco vs advance time for farm D.
required. The same simulation data were used to illustrate the relationships that exist between the rate of advance of the wetting front down an irrigated bay and the optimum time to cut-off for that bay. It is shown that if a bay is managed consistently, that is, irrigated at a consistent soil moisture deficit and consistent flow rate, a simple linear function can be used to estimate the Tco from a single advance measurement taken relatively early in the irrigation. Data from multiple irrigations of a commercial crop performed under normal farmer management demonstrated the potential of the simple relationship between Tadv and Tco to improve management of real irrigations.
Gillies, M.H., Smith, R.J., Williamson, B., Shanahan, M., 2010. Improving performance of bay irrigation through higher flow rates. In: Sydney. Australian Irrigation Conference and Exhibition 8-10 June 2010. https://www.researchgate.net/publication/ 228865098_Improving_Performance_of_Bay_Irrigation_Higher_Flow_Rates. Hume, I., 1993. Determination of infiltration characteristics by volume balance for border check irrigation. Agric. Wat. Manag 23, 23–29. https://www.sciencedirect.com/ science/article/pii/0378377493900186. Khatri, K.L., Smith, R.J., 2006. Real-time prediction of soil infiltration characteristics for the management of furrow irrigation. Irrig. Sci. 25, 33–43. https://www. researchgate.net/publication/225345932_Realtime_prediction_of_soil_infiltration_ characteristics_for_management_of_furrow_irrigation. Koech, R.K., Smith, R.J., Gillies, M.H., 2014. A real-time optimisation system for automation of furrow irrigation. Irrig. Sci. 32, 319–327. https://link.springer.com/ article/10.1007/s00271-014-0432-6. Lavis, A., Maskey, R., Qassim, A., 2006. Border-check Irrigation Design. Department of Primary Industries. Victoria, Agriculture Notes AG1262, ISSN 1329-8062. Maheshwari, B.L., McMahon, T.A., Turner, A.K., Campbell, B.J., 1988. An optimization technique for estimating infiltration characteristics in border irrigation. Agric. Wat. Manag. 13, 13–24. Mehta, B., Wang, Q.J., 2005. Soil hydraulic properties of the shepparton irrigation region. Department of Primary Industries, Tatura, Australia p176. Montgomery, J., Wigginton, D., 2008. Evaluating Furrow Irrigation Performance: Results From the 2006-07 Season. Surface Irrigation Cotton CRC Team Bulletin, CRC for Cotton Catchment Communities, Narrabri, NSW. https://www.cottoninfo.com.au/ publications/water-evaluating-furrow-irrigation-performance. Robertson, D., Wood, M., Wang, Q.J., 2004. Estimating hydraulic parameters for a surface irrigation model. Aus. J. Exp. Agr 44, 173–179. https://www.researchgate.net/ publication/248891607_Estimating_hydraulic_parameters_for_a_surface_irrigation_ model_from_field_conditions. Saha, R., Raghuwanshi, N.S., Upadhyaya, S.K., Wallender, W.W., Slaughter, D.C., 2011. Water sensors with cellular system eliminate tail water drainage in alfalfa irrigation. Calif. Agric. (Berkeley) 65 (4), 202–207. http://calag.ucanr.edu/Archive/?article= ca.v065n04p202. Selle, B., Wang, Q.J., Mehta, B., 2011. Relationship between hydraulic and basic properties for irrigated soils in southeast Australia. J. Plant Nutr. Soil Sci 1999 (174), 81–92. https://www.researchgate.net/publication/230451763_Relationship_ between_hydraulic_and_basic_properties_for_irrigated_soils_in_southeast_Australia. Skene, J.K.M., 1963. Soils and Land Use in the Deakin Irrigation Area. Technical Bulletin No 16. Victorian Department of Agriculture, Melbourne, Australia. http://vro. agriculture.vic.gov.au/DPI/Vro/gbbregn.nsf/ 0d08cd6930912d1e4a2567d2002579cb/a2bc9a9aee4b35f8ca257520001cb35b/ $FILE/Soil%20and%20Land%20Use%20in%20the%20Deakin%20Irrigation %20Area,%20Victoria.pdf. Skene, J.K.M., Freedman, J.R., 1944. Soils Survey of Part of the Shepparton Irrigation District. Technical Bulletin No 3. Victorian Department of Agriculture, Melbourne, Australia. http://vro.agriculture.vic.gov.au/dpi/vro/gbbregn.nsf/pages/soil_survey_ shepparton_irrigation. Skene, J.K.M., Harford, L.B., 1964. Soils and Land Use in the Rochester and Echuca Districts. Technical Bulletin No 17. Victorian Department of Agriculture, Melbourne, Australia. https://trove.nla.gov.au/work/27818329?q&versionId=33568343. Skene, J.K.M., Poutsma, T.J., 1962. Soils and Land Use in Part of the Goulburn Valley. Technical Bulletin No 14. Victorian Department of Agriculture, Melbourne, Australia. https://trove.nla.gov.au/work/27818350?selectedversion=NBD2583233. Smith, R.J., Raine, S.R., Minkovich, J., 2005. Irrigation application efficiency and deep drainage potential under surface irrigated cotton. Agric. Wat. Manag 71 (2), 117–130. https://www.researchgate.net/publication/222422797_Irrigation_ application_efficiency_and_deep_drainage_potential_under_surface_irrigated_cotton. Smith, R.J., Uddin, M.J., Gillies, M.H., 2018. Estimating irrigation duration for high performance furrow irrigation on cracking clay soils. Agric. Wat. Manag 206, 78–85.
References ABS, 2008. Water and the Murray Darling Basin – a Statistical Profile 2000-01 to 2005-06. Australian Bureau of Statistics, Canberra. https://www.abs.gov.au/ausstats/abs@ .nsf/mf/4610.0.55.007. Arnold, B.J., Upadhyaya, S.K., Wallender, W.W., Grismer, M.E., 2014. Sensor-based cutoff strategy for border check-irrigated fields. J. Irri. & Drain. Engg. 0401408, 1–9. https://ascelibrary.org/doi/10.1061/%28ASCE%29IR.1943-4774.0000855. Austin, N.R., Prendergast, J.B., 1997. Use of kinematic wave theory to model irrigation on cracking soil. Irrig. Sci. 18, 1–10. https://link.springer.com/article/10.1007/ s002710050038. Group, B.D.A., 2007. Cost Benefit Analyses of Research Funded by the CRDC. Report to Cotton Research and Development Corporation. BDA Group, Manuka, ACT, Australia. https://www.crdc.com.au/sites/default/files/pdf/BDA%20FINAL%20Hero %20Report%20(5-11).pdf. Butler, B.E., 1950. A theory of prior streams as a causal factor in the distribution of soils in the riverine plain of south eastern Australia. Aust. J. Agric. Res. 1, 231–252. Campbell, B., 1988. Report on Bay Hydraulics Studies at Kerang. Rural Water Commission. pp 135. Victoria. https://www.researchgate.net/publication/ 248894137_A_Theory_of_Prior_Streams_as_a_Causal_Factor_of_Soil_Occurrence_in_the_ Riverine_Plain_of_South-Eastern_Australia. Feyen, J., Zerihun, D., 1999. Assessment of the performance of border and furrow irrigation systems and the relationship between performance indicators and system variables. Agric. Wat. Manag 40, 353–362. https://www.sciencedirect.com/science/ article/pii/S0378377499000098. Finger, L., Morris, M., 2012. Implications of high flow irrigation for surface runoff and regional surface drainage management in the goulburn-Murray irrigation District (GMID). In: Australia. Irrigation Australia Ltd, Australian Irrigation Conference and Exhibition. Adelaide. GBCMA, 2012. Farm Water Frequently Asked Questions (FAQs) Round 3 - on Farm Irrigation Efficiency Program. accessed at http://www.gbcma.vic.gov.au/images/ 2012 20-faqsr3-september-updated.pdfon18March2013.. Goulburn Broken Catchment Management Authority (GBCMA). https://www.gbcma.vic.gov.au/ downloads/farm_water_program/farm_water_program_faqs_round_two_march_2011_ final.pdf. Gilfedder, M., Connell, L.D., Mein, R.G., 2000. Border irrigation field experiment. I: water balance. J. Irri. & Drain. Engg 126 (2), 85–91. https://www.researchgate.net/ publication/261760159_Border_Irrigation_Field_Experiment_I_Water_Balance. Gillies, M.H., Smith, R.J., 2015. SISCO – surface irrigation simulation calibration and optimisation. Irrig. Sci. 33, 339–355. https://link.springer.com/article/10.1007/ s00271-015-0470-8.
10
Agricultural Water Management 228 (2020) 105850
R. Smith and M. Uddin
border irrigation on cracking soils – reconciling field and infiltrometer measurements. Conference on Engineering in Agriculture and Food Processing, Engineers Australia, Paper SEA96/037. Tisdall, A.L., 1950. Infiltration studies on a red-brown earth in the Riverina. J. Aus. Inst. Agric. Sci. 26-29. https://publications.csiro.au/rpr/pub?list=BRO&pid= procite:bdacb402-63a4-4e80-a20b-3c86f552b0a1. Uniwater, 2008. Regional and Economic Benefits Through Smarter Irrigation. Final Report on STI Project. University of Melbourne. https://www.findanexpert.unimelb. edu.au/display/publication63080. Walker, W.R., 2005. Multilevel calibration of furrow irrigation infiltration and roughness. J Irrig. Drain. Engg 131, 129–136. https://ascelibrary.org/doi/10.1061/%28ASCE %290733-9437%282005%29131%3A2%28129%29.
https://www.sciencedirect.com/science/article/pii/S0378377418301604. Smith, R.J., Gillies, M.H., Shanahan, M., Campbell, B., Williamson, B., 2009. Evaluating the performance of bay irrigation in the GMID. In: Irrigation Australia, 2009 Irrigation & Drainage Conference. Swan Hill, Vic. pp. 18–21. October. https://www. researchgate.net/publication/239574224_Evaluating_the_Performance_of_Bay_ Irrigation_in_the_GMID. Smith, R.J., Uddin, M.J., Gillies, M.H., Moller, P., Clurey, K., 2016. Evaluating the performance of automated bay irrigation. Irrig. Sci. 34, 175–185. https://link.springer. com/article/10.1007/s00271-016-0494-8. Turral, H., 1996. Sensor placement for real time control of automated border irrigation. Conference on Engineering in Agriculture and Food Processing, Engineers Australia, Paper SEA96/036. Turral, H., Malano, H.M., 1996. A recirculating infiltrometer to investigate surge flow in
11
Contents lists available at ScienceDirect
Agricultural Water Management journal homepage: www.elsevier.com/locate/agwat
Selection of flow rate and irrigation duration for high performance bay irrigation
T
RJ Smitha, MJ Uddinb,* a b
Centre for Agricultural Engineering, University of Southern Queensland, Toowoomba, Qld, Australia Trangie Agricultural Research Centre, NSW Department of Primary Industries, Trangie, NSW, Australia
A R T I C LE I N FO
A B S T R A C T
Keywords: Surface irrigation Border-check irrigation Infiltration Simulation Application efficiency Optimization Time to cut-off
The maximum efficiency attainable by surface irrigation in any particular situation is determined largely by the soil infiltration characteristic and the flow rate onto the field. Performance evaluations have suggested that higher flow rates than those traditionally recommended can lead to increases of about 20% in the application efficiency of bay irrigation across the dairy regions of southern Australia. However, substantially reduced irrigation durations are required to realise these efficiency gains and greater precision is required in the selection and management of these shorter durations. In this paper, infiltration characteristic curves representative of the predominant (cracking and non-cracking) soils of the region are used in the surface irrigation simulation model SISCO, for a range of bay lengths and soil moisture deficits, to determine: (i) the flow rates required to achieve the maximum efficiency on the bay irrigated soils of southern Australia; and (ii) the means for real-time estimation of optimum irrigation durations. The aim is provision of guidance to irrigators seeking higher efficiency through the use of higher flow rates. Real-time estimation of optimum time to cut-off must recognise the substantial variations in infiltration within a soil type and with time due to changes in antecedent moisture content, and hence must be adaptive to the conditions prevailing at the time of any irrigation. A method for estimating time to cut-off (previously developed for furrow irrigation), is extended to the hydraulically more complex case of bay irrigation. The method is based on simple linear relationships between the advance time to a nominated point part way down the field and the time to cut-off. Its application to the management of irrigations is demonstrated using data from multiple irrigations of a single bay under varying soil moisture conditions.
1. Introduction Despite the evidence that well designed and managed surface irrigation systems have the potential to deliver high application efficiencies, most commercial systems still appear to be operating with significantly lower and highly variable efficiencies. The reason for this can be traced to the use of lower than optimal flow rates and more particularly the use of inappropriate times to cut-off. For example, in the Australian cotton industry, Smith et al. (2005) have shown that increasing the furrow inflow rates to 6 L/s and reducing the time to cut-off (Tco) would potentially raise the average application efficiency from the previous low of 50% to an improved 75%. More recent evaluations of furrow irrigation performance in that industry (Montgomery and Wigginton, 2008) have confirmed that adoption of higher flow rates has led to average application efficiencies in excess of 70%. As a result, the cotton industry is estimated to have
⁎
saved 28.5 G L of water per annum and achieved an industry improvement in WUE of 10% (BDA Group, 2007). Selection by farmers of appropriate times to cut-off remains problematic and the simulation work by Smith et al. (2005) suggested that raising efficiency further could only come about by selecting an individual Tco for each irrigation that gives optimum performance for the prevailing conditions. Substantial work toward real time selection of Tco has been undertaken with some success by Khatri and Smith (2006); Koech et al. (2014) and Smith et al. (2018). A similar situation prevails in the bay (border check) irrigated dairy and fodder cropping regions in southern Australia. A study by the Cooperative Research Centre for Irrigation Futures (Smith et al., 2009 and Gillies et al., 2010) demonstrated that significant gains (at least 20%) in application efficiency were possible in the bay irrigation of pasture and fodder crops, again by the simple expedient of doubling flow rates (from 0.07 to 0.22 L/s/m width to 0.24 to 0.44 L/s/m) and
Corresponding author. E-mail address: [email protected] (M. Uddin).
https://doi.org/10.1016/j.agwat.2019.105850 Received 22 February 2019; Received in revised form 4 October 2019; Accepted 6 October 2019 0378-3774/ © 2019 Elsevier B.V. All rights reserved.
Agricultural Water Management 228 (2020) 105850
R. Smith and M. Uddin
simulation model used in this study the infiltration data were converted to the form of the three parameter modified Kostiakov equation:
reducing irrigation durations. For growers seeking Federal Government funding for on-farm efficiency improvements, a minimum flow rate of 0.2 ML/d/m width is now required (GBCMA et al., 2012), based on the above work of Smith et al. (2009) and Gillies et al. (2010). No guidance was provided on selection of appropriate Tco. At these higher flow rates the very much reduced times to cut-off required mean that the ability to achieve ‘optimal’ performance is particularly sensitive to the cut-off time. For farmers to manage irrigations with these higher flow rates and shorter run times they will require, as a minimum, very clear and precise guidelines and it typically will necessitate real-time estimation of Tco. More recently, Smith et al. (2016) evaluated the irrigation performance achieved on nine farms employing automation of their irrigated bays, all of which used flow rates in excess of the recommended 0.2 ML/d/m. This study showed that with careful selection of cut-off times, application efficiencies in excess of 90% could be obtained routinely. Unfortunately some of the farmers were less diligent in their management of cut-off times and did not receive the full efficiency gains that are possible through automation. In other parts of the world researchers and farmers are wrestling with the same issues relating to management of bay irrigations. Efforts to reduce water use in the bay irrigation of lucerne (alfalfa) in California (for example, Saha et al., 2011 and Arnold et al., 2014) have tended to ignore flow rate (presumably because of infrastructure limitations) and have focussed on reducing tail-water by attempting to improve selection of time to cut-off. The purpose of the present study is to use simulation along with published infiltration and irrigation evaluation data to: (i) determine the flow rates that will provide best performance over the range of soil types across the dairy regions of northern Victoria and southern New South Wales, and (ii) assess methodologies that will allow adaptive real-time estimation of optimal times to cut-off.
I = kt a + fo τ
(1)
where I is the cumulative depth of infiltration (mm), a and k are fitted parameters, fo is the final or steady infiltration rate (mm/min) and τ is the infiltration time (min). The feature most evident from the infiltration data is the very substantial variation in infiltration within a field, within a soil type and between soil types. This is supported by the results of Selle et al. (2011) who showed that the within field and within soil type variation in final infiltration rate were not dissimilar in magnitude. Averaging infiltration over an irrigated bay (as is done by any measurement that uses the bay as the infiltrometer) masks some of the within field variation. However, the within soil type variation means that infiltration cannot be estimated from the soil descriptions provided by the available soil surveys. This poses a real challenge for the development of irrigation management guidelines which, as a consequence, need to be based on objective hydraulic criteria rather than on the basis of soil type. The data also suggest that the soils in the region are substantially more permeable than indicated by reports such as those by Mehta and Wang (2005) and Selle et al. (2011), based on point measurements of the soil hydraulic properties. This has significant implications for irrigation management, the potential magnitude of deep percolation losses and accessions to groundwater resulting from over irrigation. The other feature evident from the data is that the soils fall clearly into two key groups, cracking and non-cracking although it should be noted that different occurrences of a particular soil might fall into either group. However, soil cracking is a readily identifiable characteristic and farmers should have no difficulty in identifying into which group their soils fall. In the cracking group of soils the parameter a in the modified Kostiakov equation is zero or very small and the k parameter represents the instantaneous initial infiltration or crack fill. This in effect reduces the equation to the two parameter linear equation as used in the AIM model of Austin and Prendergast (1997):
2. Study Area This study pertains to the southern Murray Darling Basin region of northern Victoria and southern New South Wales (Fig. 1). The landform is an alluvial plain. Soils have been formed by floodplain deposition from the Murray River and tributary streams that have meandered over the floodplain for millennia. Soils range from coarse sands deposited adjacent to streams, to fine textured, cracking clays resulting from deposition of fine material remote from stream channels (Butler, 1950). Soil types across the region are as described in Skene (1963); Skene and Freedman (1944); Skene and Harford (1964), and Skene and Poutsma (1962) Border irrigation of perennial and annual pastures and fodder crops uses approximately 1600 G L of irrigation water annually on over 400,000 ha of irrigated land (ABS, 2008).
I = ZCR + fo τ
(2)
where ZCR is an instantaneous crack fill (mm). The crack fill is a function of the soil moisture deficit (Zreq) and for Lemnos loam is typically taken to be (Robertson et al., 2004):
ZCR = 0.75Zreq
(3)
The steady infiltration rate for these soils ranges from 0 to about 5 mm/h but is thought to be strongly skewed toward the lower values (Mehta and Wang, 2005). It is often simply taken to be 1 mm/h (e.g. Robertson et al., 2004). The published infiltration data compiled in this study are biased toward the cracking group of soils, partly because it contains some of the more prevalent soils of the region such as Lemnos loam and Kerang clay. Four infiltration curves (Fig. 2) were selected to represent these cracking soils: C1 – final infiltration rate 1 mm/h, deficit 40 mm (a 0.08, k 0.020, fo 0.0000167 m/min) C2 – final infiltration rate 1 mm/h, deficit 40 mm (a 0.08, k 0.020, fo 0.00005 m/min) C3 – final infiltration rate 3 mm/h, deficit 60 mm (a 0.08, k 0.032, fo 0.0000167 m/min) C4 – final infiltration rate 3 mm/h, deficit 60 mm (a 0.08, k 0.032, fo 0.00005 m/min) In the case of the non-cracking soils there is insufficient data to give any understanding of how the infiltration characteristic for a particular soil might vary (if at all) with deficit. Hence the strategy used here was to select three infiltration characteristics (Fig. 3) representative of the range of characteristics encountered, each of which is based on the published data for a named soil. The soils are:
3. Infiltration Data A detailed search of the published literature drew 51 infiltration characteristics covering 15 named soil types plus a further 8 characteristics where the soil type was not specified other than in general terms of soil texture. These data were originally gathered for a variety of purposes using of various measurement methods ranging from point measurements using ring infiltrometers (Tisdall, 1950; Robertson et al., 2004) and recirculating infiltrometers (Turral and Malano, 1996) to field averages where the irrigation bay was the infiltrometer. In this latter group, methods included volume balance approaches such as the two-point method (Maheshwari et al., 1988; Hume, 1993) to least squares fitting using hydrodynamic models (Turral, 1996; Austin and Prendergast, 1997; Smith et al., 2009; Gillies et al., 2010). Only in a very few instances was the soil moisture deficit at the time of the measurement specified in the source publication. The source data were reported in a variety of formats. To suit the 2
Agricultural Water Management 228 (2020) 105850
R. Smith and M. Uddin
Fig. 1. Location of study area (source Department of Primary Industries, Victoria).
SISCO hydraulic simulation model (Gillies and Smith, 2015) for a range of flow rates and for three bay lengths. For each combination of length and flow rate, the irrigation was optimised to determine the duration of inflow or time to cut-off (Tco) that would give maximum application efficiency (Ea). Run-off from the end of the bay was set to 5% to ensure that the irrigation always reached the downstream end of the bay. This is a small margin and may at times be insufficient compensate for errors in the estimation of parameters such as the Manning n. The implications of this and the control requirements have been discussed in later sections of the paper. For many (mostly cracking) soil types in the region the deficit is replenished rapidly (via the process of crack fill) hence the 5% runoff condition also typically ensures that the irrigation is
N1– low permeability - based on Wanalta loam or similar (a 0.4, k 0.0045, fo 0.0, approximately equivalent to a USDA 0.15 Light Clay). N2 – moderate permeability - Cobram loam (a 0.4, k 0.0055, fo 0.0, USDA 0.2 Clay Loam). N3 – higher permeability - a more permeable occurrence of Cobram loam (a 0.4, k 0.0066, fo 0.0, USDA 0.25 Clay Loam).
4. Simulations 4.1. Simulation methodology and optimisation objective For each of the case study soils, irrigations were simulated using the
Fig. 2. Representative infiltration characteristics for the cracking soils. 3
Agricultural Water Management 228 (2020) 105850
R. Smith and M. Uddin
Fig. 3. Representative infiltration characteristics for the non-cracking soils.
than those recommended in the earlier extension guidelines of Lavis et al. (2006). The exceptions are for longer bays on the more permeable soils with small moisture deficits where maximum application efficiencies reduce to below 80%. It must be stressed that these are maximum possible application efficiencies and assume perfect selection and control of irrigation duration (Tco). Few farmers would be achieving maximum efficiencies at their current flow rates and without assistance with selection of Tco would certainly not achieve them if they adopted higher flow rates. The simulations suggest preferred flow rates for cracking soils of 6 ML/d/ha for irrigation of pasture with soil moisture deficits up to about 50 mm. This should increase to 8 ML/d/ha for irrigation of deep rooted fodder crops with deficits in the order 80–100 mm. For pasture on non-cracking soils the recommended flow rates depend on the soil permeability with 6, 10 and 14 ML/d/ha recommended for lowly, moderately and more highly permeable soils, respectively. For a 400 m long 50 m wide bay, the above flow rates correspond to unit rates of 0.24, 0.4, and 0.56 ML/d/m width or 2.8, 4.6 and 6.7 L/s/m width, respectively. Again these rates should be increased by 2 ML/d/ha for higher deficits. In some instances the recommended flow rates might exceed the supply rate available, in which case growers will have to balance the potential small loss of application efficiency with the cost of reducing bay widths to increase the flow rate per unit width. At these higher flow rates the optimised times to cut-off are much shorter than those required for the lower flow rates traditionally used. It also means that cut-off will occur well before the advance has reach the halfway point down the field posing a significant challenge to any procedure employed to estimate the time to cut-off. One factor, not accounted for explicitly in the above simulations, that might limit application efficiencies on some sites is residual water stored in the micro-topography of the bay after the irrigation is finished. This factor is most significant in bays which are on heavier soils, on low slopes, imprecisely graded, with dense pasture and where traffic by stock has generated substantial micro-relief (Campbell, 1988). An early attempt to quantify this residual water is contained in the report by Campbell (1988) who measured residual depths of from 1.1 to 12.5 mm on two bays with slopes of 1:750 and 1:1500 respectively, with the larger values occurring on the flatter bay. In a detailed water balance of a particularly long duration and inefficient irrigation on a bay with heavy clay soil, Gilfedder et al. (2000) showed that on the particular site, 20 mm of water remained on the surface of the bay 20 h after the irrigation. For modern steeper and well graded bays and where high flow rates and much shorter irrigation durations are employed this term should be small. In cases where the infiltration characteristic has been determined from irrigation advance data this depth of residual
adequate in meeting the requirements of the crop. The other usual performance parameters of requirement efficiency (Er) and uniformity were also calculated for each irrigation, along with an estimate of the deep percolation loss. The optimisation objective used in studies such as this is always subjective. Individual researchers and growers place different values on the key performance parameters depending on their particular circumstances. For example, it is not uncommon for the requirement efficiency to be set at 100%, meaning complete replenishment of the soil moisture deficit (e.g. Feyen and Zerihun, 1999). Such a strategy inevitably results in substantial runoff and deep drainage volumes that reduce application efficiency significantly, and arguably is not a valid strategy on the more permeable soils or when water is in short supply. As well as reduced application efficiency it also leads to reduced water use efficiency (WUE). This was well illustrated by Uniwater (2008) who showed substantial increases in WUE on irrigated lucerne by reducing cut-off times below those required for 100% Er. An alternative strategy sometimes used is to maximise uniformity (e.g. Finger and Morris, 2012). This is only valid if Er is substantially less than 100%. If Er is 100% then the uniformity of moisture stored in the root zone is by definition 100% and this maximisation objective is non-sensical.
4.2. Simulation results The full set of results covers a very large number of combinations of soil type, moisture deficit, field length, inflow rate, slope and crop type (Manning n). Typical results for the cracking and non-cracking soil groups are provided in Figs. 4 and 5 respectively, as plots of the key performance measures Ea and Er versus flow rate. The optimum time to cut-off (Tco) and the distance reached by the irrigation advance (Xco) at the point of cut-off are also shown on these plots. In these plots, as in most cases simulated, the application efficiencies initially show a rising trend with increasing flow rate until reaching a plateau at some particular flow rate. The magnitude of this flow rate is a function of the soil type, crop density, bay length and soil moisture deficit, being higher for the more permeable soils, denser crops, longer bays and higher deficits. At or just before the point of maximum Ea, the requirement efficiency Er was generally adequate (> 95%) thus defining the preferred flow rate. At this flow rate the irrigation durations (Tco) were typically about 2–3 hours except for the very long bays and high deficits. The exceptions to the above were the cracking soil types with a short bay and low deficit, where the application efficiency was essentially constant at all flow rates. These data show that application efficiencies in excess of 90% are possible on many soils in the region through the use of flow rates higher 4
Agricultural Water Management 228 (2020) 105850
R. Smith and M. Uddin
Fig. 4. Example results for summer pasture (n = 0.26) in a 400 m long bay slope 1:750 on cracking soil (C1) and 40 mm moisture deficit.
pasture or fodder crop can be anywhere in the range 0.1 to 0.36 whereas n for a bare furrow is usually taken to be 0.04. This means that in a typical bay at times at or near cut-off there is a far greater volume of water contained in temporary storage in the advancing irrigation flow. This is illustrated in Fig. 6 for an efficient irrigation on a hypothetical irrigation bay employing a high flow rate and short time to cut-off. At the point of cut-off (2 h) between 50 and 60% of the water applied to the field is still on the field. The problem is compounded by the substantial difficulty in accurately estimating the volume of this surface ponding. Finally, because of the much shorter times to cut-off, any sensing and calculations required to estimate Tco and the subsequent decision to turn the flow off have to be made much earlier in the irrigation (typically before the advance has reached 40% of the length) and more quickly than in furrow irrigation. To be effective, methods for selection of Tco need to be simple to apply, involve the minimum of measurement and calculation on the part of farmers, but at the same time be adaptive to changes in the key variables of flow rate, soil moisture deficit, and crop density. The rate of advance of the wetting front down the bay integrates the effects of most of the relevant variables, including those above. It is also something that irrigators can measure relatively simply. It is also assumed that irrigators will at least also know the inflow rate into their bays. Some simplification of the problem can result if farmers irrigate each time at
water is most likely already embedded in the estimated infiltration characteristic, most probably in the initial infiltration term k, and has thus been accounted for in the efficiency calculations. There is no evidence in the literature of significant residual water occurring on the non-cracking soil types.
5. Toward a method for REAL-TIME selection of time to CUT-OFF 5.1. The challenge Recent work on furrow irrigation in the Australian cotton industry by Khatri and Smith (2006); Koech et al. (2014) and Smith et al. (2018) has led to two equally effective methods for estimation Tco in real time. They differ in complexity but both only require one measurement of the irrigation advance taken relatively late in the irrigation. However it is not known how these methods might translate to bay irrigation. In Australia bay irrigation differs from furrow in a number of significant ways. Firstly, with lengths ranging from about 200 to 600 m bays are much shorter than the 600 to 1500 m furrows used in the cotton industry. Secondly, the hydraulic resistance to the irrigation flow provided by the crop or soil surface (as represented by the parameter n in the Manning equation) is typically much greater and more variable than in a clean furrow. For example the Manning n for a summer
Fig. 5. Example results for summer pasture (n = 0.26) on a permeable non-cracking soil N3, length 400 m, slope 1:750 and 60 mm deficit. 5
Agricultural Water Management 228 (2020) 105850
R. Smith and M. Uddin
Fig. 6. Schematic of the water balance of an efficient bay irrigation event employing a high flow rate and short time to cut-off.
Saha et al. (2011) and Arnold et al. (2014) proposed the use of a simple volume balance approach and applied it to bay irrigated Lucerne in southern California with mixed results. The method utilises measurement of advance at two points as well as an estimate of the average depth of surface ponding. Two issues that limit the application of this method are immediately evident. First is the use of a very much simplified infiltration equation that consisted only of the initial rapid infiltration term k, and which neglected any continuing infiltration described by fo. For many of the soils of the region which display a nonnegligible continuing infiltration the method will underestimate the depth or volume of infiltration and consequently will underestimate the Tco thus introducing the real possibility that the advance will not reach the end of the field. There is no obvious way to introduce the continuing infiltration into the volume balance analysis. The second issue is the fact that the calculation of the infiltration term and hence Tco is very sensitive to the (estimated) value of the average depth of water on the field. Estimating this term from a limited number (only two) of depth measurements is highly problematic. For furrow irrigation on the heavy clay soils in the Australian cotton growing regions, Smith et al. (2018) showed that for a particular field there exists a simple linear relationship between Tco and the time of the irrigation advance to a nominated point at least 50% of the distance down the furrow. This relationship was shown to be independent of the flow rate and the soil moisture deficit, knowledge of which was not required. The method was shown to provide robust estimates of Tco that resulted in high application efficiencies. The remainder of the present paper will focus on the possible application of this approach to bay irrigated pasture and fodder crops.
the same soil moisture deficit thus elimination problems caused by variation of the infiltration characteristic of the bay, and if they use the same and constant flow rate each time they irrigate. This is now possible in the region under consideration following full automation of the supply system. 5.2. The options From an examination of the literature, three methods appeared worthy of consideration, viz: 1 The use of a hydrodynamic model in a manner similar to the Autofurrow model of Khatri and Smith (2006) and Koech et al. (2014), 2 A volume balance approach following on from the work of Saha et al. (2011) and Arnold et al. (2014), and 3 Simple unique empirical relationships for each bay linking Tco with the rate of advance to some nominated point down the bay, similar to that used in furrow irrigation by Smith et al. (2018). The Autofurrow method for the real-time selection of Tco in furrows (Khatri and Smith, 2006; Koech et al., 2014) requires measurements only of the inflow rate and advance to a single point. The main features of this optimization process are: the use of a model infiltration curve and a scaling process to describe the current soil infiltration characteristic; measurement of the inflow rate to the furrows; measurement of the water advance at a point at least midway down the furrow; an assumed constant Manning n; and a hydraulic simulation program based on the full hydrodynamic model to predict the optimum time to cut-off. To apply this method to bay irrigation requires additional measurements (for example, additional advance points and/or flow depths) to enable estimation of the Manning n parameter for the crop or pasture growing in the bay. Even so, a major difficulty arises because n and the infiltration parameter k both influence the rate of advance in a similar manner, that is, an increase in either parameter causes the advance to slow. However they have opposing effects on the volume of surface ponding. Hence multiple measurements of flow depth are required to isolate the values of these two parameters. It is easily shown through simulation that the number and location (at or near the downstream end of the field) of sensors and measurements required to identify the correct values of n and k make this approach prohibitive as a practical approach for farmers to employ. This issue with the sensor locations was first intimated by Walker (2005) and further demonstrated by Gillies and Smith (2015) during the initial testing of the SISCO model.
5.3. Relationship between Tco and advance time for irrigation bays The simulations used in the previous section to establish preferred flow rates involved optimisation to give maximum Ea and 5% runoff for a wide range of the key variables. Here these same data are used to explore the relationship between advance time and optimum Tco. For this purpose the advance down the bay was represented by the time taken to reach a point 3/8th of the distance down the bay (Tadv). This distance used for the advance time was selected arbitrarily but in all cases the advance time was less than the Tco hence would give the farmer time to make the control decision. An example of the results, in this case for each of the cracking and non-cracking soil types (C1, C2, C3, C4, N1, N2 & N3) for summer pasture (n = 0.26) on a hypothetical bay 400 m long, and with an inflow rate of 2.31 L/s/m width are presented in Fig. 7. The two lines representing the cracking soils (C1 & C3, C2 & C4) were each calculated for various values of the infiltration 6
Agricultural Water Management 228 (2020) 105850
R. Smith and M. Uddin
Fig. 7. Optimum Tco vs advance time to 150 m (3/8th distance) for summer pasture in a 400 m long bay for various soil types and moisture deficits and an inflow rate of 2.31 L/s/m width of bay.
the results for a range of soil moisture deficits ranging from 30 to 60 mm. The cause of the difference between the results for bay and furrow irrigation appears to lie in the higher value of the Manning n for pasture and the very short advance distances necessarily used in this analysis. The matter is further complicated by the fact that a different set of curves also results for different crop densities, that is for different values of the Manning n (Fig. 9). In this figure curves are presented for a single flow rate and for bare soil (n = 0.05), sparse annual pasture or fodder crop (n = 0.1), summer pasture (n = 0.26), and a very thick summer pasture or crop (n = 0.3). The difference caused by the Manning n can be explained by recourse to a simple volume balance. For a given deficit and the same flow rate, the same Tco will be required in each case to apply the correct volume of water to satisfy the deficit. The higher the Manning n means greater flow depths, greater temporary storage of water on the surface of the bay and hence slower advance. A similar situation applies to the non-cracking soils. While not presented here, the relationship between Tadv and Tco is also dependent on the length of the bay.
parameter k. For the non-cracking soils (N1, N2 and N3) a single value of the soil moisture deficit was used. For each case the results show linear relationships between Tco and Tadv. It should be noted here that changing the values of the key parameters will result in different relationships. Changing the point used for the advance measurement simply results in another family of straight lines, as does altering the inflow rate or Manning n. Similarly changing the target runoff from 5% to reflect a different risk or uncertainty also results in another family of straight lines. For the cracking soil, the soil infiltration characteristic, and in particular the rapid initial infiltration, are known to be a function of the soil moisture deficit (Eq. (3) and associated text). The majority of the infiltration occurs in the first few minutes that water is present on the surface with only relatively small depths infiltrating after that time. This tends to result in a relatively uniform rate of advance. Because the two lines for the cracking soils are each for various values of the infiltration parameter k and they therefore represent varying deficits, suggesting that the relationships for these soils are independent of the soil moisture deficit. This is consistent with the similar result for furrow irrigation on cracking soils (Smith et al., 2018). The difference between them is the effect of the different rates of continuing infiltration represented by different values of the parameter fo (1 mm/h for C1 & C3 and 3 mm/h for C2 & C4). The non-cracking soils vary from the cracking soils in that the relationship between the infiltration characteristic and the soil moisture deficit is not well known. The three non-cracking soils (N1, N2 & N3) are in order of increasing infiltration (increasing k) the results for which fall on a single straight line (Fig. 7) for the given flow rate, deficit and crop condition. This difference in infiltration between the three soils was primarily intended to show how the advance vs Tco relationship might change with increasing soil permeability. However it also might show the effect of increasing soil moisture deficit. Assuming the infiltration curve varies with deficit by varying the parameter k (as with the cracking soils) then the relationship for these soils might also be largely independent of the soil moisture deficit. (Note that while the selection of the optimum Tco might be independent of the deficit the resulting application efficiency is very much dependent on it.) The values of Tco for the non-cracking soils are typically higher than for the cracking soils at the same flow rate. This reflects the very different shape of the infiltration curves (and the greater ongoing infiltration that occurs at later times). The greater permeability also results in lower maximum application efficiencies. The previous furrow irrigation work by Smith et al. (2018) showed that for furrows the relationship between advance and Tco was also independent of inflow rate. This not the case for bay irrigation, as illustrated in Fig. 8 which shows the relationships for a cracking (C1) soil, each line corresponding to a particular flow rate, and represents
5.4. Sensitivity analysis The relationship for the cracking soil (C1) presented in Fig. 7 is: Tco = 3.19 Tadv −235
(4)
Using the same data as used to develop this relationship the sensitivity of the irrigation performance to changes or errors in the soil moisture deficit, flow rate and Manning n were assessed by varying each of those parameters by ± 10% (Table 1). As expected a change in the deficit of 10% causes only a minor change in performance with the application efficiency changing by 2%. However performance is a little more sensitive to changes in the inflow rate and Manning n. A 10% change in these parameters causing about a 5% change in performance and possible instances of failure to reach the end of the field. For irrigators in the Goulburn Murray Irrigation District of northern Victoria, full automation of the supply system has provided them with constant and precisely known flow rates, removing the likelihood of errors or uncertainties in flow rates. The uncertainty involved in estimating the Manning n is more problematic. However this can be countered effectively by either erring toward higher values of n or by aiming for a slightly higher runoff volume when establishing the Tco/Tadv relationship. Either approach provides an additional margin of safety. Irrigating with a consistent flow rate at a consistent soil moisture deficit will help remove some uncertainty.
7
Agricultural Water Management 228 (2020) 105850
R. Smith and M. Uddin
Fig. 8. Optimum Tco vs advance time to 150 m for summer pasture in a 400 m long bay on a cracking soil at various inflow rates.
5.5. Management implications
Table 1 Results of sensitivity analysis of irrigation performance with the variation of soil moisture deficit, flow rate and Manning n/.
To illustrate application of the method to real on-farm management of irrigations, data from five sequential irrigations in a single bay planted to maize being grown for fodder were used. These data were obtained in an evaluation of automated bay irrigation undertaken in 2014 by Smith et al. (2018) and reported in that paper as farm D. The bay is 400 m long and 50 m wide with an average slope of 1:916. Soil type is a heavy clay loam. Each of the five irrigations were monitored and evaluated, with continuous measurement of the inflow hydrograph and the flow depths at three points down the field. Infiltration parameters and the Manning n were determined by a calibration process using the simulation model SISCO (Gillies and Smith, 2015) and the performance measures (Ea, Er, runoff and deep drainage) estimated again using the SISCO model. Inflow rate was the same for each irrigation at 14 ML/d but the soil moisture deficit prior to each irrigation varied wildly (Table 2) resulting in infiltration characteristics of very different magnitude (Fig. 10). The Manning n increased from 0.06 to 0.09 throughout the sequence (Table 2). In addition to the actual performance achieved by the farmer, each irrigation was optimised using SISCO with the results
Control Deficit +10% Deficit -10% Inflow +10% Inflow-10% n +10% n -10% a
Tadv
Tco
Ea
Er
112.0 118.0 105.5 104.5 119.0 116.0 107.0
122.3 141.4 101.5 98.4 144.6 135.0 106.3
92.8 89.5 94.8 99.1 87.5 84.5 99.6
98.2 99.7 92.6 92.9a 98.7 98.8 91.7a
Advance stopped 5% short of the end of field.
also given in Table 2. To develop the relationship between Tco and the time of advance to the 3/8th distance for a particular bay at least one irrigation at the site must be evaluated as described above. For farm D the relationship was developed from the data from the first irrigation (29 Jan 14) by varying the deficit and hence the infiltration parameter k and optimising as above with the result given by the trend line in Fig. 11. The relationship
Fig. 9. Optimum Tco vs advance time to 150 m for various crop conditions (different Manning n values) in a 400 m long bay on a cracking soil and inflow rate of 3.09 L/s/m width. 8
Agricultural Water Management 228 (2020) 105850
R. Smith and M. Uddin
Table 2 Comparison of farmer, optimum and real-time managed irrigation performance – farm D. Farmer
Optimum
Managed (equation 5)
Date
Deficit (mm)
n
Tco (min)
Ea %
Tco (min)
Ea %
Tco (min)
Ea %
Runoff %
29 Jan 14 4 Feb 14 11 Feb 14 23 Feb 14 8 Mar 14
29 45 57 31 43
0.060 0.060 0.070 0.083 0.090
110 110 110 115 110
55 70 77 48 67
43.0 67.6 82.1 33.5 64.0
95 95 95 95 95
43.0 67.4 92.0 48.7 73.2
95.0 95.0 87.4 76.4 87.7
5.0 5.0 12.6 23.6 12.3
bay the efficiencies obtained would have been very much improved from those obtained by the farmer (Table 2). The equation accommodates the varying soil moisture deficits and varying infiltration but does not account for the increases in the Manning n with time as the crop develops. Hence the efficiencies that would have been obtained in the latter part of the season are slightly sub-optimal. While one data set is not sufficient to allow firm conclusions to be drawn it does illustrate the potential of the simple approach for estimating optimal Tco based on a measured time of advance to a nominated point. It has the advantage of requiring only a single sensor in each bay but does require full monitoring and evaluation of one irrigation to enable the relationship between advance time and optimal Tco to be developed. High flow rate bay irrigation with its attendant early cut-off is a radical departure from the traditional practice in northern Victoria. It does carry some risk but offers high rewards in terms of high application efficiencies, minimal deep percolation losses, reduced waterlogging of surface soils and reduced accessions to the water-table. Those irrigators who have adopted the practice have not all achieved these rewards (Smith et al., 2018) and those who have did so by trial and error. The method described in this paper offers improved performance to all adopters.
is: Tco = 2.89 Tadv – 65.5
(5)
Tco for all five optimised irrigations are also plotted against the corresponding advance time to the 3/8th distance in Fig. 11. Finally, the irrigation performance that would have been achieved had the remaining irrigations been controlled by equation 5 are presented in the final columns of Table 2. The significant issues with the management of irrigations on Farm D are:
• a lowly permeable soil with an unusual infiltration characteristic, • the uncertain and varying deficit, and • the variation in Manning n. In his attempt to satisfy the soil moisture deficit, the farmer used excessively long irrigation durations (Tco) the consequences of which were low application efficiencies and large volumes of runoff (Table 2). Runoff in fact was the sole loss at each irrigation, there were no deep percolation losses. The optimisations of the individual irrigations, undertaken with the benefit of hindsight, show that very much higher application efficiencies could have been obtained through substantial reductions in irrigation duration (which would vary in accordance with the deficit). However in all cases the optimised irrigations failed to fully satisfy the deficit. Because the farm is fully automated this under-irrigation is not a major issue, it simply means that the next irrigation will be required a little sooner. It also might lead to some yield improvement through a reduction in waterlogging of these heavy soils. It is worth noting at this point that the deficits were measured using an uncalibrated capacitance probe and that while the relativity between irrigations would likely be correct the magnitudes of the deficits may not be. Had the irrigations been controlled by equation 5 in conjunction with a single advance measurement located 150 m from the head of the
6. Conclusions Published data on soil infiltration properties across the Murray Dairy region of southern Australia were used in a simulation study of bay irrigation performance. It has confirmed that flow rates higher than those recommended by previous guidelines have the potential to deliver higher application efficiencies for bay irrigated pasture and fodder crops across this region. Preferred flow rates for the major soil groups are recommended. These higher flow rates result in much shorter irrigation durations (or times to cut-off) and consequently greater accuracy and precision in the selection and management of time to cut-off is
Fig. 10. Infiltration characteristics for farm D (from Smith et al., 2017). 9
Agricultural Water Management 228 (2020) 105850
R. Smith and M. Uddin
Fig. 11. Optimum Tco vs advance time for farm D.
required. The same simulation data were used to illustrate the relationships that exist between the rate of advance of the wetting front down an irrigated bay and the optimum time to cut-off for that bay. It is shown that if a bay is managed consistently, that is, irrigated at a consistent soil moisture deficit and consistent flow rate, a simple linear function can be used to estimate the Tco from a single advance measurement taken relatively early in the irrigation. Data from multiple irrigations of a commercial crop performed under normal farmer management demonstrated the potential of the simple relationship between Tadv and Tco to improve management of real irrigations.
Gillies, M.H., Smith, R.J., Williamson, B., Shanahan, M., 2010. Improving performance of bay irrigation through higher flow rates. In: Sydney. Australian Irrigation Conference and Exhibition 8-10 June 2010. https://www.researchgate.net/publication/ 228865098_Improving_Performance_of_Bay_Irrigation_Higher_Flow_Rates. Hume, I., 1993. Determination of infiltration characteristics by volume balance for border check irrigation. Agric. Wat. Manag 23, 23–29. https://www.sciencedirect.com/ science/article/pii/0378377493900186. Khatri, K.L., Smith, R.J., 2006. Real-time prediction of soil infiltration characteristics for the management of furrow irrigation. Irrig. Sci. 25, 33–43. https://www. researchgate.net/publication/225345932_Realtime_prediction_of_soil_infiltration_ characteristics_for_management_of_furrow_irrigation. Koech, R.K., Smith, R.J., Gillies, M.H., 2014. A real-time optimisation system for automation of furrow irrigation. Irrig. Sci. 32, 319–327. https://link.springer.com/ article/10.1007/s00271-014-0432-6. Lavis, A., Maskey, R., Qassim, A., 2006. Border-check Irrigation Design. Department of Primary Industries. Victoria, Agriculture Notes AG1262, ISSN 1329-8062. Maheshwari, B.L., McMahon, T.A., Turner, A.K., Campbell, B.J., 1988. An optimization technique for estimating infiltration characteristics in border irrigation. Agric. Wat. Manag. 13, 13–24. Mehta, B., Wang, Q.J., 2005. Soil hydraulic properties of the shepparton irrigation region. Department of Primary Industries, Tatura, Australia p176. Montgomery, J., Wigginton, D., 2008. Evaluating Furrow Irrigation Performance: Results From the 2006-07 Season. Surface Irrigation Cotton CRC Team Bulletin, CRC for Cotton Catchment Communities, Narrabri, NSW. https://www.cottoninfo.com.au/ publications/water-evaluating-furrow-irrigation-performance. Robertson, D., Wood, M., Wang, Q.J., 2004. Estimating hydraulic parameters for a surface irrigation model. Aus. J. Exp. Agr 44, 173–179. https://www.researchgate.net/ publication/248891607_Estimating_hydraulic_parameters_for_a_surface_irrigation_ model_from_field_conditions. Saha, R., Raghuwanshi, N.S., Upadhyaya, S.K., Wallender, W.W., Slaughter, D.C., 2011. Water sensors with cellular system eliminate tail water drainage in alfalfa irrigation. Calif. Agric. (Berkeley) 65 (4), 202–207. http://calag.ucanr.edu/Archive/?article= ca.v065n04p202. Selle, B., Wang, Q.J., Mehta, B., 2011. Relationship between hydraulic and basic properties for irrigated soils in southeast Australia. J. Plant Nutr. Soil Sci 1999 (174), 81–92. https://www.researchgate.net/publication/230451763_Relationship_ between_hydraulic_and_basic_properties_for_irrigated_soils_in_southeast_Australia. Skene, J.K.M., 1963. Soils and Land Use in the Deakin Irrigation Area. Technical Bulletin No 16. Victorian Department of Agriculture, Melbourne, Australia. http://vro. agriculture.vic.gov.au/DPI/Vro/gbbregn.nsf/ 0d08cd6930912d1e4a2567d2002579cb/a2bc9a9aee4b35f8ca257520001cb35b/ $FILE/Soil%20and%20Land%20Use%20in%20the%20Deakin%20Irrigation %20Area,%20Victoria.pdf. Skene, J.K.M., Freedman, J.R., 1944. Soils Survey of Part of the Shepparton Irrigation District. Technical Bulletin No 3. Victorian Department of Agriculture, Melbourne, Australia. http://vro.agriculture.vic.gov.au/dpi/vro/gbbregn.nsf/pages/soil_survey_ shepparton_irrigation. Skene, J.K.M., Harford, L.B., 1964. Soils and Land Use in the Rochester and Echuca Districts. Technical Bulletin No 17. Victorian Department of Agriculture, Melbourne, Australia. https://trove.nla.gov.au/work/27818329?q&versionId=33568343. Skene, J.K.M., Poutsma, T.J., 1962. Soils and Land Use in Part of the Goulburn Valley. Technical Bulletin No 14. Victorian Department of Agriculture, Melbourne, Australia. https://trove.nla.gov.au/work/27818350?selectedversion=NBD2583233. Smith, R.J., Raine, S.R., Minkovich, J., 2005. Irrigation application efficiency and deep drainage potential under surface irrigated cotton. Agric. Wat. Manag 71 (2), 117–130. https://www.researchgate.net/publication/222422797_Irrigation_ application_efficiency_and_deep_drainage_potential_under_surface_irrigated_cotton. Smith, R.J., Uddin, M.J., Gillies, M.H., 2018. Estimating irrigation duration for high performance furrow irrigation on cracking clay soils. Agric. Wat. Manag 206, 78–85.
References ABS, 2008. Water and the Murray Darling Basin – a Statistical Profile 2000-01 to 2005-06. Australian Bureau of Statistics, Canberra. https://www.abs.gov.au/ausstats/abs@ .nsf/mf/4610.0.55.007. Arnold, B.J., Upadhyaya, S.K., Wallender, W.W., Grismer, M.E., 2014. Sensor-based cutoff strategy for border check-irrigated fields. J. Irri. & Drain. Engg. 0401408, 1–9. https://ascelibrary.org/doi/10.1061/%28ASCE%29IR.1943-4774.0000855. Austin, N.R., Prendergast, J.B., 1997. Use of kinematic wave theory to model irrigation on cracking soil. Irrig. Sci. 18, 1–10. https://link.springer.com/article/10.1007/ s002710050038. Group, B.D.A., 2007. Cost Benefit Analyses of Research Funded by the CRDC. Report to Cotton Research and Development Corporation. BDA Group, Manuka, ACT, Australia. https://www.crdc.com.au/sites/default/files/pdf/BDA%20FINAL%20Hero %20Report%20(5-11).pdf. Butler, B.E., 1950. A theory of prior streams as a causal factor in the distribution of soils in the riverine plain of south eastern Australia. Aust. J. Agric. Res. 1, 231–252. Campbell, B., 1988. Report on Bay Hydraulics Studies at Kerang. Rural Water Commission. pp 135. Victoria. https://www.researchgate.net/publication/ 248894137_A_Theory_of_Prior_Streams_as_a_Causal_Factor_of_Soil_Occurrence_in_the_ Riverine_Plain_of_South-Eastern_Australia. Feyen, J., Zerihun, D., 1999. Assessment of the performance of border and furrow irrigation systems and the relationship between performance indicators and system variables. Agric. Wat. Manag 40, 353–362. https://www.sciencedirect.com/science/ article/pii/S0378377499000098. Finger, L., Morris, M., 2012. Implications of high flow irrigation for surface runoff and regional surface drainage management in the goulburn-Murray irrigation District (GMID). In: Australia. Irrigation Australia Ltd, Australian Irrigation Conference and Exhibition. Adelaide. GBCMA, 2012. Farm Water Frequently Asked Questions (FAQs) Round 3 - on Farm Irrigation Efficiency Program. accessed at http://www.gbcma.vic.gov.au/images/ 2012 20-faqsr3-september-updated.pdfon18March2013.. Goulburn Broken Catchment Management Authority (GBCMA). https://www.gbcma.vic.gov.au/ downloads/farm_water_program/farm_water_program_faqs_round_two_march_2011_ final.pdf. Gilfedder, M., Connell, L.D., Mein, R.G., 2000. Border irrigation field experiment. I: water balance. J. Irri. & Drain. Engg 126 (2), 85–91. https://www.researchgate.net/ publication/261760159_Border_Irrigation_Field_Experiment_I_Water_Balance. Gillies, M.H., Smith, R.J., 2015. SISCO – surface irrigation simulation calibration and optimisation. Irrig. Sci. 33, 339–355. https://link.springer.com/article/10.1007/ s00271-015-0470-8.
10
Agricultural Water Management 228 (2020) 105850
R. Smith and M. Uddin
border irrigation on cracking soils – reconciling field and infiltrometer measurements. Conference on Engineering in Agriculture and Food Processing, Engineers Australia, Paper SEA96/037. Tisdall, A.L., 1950. Infiltration studies on a red-brown earth in the Riverina. J. Aus. Inst. Agric. Sci. 26-29. https://publications.csiro.au/rpr/pub?list=BRO&pid= procite:bdacb402-63a4-4e80-a20b-3c86f552b0a1. Uniwater, 2008. Regional and Economic Benefits Through Smarter Irrigation. Final Report on STI Project. University of Melbourne. https://www.findanexpert.unimelb. edu.au/display/publication63080. Walker, W.R., 2005. Multilevel calibration of furrow irrigation infiltration and roughness. J Irrig. Drain. Engg 131, 129–136. https://ascelibrary.org/doi/10.1061/%28ASCE %290733-9437%282005%29131%3A2%28129%29.
https://www.sciencedirect.com/science/article/pii/S0378377418301604. Smith, R.J., Gillies, M.H., Shanahan, M., Campbell, B., Williamson, B., 2009. Evaluating the performance of bay irrigation in the GMID. In: Irrigation Australia, 2009 Irrigation & Drainage Conference. Swan Hill, Vic. pp. 18–21. October. https://www. researchgate.net/publication/239574224_Evaluating_the_Performance_of_Bay_ Irrigation_in_the_GMID. Smith, R.J., Uddin, M.J., Gillies, M.H., Moller, P., Clurey, K., 2016. Evaluating the performance of automated bay irrigation. Irrig. Sci. 34, 175–185. https://link.springer. com/article/10.1007/s00271-016-0494-8. Turral, H., 1996. Sensor placement for real time control of automated border irrigation. Conference on Engineering in Agriculture and Food Processing, Engineers Australia, Paper SEA96/036. Turral, H., Malano, H.M., 1996. A recirculating infiltrometer to investigate surge flow in
11