Using a risk based, site-speci c Decision Support System to determine the suitability of mine water for irrigation John G Annandale1, H Meiring du Plessis1, Phil D Tanner1, Jo Burgess2

1Department of Plant and Soil Sciences, University of Pretoria, Private Bag X20, Hat eld 0028, South Af- rica, [email protected] 2Water Research Commission, Private Bag X03, Gezina, 0031, South Africa

Abstract Use of irrigation as a mine water management option o en presents a cost-e ective means of utilising poor quality mine-impacted waters that would otherwise need to be treated before release into surface water bodies. However, there are some waters that should not be considered for long-term irrigation, and site selection is critical. A newly developed risk based, site-speci c irrigation water quality assessment tool de nes key factors to consider when determining the viability and sustainability of irrigation with speci c mine waters. Keywords: Irrigation water quality, crop yield, crop quality, soil quality, Decision Sup- port System

Introduction ies will likely be permitted and desirable. As  ere is much interest in the bene cial use this is unlikely with most mine-impacted of mine water for agricultural irrigation. Ir- waters, Tier 2 supplies more site-speci c rigation is o en a cost-e ective means for guidelines, enabling the user to design a crop operating mines to manage surplus water. production system to best accommodate the Upon closure, irrigation may present a sus- speci c water quality. If there are still con- tainable means for communities to diversify cerns about the usability of water for irriga- away from mining, by producing food and tion, then a Tier 3 assessment is indicated.  bre sustainably, and creating employment.  is will require detailed expert input to as- Large savings in water treatment costs are sess whether or not irrigation is at all feasible, also likely to follow. However, not all mine and if concerns highlighted by the Tier 2 as- waters are suitable for irrigation, and support sessment can be mitigated. Finally, the DSS is necessary to make informed decisions on is electronic and user-friendly, with colour suitability. coding to make the suitability of waters for ir- An irrigation water quality Decision Sup- rigation intuitive. Help  les provide informa- port System (DSS) has recently been devel- tion regarding the current state of knowledge oped (du Plessis et al 2017). Fitness-for-use of for suitability indicators, and describe the water is presented as being ‘ideal’, ‘acceptable’, approach and calculating procedures used in ‘tolerable’ or ‘unacceptable’.  e DSS is novel the DSS. in a number of ways. Firstly, it is risk based, enabling the user to assess the implications Methods of irrigating with a range of waters, includ- Water quality assessments at Tier 2 employ a ing mining impacted waters on soil and crop scaled down version of the Soil Water Balance resources, as well as on irrigation equipment. (SWB) crop growth and solute balance model Secondly, the guidelines are structured in (Annandale et al 2011), to dynamically simu- three tiers. Tier 1 provides generalised, con- late the interactions between irrigation water servative estimates of the suitability of water constituents and the soil-crop-atmosphere for irrigation. If mine waters are shown to be system. A simple cascading 11 layer soil wa- ideal or acceptable at this level, there may be ter balance is used, with default parameters no need to treat water or to utilise it through ( eld capacity, permanent wilting point, bulk irrigation, and release into surface water bod- density, and drainage characteristics) for pre-

Wolkersdorfer, Ch.; Sartz, L.; Weber, A.; Burgess, J.; Tremblay, G. (Editors) 239

4_Saline and Neutral Mine water BOOK.indb 239 9/3/18 11:15 AM 11th ICARD | IMWA | MWD Conference – “Risk to Opportunity”

de ned soil textures (clay, sandy loam, sand worked with sodium bicarbonate rich water and coarse sand) populated in a database. A from a coal bed methane operation (Beletse simple crop factor model is used (with many et al 2008) and showed this particular mine crop species parameters included in the da- water to be problematic for irrigation due to tabase) to estimate water use, and a seasonal, the high levels of salinity and sodium, with root density-weighted pro le salt content is no opportunity to precipitate gypsum and estimated to predict yield reduction due to reduce soil salinity. Beletse also considered salinity using the Maas and Ho mann (1977) sodium sulphate rich waters on heavy clay approach. Soil pro le chemical equilibrium soils, and although pastures did grow with reactions are modelled a er Robbins (1991) this water, the sustainability of the practice in order to predict gypsum precipitation that was questioned (Beletse 2008). reduces root zone salinity. Simulations are For the demonstration of the irrigation run over periods of 10 to 45 years in order to water quality DSS, three mine waters that we quantify the probability and severity of a spe- have experience with for irrigation were con- ci c e ect occurring. Site-speci city is con- sidered (lime treated AMD, sodium bicar- sidered by means of populated databases that bonate rich and sodium sulphate rich waters). allow the user to select an appropriate weath- In addition, a chloride rich water was “gener- er station, soil texture, crop specie, irrigation ated” by keeping the properties of our actual management strategy, and irrigation system. lime treated AMD water, except that sulphate In this paper, the functionality of the DSS and chloride concentrations on a mol charge is demonstrated by simulating the e ect on basis were swopped, to see if the DSS would crop yield of 45 years of irrigation with  ve come to the same conclusions that du Plessis poor quality mine waters.  e role of crop (1983) did. Finally, because we are currently choice, climate and irrigation management interested in investigating whether it is fea- on crop yield are also highlighted. In addi- sible to directly apply AMD to heavily limed tion, nutrient supply to crops and trace ele- soil (to negate the need for a liming plant), we ment addition to the soil pro le with irri- also considered an acidic water of pH 3.  ese gation water are demonstrated. Finally, the waters’ analyses are presented in Table 1. e ect of mine impacted waters on corrosion and scaling of irrigation systems, and on soil Relative crop yields physical properties are presented. Model pre-  e DSS considers a relative crop yield above dictions are discussed in the light of previ- 90% to be ideal, between 80 and 90% is ac- ous experience with irrigation using some of ceptable, between 70 and 80% is tolerable, these waters. and an unacceptable yield is below 70% of potential yield. Of course, in reality, this will Water qualities depend on the grower and the pro tability of Much research on irrigation with mine wa- producing a particular crop, but it is consid- ters has been carried out in South Africa. ered a useful guideline. Crops di er greatly Du Plessis (1983) modelled the advantage in their sensitivity to or tolerance of salin- of sulphate rich waters over chloride domi- ity, but tend to show no yield penalty until a nated waters, suggesting that irrigation with certain threshold salinity is exceeded, where gypsiferous mine waters would be feasible. a er yield decline is linear, with the slope of  is was proved correct by Jovanovic et al this decline depending on the particular crop. (1998), who showed that irrigation of a wide  e simulations presented are for a “Medium range of crops was feasible using lime treated Sensitive” summer crop, maize, with a thresh- acid mine drainage (AMD). Annandale et al old saturation paste salinity of 170 mS/m and (1999 and 2002) investigated the long-term a 12% decrease in yield for each 100 mS/m sustainability of irrigation with gypsiferous increase in salinity, and two “Medium Toler- mine waters, and demonstrated that gypsum ant” crops, soybeans, a summer crop with a precipitation in the soil kept soil solution sa- threshold of 500 mS/m and slope of 20% and linity levels relatively low, thereby facilitating wheat, a winter crop with a threshold of 600 sustainable crop production, without nega- mS/m and slope of 7.1%. tive consequences for soil resources. Beletse

240 Wolkersdorfer, Ch.; Sartz, L.; Weber, A.; Burgess, J.; Tremblay, G. (Editors)

4_Saline and Neutral Mine water BOOK.indb 240 9/3/18 11:15 AM 11th ICARD | IMWA | MWD Conference – “Risk to Opportunity”

Table 1 Composition of mine waters evaluated with the DSS Lime treated Chloride Sodium Sodium sulphate AMD rich water bicarbonate rich rich water AMD water Major constituent

Calcium (mg/L) 615 615 25 32 227 Magnesium (mg/L) 208 208 0 88 132 Sodium (mg/L) 10 10 2000 796 13 Bicarbonate (mg/L) 41 41 5000 450 0 Chloride (mg/L) 5 1500 375 18 3.5 Sulphate (mg/L) 2082 7 7 1647 2919 pH 5.7 5.7 7.5 8.9 3 EC (mS/m) 377 377 750 372 360 TDS (mg/L) 2961 2961 7407 3031 3295 SAR (mol/L)0.5 0.1 0.1 111 16.5 0.2 Trace elements Aluminium (mg/L) 2 158 Iron (mg/L) 31 233 Manganese (mg/L) 28 72 Nutrients Nitrogen (mg/L) 16 19.5 Phosphorus (mg/L) 0.4 0.5 Potassium (mg/L) 6 7

Relative yields were simulated for 45 years of for these simulations was to apply 20 mm ev- irrigation with  ve di erent water qualities ery time the soil pro le had a de cit to  eld using weather data from Loskop Dam, which capacity of 25 mm, thereby leaving room for is reasonably representative of the coal  elds at least 5 mm of rain. Results are presented of Mpumalanga in South Africa.  e long in Table 2. simulation period is used to estimate the risk It is clear that production of a medium of irrigation with these waters, so that rela- sensitive maize crop is only feasible with the tive yields can be presented as the fraction of gypsiferous lime treated AMD water. Soy- time they fall in a particular  tness for use bean is clearly a summer crop that can be category.  e irrigation management practice produced with a wider range of poor qual-

Table 2 Simulated % of time that yields of maize, soybean and wheat at Loskop would fall within speci c tness for use categories. Fitness Relative Lime Chloride Sodium Sodium AMD for crop treated rich bicarbonate sulphate rich Use yield AMD water rich water water

% Maize Maize Maize Maize Maize Wheat Wheat Wheat Wheat Wheat Soybean Soybean Soybean Soybean Soybean

Ideal 90-100 91 100 100 36 82 39 91 100 27 Acceptable 80-90 9 27 12 9 24 9 30 Tolerable 70-80 9 6 9 21 36 9 24 Unacceptable <70 100 27 100 82 15 55 100 100 18

Wolkersdorfer, Ch.; Sartz, L.; Weber, A.; Burgess, J.; Tremblay, G. (Editors) 241

4_Saline and Neutral Mine water BOOK.indb 241 9/3/18 11:15 AM 11th ICARD | IMWA | MWD Conference – “Risk to Opportunity”

ity waters than maize, and wheat is able to be yields are predicted for Vaalharts when e ec- produced even with very poor quality waters. tive leaching is increased from 5.7 to 7.5%. Clearly crop choice is one of the manage- ment practices available when assessing the Plant nutrient supply through mine feasibility of irrigating with a particular water water irrigation quality. Directly irrigating with AMD does Some mine waters are quite high in essential not look very promising, but initial pot trials plant macro-nutrients, like nitrogen, phos- undertaken at the University of Pretoria are phorus and potassium. Irrigation with nu- encouraging when such waters are applied to trient rich waters can be bene cial for crop highly bu ered or limed soils. production, as this may represent a saving in  e e ect of location and irrigation man- the cost of buying fertiliser and the expenses agement are illustrated in Table 3. Warmer, incurred in applying it. However, some crops drier climates will be less suited to irrigation may be negatively a ected by high con- with saline waters than cooler wetter climates, centrations of nutrients, through excessive where atmospheric evaporative demand is vegetative growth and lodging, delayed ma- lower and a higher rainfall induced leaching turity and reduced crop quality. High nutri- environment is encountered. Simulations for ent levels may complicate fertiliser manage- Loskop are compared to those for Vaalharts, ment and limit control over nitrate leaching the biggest irrigation scheme in South Africa, and P wash o .  e rationale adopted in the which has a drier climate, with a far cooler DSS is that the higher the nutrient content winter. It is expected that maize will perform and the greater the supply of nutrients to the better in the more humid environment. In ad- crop, the harder it becomes to manage crop dition, two irrigation management strategies nutrient requirements. However, crops vary are compared.  e “De cit” irrigation strat- greatly in their nutrient requirements, and egy is designed to minimise leaching and use pasture crops in particular, will not easily be rainfall more e ciently.  is is a wise strat- adversely a ected by high nutrient loads, so egy to follow when using good quality water, crop selection is once again important when but irrigation is a salt concentrating practice, irrigating with mine impacted waters. Table 4 as crop roots extract water and exclude salts, illustrates the nutrient supply to soybean and and the higher the salinity, the greater the wheat from our treated and untreated AMD leaching requirement for sustainable irriga- waters. tion. It is therefore not surprising that better

Table 3 Demonstration of the e ect of location (climate) and irrigation management (de cit irrigation vs a salt leaching strategy) on water requirements, e ective leaching and relative yield.

Fitness for Relative Loskop Dam Vaalharts Use crop De cit irrigation Leaching requirement De cit irrigation Leaching requirement yield (%) Maize Wheat Maize Wheat Maize Wheat Maize Wheat Ideal 90-100 100 100 100 100 Acceptable 80-90 9 9 Tolerable 70-80 55 45 27 36 Unacceptable <70 36 45 73 64 Irrigation (mm) 731 557 763 560 978 523 1003 527 E ective (%) 8.9 1.1 11.1 2.2 5.7 1.7 7.5 2.9 Leaching Simulations are for the sodium sulphate rich water on a maize-wheat rotation in the Loskop Dam area and at Vaalharts (drier region with cooler winters).  e De cit irrigation strategy is to irrigate with 15 mm when- ever the de cit to eld capacity reaches 20 mm, thereby leaving some room for rain.  e leaching requirement strategy applies 15% more water than is required to return the pro le to eld capacity.

242 Wolkersdorfer, Ch.; Sartz, L.; Weber, A.; Burgess, J.; Tremblay, G. (Editors)

4_Saline and Neutral Mine water BOOK.indb 242 9/3/18 11:15 AM 11th ICARD | IMWA | MWD Conference – “Risk to Opportunity”

Table 4 Simulated plant nutrient (nitrogen) supply to a wheat-soybean rotation irrigated with lime treated AMD and AMD for Loskop weather data with irrigation leaving 5 mm room for rain once de cit to eld capacity reaches 25 mm. Lime treated AMD AMD Fitness for Contribution Soybean Wheat Soybean Wheat Use to crop N removal Time Applied Time Applied Time (%) Applied Time (%) Applied (%) (kg/ha) (%) (kg/ha) (kg/ha) (kg/ha) Ideal 0-10% Acceptable 10-30% 53 102 7 107 Tolerable 30-50% 47 130 100 88 91 141 55 103 Unacceptable >50% 2 191 45 114

Trace element accumulation in soil Corrosion and scaling of irrigation A concern o en raised with the use of mine equipment and e ect of mine water water for irrigation is the fate of trace ele- on soil physical properties ments in the water. Especially with circum-  e DSS uses the Langelier Index to estimate neutral waters this is usually not of great con- corrosion or scaling of irrigation equipment. cern, but with lower pH waters this should  is may not be the best index for sulphate certainly not be ignored.  e DSS calculates rich waters, as it was developed for carbonate how many years it will take to reach protec- rich waters. However, it can be seen in Table tive threshold soil values in the top 150 mm 6 that the sodium bicarbonate and sodium of the soil pro le. Aluminium, iron and man- sulphate rich waters are predicted to be scal- ganese are identi ed as the elements most ing, with the other waters indicated to be cor- likely to be  agged by regulators. However, rosive, especially the AMD. In addition, the the fact that they are found in abundance in DSS also predicts the e ect of water quality natural soils raises the question whether or on in ltration and hydraulic conductivity. not these guidelines should be relaxed some- Sodium is particularly problematic, but the what. To ascertain the real risk of detrimental negative e ect is somewhat counteracted by food or forage safety due to trace element ac- high salinity levels. cumulation, more research needs to be done.

Table 5 Predicted e ect of irrigation with di erent mine a ected waters on corrosion (negative Langelier Index) or scaling (positive Langelier Index) of irrigation equipment, and the % of time soil physical properties will fall in speci c tness for use categories. Fitness Lime Chloride Sodium Sodium AMD Parameter for Use treated rich bicarbonate sulphate rich AMD water rich water water Corrosion (-) Ideal Scaling (+) Acceptable 0.84 Langelier Index Tolerable -1.62 -1.61 1.34 Unacceptable -6 Ideal 92 92 92 92 92 Surface Acceptable 8 8 8 8 8 In ltration (% of time) Tolerable Unacceptable Ideal 100 89 42 78 98 Soil Acceptable 28 2 Hydraulic Conductivity Tolerable 17 (% of time) Unacceptable 11 13 20 2

Wolkersdorfer, Ch.; Sartz, L.; Weber, A.; Burgess, J.; Tremblay, G. (Editors) 243

4_Saline and Neutral Mine water BOOK.indb 243 9/3/18 11:15 AM 11th ICARD | IMWA | MWD Conference – “Risk to Opportunity”

Conclusions Annandale JG, Jovanovic NZ, Tanner PD, Benadé Not all mine waters are suitable for irrigation. N, du Plessis HM (2002)  e sustainability of ir- However, the user-friendly DSS is able to as- rigation with gypsiferous mine water and impli- sess site-speci c factors that in uence the cations for the mining industry in South Africa. suitability of mine waters for irrigation, and Mine Water and the Environment, 21, 81-90. present the risk taken in using such waters, as Annandale JG, Stirzaker RJ, Singels A, van der far as crop yield and quality, soil factors and Laan M, Laker MC (2011) Irrigation scheduling irrigation equipment is concerned. In some research: South African experiences and future cases, a more detailed Tier 3 assessment is re- prospects. Water SA. 37, 5, Special Issue, 751- quired, to ascertain if negative Tier 2 assess- 763. ment issues, like for trace element loading in Beletse YG, Annandale JG, Steyn JM, Hall I, Aken the simulations presented here, can be miti- ME (2008) Can crops be irrigated with sodium gated.  is may require expert input by crop bicarbonate rich CBM deep aquifer water?  e- or soil scientists.  e DSS will assist regulat- oretical and  eld evaluation. Ecolog. Eng. 33(1), ing authorities to make decisions on permit- 26-36. ting mine water irrigation, and in consider- Beletse YG (2008)  e environmental impact and ation of mine closure options. sustainability of irrigation with coal-mine water. Future research should focus on the as- PhD thesis, University of Pretoria. sessment of food and forage safety when using Du Plessis HM (1983) Using lime treated acid mine waters for irrigation, and improvements mine water for irrigation. Wat. Sci. Tech. 15, to algorithms for scaling and corrosion using 145-154. gypsiferous waters. In addition, the need to Du Plessis HM, Annandale JG, van der Laan M, discriminate against waters high in trace el- Jooste S, du Preez CC, Barnard J, Rodda N, Dab- ements that are abundant in soils should be rowski J, Genthe B, Nell P (2017) Risk based, considered. Finally, an e ort should be made site-speci c, irrigation water quality guidelines. to improve and expand parameters used in Water Research Commission Report No TT the model, especially for e ects like scorch- 727/17, ISBN No 978-1-4312-0910-1 ing of foliage with low pH waters. Jovanovic NZ, Barnard RO, Rethman NFG, An- Acknowledgements nandale JG (1998) Crops can be irrigated with limetreated acid mine drainage. Water SA, The authors thank the Water Research 24(2), 113122. Commission for funding the development Maas EV, Ho man GJ (1977) Crop salt tolerance of the Irrigation Water Quality DSS. – current assessment. J. Irrig. Drain Div. ASCE 103(IR2), 115-134. References Robbins CH (1991) Solute transport and reactions Annandale JG, Jovanovic NZ, Benadé N, Tanner in salt-a ected soils. In: Hanks RJ, Ritchie JT PD (1999) Modelling the longterm e ect of irri- (Eds) Modelling Plant and Soil Systems. Agron- gation with gypsiferous water on soil and water omy Monograph No. 31, ASA-CSSA-SSSA, resources. Agric. Ecosyst. Environ. 76: 109-119. Madison, Wisconsin, USA, p 365-395

244 Wolkersdorfer, Ch.; Sartz, L.; Weber, A.; Burgess, J.; Tremblay, G. (Editors)

4_Saline and Neutral Mine water BOOK.indb 244 9/3/18 11:16 AM