Model-based approach for treated wastewater reuse strategies focusing on water and its nitrogen content ”A case study for olive growing farms in peri-urban areas of Sousse, Tunisia”

:


Introduction
Population growth and its high concentration in urban and peri-urban areas com- 44 bined with climate change induce an increasing pressure on water resources and an im- 45 portant impact on the degradation of water quality [1,2]. It is estimated that about 2.3 46 billion people are in water-stressed countries, of which 733 million people are located in 47 high water-stressed countries [3]. Indeed, the highest stress levels occur in Northern Af-48 rica and in Western, Central and Southern Asia [4]. Due to the gap between water supply 49 and demand and the competition between economic sectors, water scarcity situation is 50 becoming more severe. Agriculture sector is the largest water consumer with more than 51 70% of all withdrawals globally and the water withdrawal ratio for agriculture can reach 52 90% in some arid countries [4]. Indeed, irrigated agriculture is essential to preserve agri- 53 culture productivity, food security, to attenuate the effect of climate and to contribute to 54 the national economy. In this context, international communities are more conscious 55 about water issues and the United Nations has included among the 17 Sustainable Devel-56 opment Goals (SDGs), goal 6 which is dedicated to water and sanitation. Additionally, 57 climate change and anthropogenic activities have significant effects on water availability 58 which may cause decision and policy makers to focus on new strategies for water resource 59 management and water security for sustainable allocation and use [5]. 60 In water-scarce countries and regions, the reusing of wastewater provides a signifi-61 cant opportunity to substitute limited freshwater resources with reclaimed water for spe-62 cific purposes [6] [7]. In addition to efficient water distribution systems and sustainable 63 agriculture, reuse of wastewater is a relevant action in reducing water stress [4]. Certainly, 64 wastewater is a possibly inexpensive and sustainable source of water, energy, nutrients, 65 organic matter and other useful by-products [8]. However, several barriers such as public 66 perception, pricing, technical and regulation are affecting the possibility of implementing 67 efficient water reuse strategies [9]. Like many countries in the MENA regions,Tunisia also 68 suffers from the problem of water shortage. Annual water resource potential is estimated 69 at 4898 million m 3 with about 2700 million m 3 is surface water and 2198 m 3 is groundwater 70 [10] [24]. In fact, the total renewable water resource per capita is estimated at 420 m 3 /in- 71 habitant/year which is considered a Key indicator of water scarcity [10] . Freshwater is not 72 used only for domestic purposes such as drinking but also for economic activities such as 73 agriculture or Industry. Furthermore, the agriculture sector has a great importance due to 74 its social impact. Agriculture is the first user of water compared to the other sectors, ac-75 counting for about 79% of freshwater [11]. However, water drinking is estimated at 15%, 76 Industry at 3% and tourism sector at 1% and other use at 2% [11]. In addition, population 77 growth and rapid development of the economic sectors have increased the problem of 78 water scarcity in Tunisia. Therefore, the government is facing a major challenge that deals 79 with preserving and protecting this scarce resource to fit the supply and demand of water. 80 For this reason, the government adopted several strategies to protect it and to maintain 81 balance between water demand and supply. The main strategies can be summarized as (i) 82 water surface mobilization through appropriate infrastructures such as dam ; (ii) support 83 farmers to adopt techniques of water saving with incentive allowance ; (iii) implementa-84 tion of appropriate legislation and institutional systems for water resource management ; 85 (iv) promote non-conventional water use in agriculture such as treated wastewater reuse 86 or brackish water desalination (v) improvement the involvement of local people in the 87 strategy through the etablishement of local water user association. Additionally, the Tu-88 nisian government developed two key strategies for the year 2050 related to water re-89 source management "Water 2050"and reuse "WATER REUSE 2050". Both strategies focus 90 on developing appropriate action plans to support and guide decision makers and water 91 managers. The Water 2050 strategy included several recommendations for water re-92 sources management based on forecasting models of supply and demand. These recom-93 mendations ar primerly related to water, infrastructure, governance, economy, and ecol-94 ogy [12]. However, the WATER REUSE 2050 focuses on reuse as an alternative solution 95 to conserve freshwater. The goal of this strategy is to implement a sustainable action plan 96 in terms of assessment, technologies, regulation, financing of treated wastewater [13] 97 The reuse framework in Tunisia started by launching research programs, the con-98 struction of several WasteWater Treatment Plants (WWTP) with advanced technologies 99 and appropriate sanitation systems, the involvement of several actors and the adoption of 100 several standards and guidelines for safe use [14]. Despite the efforts provided by the Tu-101 nisian government, the reuse rate is still low compared to potential treatment of 102 wastewater. Among 122 WWTP, only 61 treated plants are designed for reuse. In 2019, 103 about 284 million m3 were generated but only 13,4 million m3 are recycled for agriculture 104 purposes [15]. 105 From a circular economy perspective, recycling and reuse are the central concern and 106 water supply can be improved through better wastewater management strategies [16]. 107 Conversely, risks associated with water quality and human health must also be taken into 108 account [16]. Furthermore, it was estimated that 80% of all industrial and municipal 109 wastewater are rejected to the environment without treatment affecting overall water 110 quality, leading to negative impacts on human health and ecosystems [8]. Therefore, the 111 focus on appropriate technologies for an efficient water treatment is important to deter-112 mine reuse purpose [9]. Wastewater treatment is based on a combination of physical, 113 chemical, and biological processes to eliminate wastewater components [8]. Several tech-114 niques and methods for wastewater treatment are applied. Indeed, conventional methods 115 for removing metals are becoming inappropriate to meet rigorous permissible effluent 116 standards for an intended use [17]. Additionally, the implementation of advanced tech-117 niques as a tertiary treatment process may lead to good water quality for supplying irri-118 gation or domestic uses [18]. For example, Kalboussi et al. [19] conducted a life cycle as-119 sessment study to evaluate the environmental efficiency of water reclamation for agricul-120 tural irrigation among other conventional options. They found that the environmental im-121 pact of reclaimed water depends directly on the type of tertiary treatment technology and 122 the location of the treatment plant in relation to the field and other water sources. Natural 123 landscapes such as forests and wetlands have an important contribution in improving 124 water quality by decreasing sediment loadings, capturing and holding pollutants and re-125 cycling nutrients [8]. Nature-Based Solution (NBS) creates opportunities as an innovative 126 solution to improve ecosystem services, boost resilience and livelihood in water planning 127 and management [8]. 128 Treatment may improve the quality of treated wastewater to meet standards, but, 129 should also preserve nutrients. As wastewater is rich in nitrogen and phosphorus which 130 can provide nutrients to crops, the serious challenge for reuse agricultural irrigation is not 131 only to preserve quantities of nitrogen and phosphorus contained in the wastewater, be-132 cause these nutrients are essential for plant growth [20] but also to respect appropriate 133 guidelines for safe use [21] , [22] . In order to implement sustainable and effective reuse 134 strategies, a good knowledge of soil-plant-water interactions is required. In this context, 135 crop models have been developed by several teams and have led to several software such 136 as AquaCrop [23] , STICS [24], OPTIRRIG [25]among other ones. The simulations pro-137 vided by these models serve as predictive and decision support tools for agricultural prac-138 tices. More complex and comprehensive models have been developed as Global Change 139 decision support system DANUBIA [26]. For the processing, DANUBIA crop growth 140 model needs several data such as meteorological date, site-specific information, soil char-141 acteristics and farming practices. Additionally, the Nitrogen cycle was also integrated in 142 DANUBIA model to determine nitrogen turnover, nitrogen fluxes and storages [27]. These 143 approaches are based on relatively complex models with many variables and parameters, 144 which provide quite precise descriptions of the state of the soil-crop-climate system, but 145 are also quite heavy to conduct intensive optimization over a tactic time horizon [23,24]. 146 Other approaches are based on much simpler models (i.e. reduced models) that do not 147 intend to give a precise description of the internal functioning of the soil-crop system, but 148 rather focus on flux balance, and can therefore predict soil composition, water consump-149 tion, and biomass production at the field scale only [28,29,30,31,32]. This kind of models 150 is thus better suited to apply optimization tools, because of their relatively small size. 151 Moreover, the manipulated variables that typically describe irrigation and fertilization, 152 and measurements such as soil humidity and crop water demand are usually considered 153 at the field scale by practitioners. These reduced models can be validated on the more 154 sophisticated models, which can also provide parameters sensitivity [28]. In this context, 155 Pelak et al. [29] focused on the relationship between canopy cover, soil moisture and soil 156 nitrogen content to optimize strategies of fertilization and irrigation. Moreover, Kalboussi 157 et al.
[30], [31] proposed a generic crop model named "TOYCROP" which is the basic ver-158 sion of the more advanced model "OPTIRRIG model". TOYCROP was developed to de-159 termine optimal irrigation and nitrogen management via treated wastewater [31], [32]. 160 Considering water scarcity in Tunisia, this research focuses on promoting reuse as an 161 alternative solution to water saving and implementing of crop models for a sustainable 162 reuse scheme. A feature of this study is the development of a model based on the combi-163 nation of treated wastewater and nitrogen as nutrient for olive production. Therefore, the 164 main objectives of this study are (i) to analyze the value chain of treated wastewater for 165 olive growing farm and (ii) to apply a mathematical model considering water and nitro-166 gen content in order to maximize olive yield in the treated wastewater (TWW) irrigated 167 perimeter of Msaken, Sousse (Tunisia). This research may be useful for local decision mak-168 ers to provide appropriate guidance and recommendations for fertigation scheduling. 169 The next section presents the research framework and the description of the study 170 area, as well as the approach used to characterize the optimal irrigation and nitrogen strat-171 egy. The main results related to the value chain and modeling are proposed and discussed 172 in Section 3. Finally, section 4 summarizes the main outcomes of this research. In this study, we focused on the reuse of treated wastewater in the irrigated perimeter 177 in Msaken, Sousse. Figure 1 illustrates the main components of the approach applied. Spe-178 cific parameters and datasets were used to implement the wastewater value chain and 179 detect the interaction between irrigation and nitrogen based on a crop model analysis. The irrigated perimeter of Msaken from Sousse governorate is selected as a study 219 area. Sousse is characterized by water stress and overexploited and saline groundwater. 220 In order to provide safe water for users and to ensure water security, the local decision 221 makers adopted a strategy based on a transfer of water from neighboring governorates 222 (Zaghouan and Kairouan) [33]. It was estimated in 2021 that about 65% of distributed wa-223 ter resources is from internal resources including water surface, groundwater and treated 224 wastewater [33]. However, about 35% are external water resources. Among the internal 225 resources distributed, only 7% are coming from treated wastewater [33]. For this reason, 226 the reuse for agriculture purposes can be a way of mitigating water shortage problems in 227 Sousse. This irrigated perimeter of Msaken is located between 10°36"-10°38" N latitude 228 and 35°45"-35°43"E longitude (figure 2). This perimeter was implemented in 2002 and it 229 was developed to reuse treated wastewater. This region is characterized by a semi-arid 230 climate with mild winter. Average annual rainfall is about 319 mm. The average monthly 231 maximum temperature is around 35°C in July and the lowest monthly average tempera-232 ture is around 6°C in January The concept of value chain is defined as all activities required to take a product from 248 the initial input-supply phase, through numerous stages of production, to its final market 249 destination [35]. In addition, value chain analysis is a process of breaking a chain into its 250 component parts to understand its structure and operation in detail [35] . In the case of 251 treated wastewater reuse, value chain was required to (i) identify the main actors involved 252 in the process from the wastewater collection to reuse; (ii) describe the main components 253 of the wastewater treatment system; (iii) monitor the water quality and quantity used; (iv) 254 identify local farmers' perceptions of treated wastewater. The value chain analysis was 255 useful in providing a SWOT (Strengths, Weaknesses, Opportunities and Threats) analysis 256 for reuse in our study area. Moreover, SWOT analysis was conducted to identify the main 257 gap of reuse.
The function is used to capture the plant stomatal response to soil moisture con-279 dition, as given by equation 4 [30]: where is the wilting point and * is the point at which stomata closure starts. 283 A similar function, , is used to module evaporation, depending on the hygroscopic 284 point of soil, ℎ , below which no soil moisture losses occur (equation 5). 285 A diagram of and as a function of S is shown in Figure 3 [30]:. Where K t is a climatic coefficient, T m is the mean monthly temperature (°C) and p is the mean daily 297 percentage of annual daytime hours. The transpiration of olive is assumed to be proportional to the crop 298 radiation interception efficiency (t), which is a function between 0 and 1 that reflects the plant cover, 299 as mentioned in equation 7.
The soil evaporation is considered proportional to the uncovered part of the soil (1-( )).  The function of nitrogen uptake limitation is given by equation 11 [29].
The nitrogen leaching is proportional to the water percolation, , and the nitro-327 gen concentration / as indicated in equation 12 [29]. Where * is the normalized daily water productivity and olive transpiration is given 335 by equation 14.
2.3.3 The sensitivity analysis 338 All mathematical models are approximate and their usefulness depends on the under-339 standing of the uncertainty associated with the predictions [36]. Uncertainty can affect 340 the accuracy of the results at every calculation stage [36]. Sensitivity analysis can deter-341 mine how variability in inputs leads to variability in outputs. In other words, it is an 342 approach to determine which parameters have the most or least impact on the output 343 solution. It quantifies the ratio of output disturbances to input disturbances.  Schlumberger array was adopted. Based on this approach and considering the value of resistivity 469 (localization of a prominent elongated low resistivity beneath the tree), it was detected that the 470 average root depth in the selected olive growing farm is about 0,8 m. Figure 5 shows the ERT 471 profile related to the analysis. 3. Figure 5. ERT profile to assess root depth Results and discussion 478 In this section, the results of value chain analysis were presented and the SWOT was also elaborated. 479 Then, the optimal irrigation and nitrogen plan is developed through the mathematical simulation. The reuse was assessed based on water volume and water quality.  Figure 7 shows the variation of the volume of effluent consumed by the farmers. We found that the 511 amount of treated wastewater that farmers consume varies. The annual average from 2012 and 2020 is 512 185 277 m3. The highest usage was in 2016 and the lowest is in 2020. The volume in 2020 was estimated 513 to be 69650 m3. This volume is only about 2% of overall treated wastewater provided by WWTP of 514 Msaken. Based on the survey and discussions with local users and managers, we found that farmers 515 mainly used TWW to irrigate their olive growing farms, and the main cultivar is Chemlali olive (Olea 516 europaea L.). The irrigation scheme depends on rainfall and the amount of TWW provided by ONAS. 517 In fact, olive trees can grow in difficult climatic conditions and with poor water quality [47]. Addition-518 ally,olive trees are an alternate bearing species which is characterized by low-yield "off-year" followed 519 by a high-yield "on-year" [48]. This situation can explain the fluctuations in water consumption in 2014, 520 2019 and 2020. However, water usage in 2016 was exceptional. This is because in the past, several local 521 farmers were dairy producers and they irrigated their land to grow pastures for their livestock. How-522 ever, due to various reasons such as livestock insecurity and declining subsidies for seeds, many farm-523 ers stopped this activity and focused solely in irrigating their olive trees in appropriate period. In fact, 524 supplemental irrigation of the Chemlali olive cultivar helps ensure and maintain olive yields [49]. As it 525 was presented in figure 8 ((a);(b) The treated wastewater used on this irrigated perimeter was monitored for the year 2020 to evaluate the 537 efficiency of the treatment plant of Msaken and its appropriateness for use on agricultural irrigation. In 538 this study, we focused on the 4 important parameters (Conductivity, Total Suspended Solids (TSS), BOD5 539 and COD). It was observed that the concentrations of these parameters were high before treatment and 540 they reduced after treatment as described in Figure 9. The conductivity is varied from 2830 to 3365 µS/cm 541 before treatment and after treatment, the conductivity of treated wastewater is from 2839 to 3104 µS/cm. 542 The removal efficiency of conductivity is 4%. For the case of TSS, the amount is varied from 274 to 579 543 mg/L before treatment but the concentration decreases after treatment and it is ranged from 16 to 27 544 mg/L. The removal efficiency of TSS is 95%. However, BOD5 before treatment is varied from 344 to 660 545 mg/L. After treatment, the value of BOD5 is from 9 to 30 mg/L. The performance of the BOD5 treatment 546 is high, about 96%. In addition, the amount of COD before treatment varies from 414 to 1377 mg/L. This 547 value has decreased drastically: COD ranges from 50 to 86 mg/L after treatment. The removal efficiency 548 is about 93%. The high efficiency of the treated plant is in accordance with the Tunisian standard (NT 549 106-03) for reuse in agriculture, as shown in Table 1. We can confirm that the quality of treated 550 wastewater is suitable for irrigation and does not pose any risk to human health. However, the salinity 551 of the soil must be monitored because the conductivity value is high. In this context, the impact of treated 552 wastewater on soil properties is also important.  Nitrogen and phosphorus are two important and basic natural components for the growth of living 560 organisms [50]. Additionally, in the case of soil nutrients deficiency, the use of synthetic fertilizers is a 561 significant factor in securing and increasing global food production [51]. Moreover, the high potentials 562 of the recuperated nutrient for reuse as fertilizer in agriculture is recognized [52]. In our research, we 563 focus on the variation of Total Nitrogen (N) and Phosphorus (P2O5)in soil as a function of irrigation 564 with TWW. The results presented in Table 2 show that the nutrients concentration increased under the 565 effect of irrigation. It is found that the total Nitrogen content at 0-20 cm depth in plot 1 (without irriga-566 tion) is 440 mg/kg, while in the most irrigated plot it was 1120 mg/kg. The same tendency is observed 567 for phosphorus. The highest concentration was obtained in plot 3 compared to the amounts in plots 2 568 and 1. Similar results were found in the research work of Hidri et al. [53].  The main outcomes of the SWOT analysis are presented in Table 3. The SWOT analysis pointed out 578 several advantages and barriers to the development of the reuse in the irrigated perimeter of Msaken. 579 Indeed, the application of reuse based on mathematical simulation can lead to ensuring sustainable 580 irrigation and fertilization schedules.   Figure 10 shows the results of the model compared to the data used. The results show a good 602 fit of the model to the field data. In fact, the olive biomass production simulated by the model 603 is quite similar to the field production. In addition, the soil nitrogen content determined in the 604 laboratory is very close to the model results.

Soil Humidity simulation
The model is also applied to simulate soil moisture. Namely, this parameter is estimated based 617 on water inputs from rainfall and irrigation, and losses from evaporation, transpiration, and 618 deep percolation. The variation of soil humidity over time is illustrated in Figure 11. Due to 619 the increase of drought period and rainfall deficit, the selected farmer applied two intensive 620 rounds of irrigation with 350 m3 of TWW per day per hectare: 12 days for the first round and 621 13 days for the second one. According to the simulation results, the volumes brought by each 622 irrigation overflow the soil and may induce to deep percolation of water. Soil evaporation and 623 crop transpiration simulated by the model are shown in Figure 12.b. In this context, data on 624 the reference evapotranspiration ET0 and the radiation interception efficiency of olive crop 625 (that is the total ( )) and the one specific to olive production ( )) are required (cf. 626 The nitrogen content of the soil was also increased by fertigation (cf. Figure 10.b From 100 iterations, the results shown in figure 13 reveal that the biomass is sensitive to the normalized 655 daily water productivity (W*). However, the nitrogen response is sensitive to both parameters of the 656 nitrogen leaching function ( and ). On the other hand, a perturbation of the initial soil N concentra-657 tion or of the plant N uptake limit does not affect the model output. For the analysis of viability, we found that nitrogen stress is always largely overcome by water stress, 667 which is then the only limiting factor for biomass production. Figure 14 shows the domain K of maximum 668 biomass production in grey and two boundaries: the red boundary for nitrogen stress and the blue for 669 hydric stress. The finding indicates that the magenta curve representing the separation between the hydric 670 and nitrogen stresses, does not belong to the domain K. Therefore, any trajectory of the system depicted in 671 the (S,N) plane that starts in the domain K will only touch the S=S* boundary of the hydric stress and never 672 the S/N= ηc boundary of the nitrogen stress. Moreover, due to the fertigation, we can observe that the 673 trajectory which remains on the S=S* boundary goes upward. This means that the amount of nitrogen in-674 creases with irrigation, which implies that irrigation with TWW can stay away from the nitrogen stress. 675 Wastewater reuse is a sustainable solution for water resource management to cope 732 with water scarcity. In addition, treated wastewater is also considered as a fertilizer source 733 that can provide necessary inputs for plant growth. This study investigated local 734 farmers'perceptions of the use of treated wastewater in agriculture in irrigated perimeter 735 of Msaken. We also applied a crop model and mathematical simulations to identify 736 optimal and safe conditions for wastewater reuse. 737 Results related to the reuse value chain show that the quality of TWW is suitable for 738 irrigation and does not have a risk to human health. However, local farmers only focus on 739 supplemental irrigation to ensure the olive production. Moreorer, the viability analysis 740 indicates that nitrogen is not a limiting factor for olive production in the Msaken irrigated 741 area. In addition, in this study, we have identified a theoretical minimum irrigation scheme 742 that could guarantee maximum olive production, taking into account soil and water reuse 743 characteristics. We found that maximum irrigation is 5.77 m 3 /day/ha and the total water 744 required per year is 1240 m 3 /ha. 745 Further viability analysis can be elaborated in the future studies to investigate the 746 minimal total quantity of water or total supplied nitrogen to ensure a given biomass 747 production and to consider phosporus needs. 748 The next step of the research is to calculate in real conditions optimal trajectories that 749 maximize olive production (both in terms of water and nutrient content), taking into ac-750 count weather effects. It could even be interesting to intervene in the reuse chain (water 751 treatment system) and act at that level to irrigate olive trees with optimal quality water. 752 In other words, it would be possible to treat water so that it exactly meets the needs of 753 olive trees. 754 Finally, this study could be improved with additional experimental data. For exam-755 ple, the use of appropriate sensors could provide more accurate estimation of soil mois-756 ture and thus, model calibration and prediction.