Impact of the shift from groundwater to surface water irrigation on aquifer dynamics and hydrochemistry in a semi-arid region in the south of Portugal

Impact of the shift from groundwater to surface water irrigation on aquifer dynamics and hydrochemistry in a semi-arid region in the south of Portugal

agricultural water management 85 (2006) 121–132 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/agwat Impact of the sh...

1MB Sizes 0 Downloads 3 Views

agricultural water management 85 (2006) 121–132

available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/agwat

Impact of the shift from groundwater to surface water irrigation on aquifer dynamics and hydrochemistry in a semi-arid region in the south of Portugal T.Y. Stigter a,*, A.M.M. Carvalho Dill a, L. Ribeiro b, E. Reis c a

CVRM/FCMA, Universidade do Algarve, Campus de Gambelas, 8005-139 Faro, Portugal CVRM, Instituto Superior Te´cnico, Av. Rovisco Pais, 1049-001 Lisbon, Portugal c CCDR-Algarve, Rua Dr. Jose´ Matos 13, 8000-503 Faro, Portugal b

article info

abstract

Article history:

In semi-arid regions, the shift from locally extracted groundwater to regionally supplied

Accepted 2 April 2006

surface water for irrigation can have a large impact on the aquifer dynamics and hydro-

Published on line 23 May 2006

chemistry, as is shown for an area of intensive citrus culture in the south of Portugal. Studying the changes that occur is important, particularly in the light of European and

Keywords:

national policies that seek to preserve the quality of groundwater and the involved eco-

Irrigation

systems. In September 2005, the application of Nitrates Directive 91/676/CEE to the study

Groundwater

area led to its designation as a vulnerable zone and the obligation of implementing

Surface water

measures that reduce the risk of nitrate leaching to groundwater. In addition, the area is

Nitrates Directive

bordered to the south by the internationally recognized ecosystem of the Ria Formosa

Salinisation

lagoon, which requires a careful management of water quantity and quality. This article

Dilution

shows that the substitution of the irrigation source indirectly caused the reduction of the nitrogen load on soil and groundwater. The study is based on a large number of groundwater head and quality data gathered over the past decade. It is observed that irrigation with surface water triggered freshening of the upper aquifer, inverting a long-existing trend of increasing groundwater salinities caused by irrigation with local groundwater. The occurrence of both phenomena in the same aquifer is rather unique and is simulated with the hydrogeochemical software PHREEQC. In terms of aquifer dynamics, a sharp rise of the water table is observed, together with an almost complete attenuation of its interseasonal oscillations. A further rise, which is to be expected in years with average or higher rainfall, could cause the water table to enter the root zone of the citrus trees in some areas, with potentially negative consequences. Therefore, an integrated mixed-source irrigation system using both surface and groundwater is highly recommended. # 2006 Elsevier B.V. All rights reserved.

1.

Introduction

It is a well-known fact that certain agricultural practices, such as irrigation and the application of fertilisers, can affect the quality of groundwater, through the leaching of nutrients and

salinisation, the occurrence of which continues to be an important topic of publications worldwide in 2005 (e.g. Chen et al., 2005; Jalali, 2005; Liu et al., 2005; Oenema et al., 2005; Rajmohan and Elango, 2005; Wolf et al., 2005). In Europe, Directive 91/676/EEC, typically labelled the Nitrates Directive,

* Corresponding author. Tel.: +351 289 800995; fax: +351 289 818353. E-mail address: [email protected] (T.Y. Stigter). 0378-3774/$ – see front matter # 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.agwat.2006.04.004

122

agricultural water management 85 (2006) 121–132

was set up to protect waters against pollution caused by nitrates from agricultural sources. More than a decade after the implementation the problems continue to exist throughout Europe (e.g. EC, 2002; EEA, 2003), partly due to the sluggish implementation of Directive. According to Goodchild (1998), the designation of so-called nitrate vulnerable zones and the application of action programmes are strongly hindered by the agricultural lobby, which fears that the production and consequently the financial situation of many farmers will be seriously affected. Moreover, the lack of co-operation between ministries of agriculture and environment is a serious obstacle on the way to a swift implementation of the Directive. Disagreements are worsened by the complex nature of the N cycle and the various human activities that affect it. Possibly one of the most effective measures to be included by EU Member States in their action programmes is by limiting the (organic and mineral) nitrogen fertiliser application on the basis of a balance between crop requirements and supply from the soil and irrigation water. According to the EC (2002), several country reports mention that crop yields remained the same despite the lowering of N inputs, due to the positive impact of improved practices. In a three-year experiment performed in an area near Madrid, Spain, Diez et al. (1997) show that different irrigation schedules and fertiliser types (including a unfertilised control plot) do not affect maize grain yields. Cameira et al. (2003) conclude from their experiments in an alluvial soil under irrigated maize in the Sorraia Watershed in Portugal, that organic nitrogen mineralisation is significant yet unaccounted for during fertilisation practices. In a case study in the region of Valencia (Spain), Paz and Ramos (2004) find that nitrogen fertilisation rates for citriculture can be reduced 50% to a rate similar to that recommended by the Code of Good Agricultural Practice in the area, without affecting crop yields. Oenema et al. (2004, 2005) and Wolf et al. (2005) study the effects of lowering nitrogen (N) and phosphorus (P) surpluses in agriculture on the quality of groundwater and surface water in the Netherlands. They conclude that lowering these surpluses via the implementation of the Dutch mineral accounting system MINAS at farm level is effective in reducing nutrient load on the soil and nitrate leaching to groundwater, based on nutrient emission models. The present article aims to demonstrate the consequences of the reduction of the nitrogen load on soil and groundwater in a semi-arid region in the south of Portugal, where in 2001 groundwater was replaced by surface water for the irrigation of citrus orchards. Until then water had been extracted by the farmers from private wells, which led to an increase in nitrate and salinity levels and triggered several hydrochemical processes. Moreover, there was a strong tendency to apply fertilisers in excess to crop requirements, as the high levels of nitrogen already present in the groundwater, which had the potential of supplying more than 50% of the nitrogen recommended for citrus orchards, were never taken into account. From 2001 onwards, farmers started using surface water supplied by a regional irrigation district, which contained no dissolved nitrogen, therefore causing an accidental improvement of fertiliser application practices. The use of surface water for irrigation triggered a groundwater freshening mechanism, therefore inverting the trend of

increasing salinities. The sequence of groundwater salinisation and freshening in the same aquifer as a result of changes in agricultural practices is rather unique and was not found documented in the literature. The shift towards surface water for irrigation also affected the groundwater heads, which have been closely monitored in the past decade. Therefore, the case study provides a unique possibility to analyse the response of both aquifer dynamics and hydrochemistry to the evolution of agricultural activity and the changes that occurred. The regional importance of the study is mainly twofold: first, groundwater is still an important source for domestic water supply in many of the rural households; second, the study area is bordered in the south by a shallow lagoon with an area of 100 km2 that forms an extremely valuable and sensitive ecosystem, recognised at both a European and an international level (Newton et al., 2003).

2.

Study area

The study area, named Campina da Luz, is located the Algarve, the southernmost province of Portugal, indicated in Fig. 1. It measures approximately 4 km  6 km, and is bordered by the Ria Formosa lagoon (Atlantic Ocean) in the south and nonperennial streams in the east and west. Characterised by a warm Mediterranean climate, mean annual air temperature and precipitation measured at Faro airport are 17.3 8C and 531 mm, respectively. Average rainfall measured at the rain station in Quelfes (Fig. 1) between 1981 and 2004 is 602 mm, but the hydrological year of 2004/2005 has been extremely dry, with 211 mm of rainfall registered until August, largely affecting the urban and agricultural water supply sectors. Real annual evapotranspiration losses are estimated around 80% of the precipitation (Silva, 1984; De Bruin, 1999; Stigter and Carvalho Dill, 2001). According to the national inventory of aquifer systems in Continental Portugal (Almeida et al., 2000), the largest part of the study area is underlain by the aquifer system of LuzTavira. Silva (1984) and Bonte (1999) provide detailed descriptions of the hydrogeology of the study area. A hydrogeological map is presented in Fig. 2. The oldest sediments that crop out in the area were deposited in the Jurassic, subdivided into four units (Bonte, 1999). The first two and the last units constitute the main aquifers of the region and consist of limestones and dolomites, exceeding 500 m in thickness. The third unit is mainly made up of marls with a thickness of 450 m and forms an aquitard. A second aquitard is largely made up of Cretaceous sandy marls that crop out in the southwest, uplifted by the NW–SE trending faults. Miocene sediments cover the Cretaceous and Jurassic units in the centre of area, where they form an upper aquifer of 75 m estimated thickness. Although rather heterogeneous, the main lithology is formed by sandy limestones. Sands and gravels of PlioQuaternary age have local outcrops in the centre and along the coast, but their thickness does not exceed 20 m. Holocene silts and clays have been deposited in river valleys. Recharge of the aquifers occurs in the north and centre and the general direction of groundwater flow is N–S, as is illustrated by the water table contours plotted in Fig. 2. The steeper hydraulic gradients in the south are associated to the

agricultural water management 85 (2006) 121–132

123

Fig. 1 – Location of the study area; also shown are the irrigation district, the water reservoirs that supply it and the nitrate vulnerable zone (NVZ) of Faro, designated in 1997. The new NVZ designated in September 2005 covers a large part of blocks D4.2 and D4.3 of the irrigation district.

presence of faults and the uplifting of the Cretaceous marls. These faults form the southern limit of the aquifer system of Luz-Tavira (Almeida et al., 2000). Agriculture has been a major form of land use in the area for centuries. Initially dominated by almond, fig, olive and carob trees, agriculture became more intensive with the introduction of irrigation at the end of the 19th century. Nowadays, citriculture is the dominant land use in the centre, whereas towards the south the presence of viticulture becomes significant, as illustrated in Fig. 3. Average values of irrigation water requirements are 770 mm/year for citriculture and 650 mm/year for viticulture, according to the Instituto de Desenvolvimento Rural e Hidra´ulica (IDRHa). Groundwater was the single source for irrigation until 2001, at which point the area was included in the irrigation district that supplies the Eastern Algarve using water reservoirs of dams built in two principal tributaries of the Guadiana river, which forms the border with Spain (Fig. 1). These reservoirs have also been used for urban water supply since 1999, resulting in the shutdown of a large number of municipal extraction wells. The total area of the irrigation district is 8621 ha and the average irrigation volume is estimated at 57 hm3/year (Mota and Dias, 2000). Until 2003 only three nitrate vulnerable zones (NVZs) had been designated on the Portuguese mainland in compliance with the Nitrates Directive, one of which is located west of the study area (NVZ of Faro, see Fig. 1). In the past 3 years five new

NVZs were designated, one of which covers the study area and whose designation occurred in September 2005 after a longlasting period of arduous debate. Although still not satisfactory in the eyes of the European Commission, the country has undeniably made progress in this field, which can partly be attributed to a number of studies that pointed out the need and suggested the way to revise and further develop the NVZ designation process in Portugal (e.g. Ribeiro et al., 2002a,b).

3.

Methods

The groundwater heads in the Algarve are monitored on a monthly basis in a well network maintained by the Regional Environmental Directorate. Those wells located within or near the study area are mapped in Fig. 3, which also shows the location of the former urban water supply wells. For the purpose of this study, groundwater head time series were analysed for an interval of 10 years, to detect interseasonal and interannual oscillations and their response to the (urban supply and irrigation) source substitutions. Another monitoring programme, based on a much denser network of observation points, was set up to study the hydraulic behaviour of the upper aquifers between September 1997 and August 1999 (Stigter and Carvalho Dill, 2001). The water table contours in Fig. 2 were created on the basis of this network. Measurements were repeated in October 2003, in

124

agricultural water management 85 (2006) 121–132

Fig. 2 – Hydrogeological map of the study area (geology adapted from Bonte, 1999).

order to compare water table depths before and after the start of surface water irrigation. The contamination and salinisation of the upper aquifer induced by agricultural practices, was studied on the basis of hydrogeochemical data gathered in two sampling campaigns, namely January 1998 and March 1999. For each well, measurements were of groundwater level, pH, electrical conductivity (EC), temperature (T), nitrate (semi-quantitative) and alkalinity (titrimetric) were performed in the field. Normal and acidified groundwater samples were gathered in plastic bottles for a detailed analysis of major anions and cations, respectively, at the laboratory of the Vrije Universiteit Amster-

dam (Bonte, 1999; De Bruin, 1999; Stigter and Carvalho Dill, 2001). In June 2003 10 wells, located in Fig. 3, were resampled with the aim of studying the potential dilution effect caused by surface water irrigation. A spring was also sampled. Following the same sampling procedure as before, the only difference was that it was not possible to measure T in the field. The analyses were performed by high-pressure liquid chromatography at the Universidade do Algarve in Faro, in collaboration with the CVRM/Geo-Systems Centre of the Instituto Superior Te´cnico (IST). The software PHREEQC (Parkhurst and Appelo, 1999) is used to simulate and confirm some of the hydrogeochemical

125

1.39 2.01 1.34 1.88 1.84 1.74 1.59 1.40 1.50 1.56 2.36 1.68 1.90 2.03 2.05 1.94 1.85 1.95 2.32 1.92 1.81 1.98 0.69 0.63 0.61 0.81 0.11 0.05 0.06 0.30 0.05 0.45 1.40 0.07 0.53 0.02 0.59 0.30 0.16 0.13 0.05 0.10 0.02 0.92 75 23 80 15 13 16 17 20 21 25 0* T2 T8 T9 T15 T20 T22 T24 T25 T39 T81 T49

Solutes in mmol/l, EC in mS/cm; *spring, Cm = cretaceous marls, Ml = miocene sandy limestones, SI = saturation Index.

0.69 0.46 0.52 0.41 0.50 0.70 0.36 0.46 0.34 0.44 0.49 0.98 1.44 0.92 1.12 1.10 0.79 1.50 0.99 2.03 1.73 1.22 1.31 0.92 0.79 0.54 1.09 1.32 1.31 0.58 1.13 1.40 0.76 0.54 0.40 0.25 0.26 0.34 0.41 0.42 0.14 0.44 0.52 0.31 5.64 6.42 6.50 7.80 5.90 5.24 7.12 6.76 7.54 6.20 6.24 5.50 4.56 2.34 2.80 3.38 7.72 4.71 1.78 3.31 4.12 3.33 1.10 1.13 0.90 1.40 0.95 1.57 1.59 1.20 1.41 0.87 1.05 3.92 2.98 3.41 3.20 2.96 3.64 2.53 3.08 2.57 2.70 3.09 0.07 0.01 0.04 0.10 0.23 0.44 0.02 0.03 0.05 0.03 0.08 5.40 6.57 2.15 3.13 3.72 6.06 7.08 1.77 6.71 7.15 4.06 1406 1296 1035 1165 1178 1580 1388 941 1321 1357 1166 6.85 7.52 6.87 7.47 7.32 7.16 7.15 6.95 7.09 7.06 7.87

Ca2+ K+ Na+ EC pH Aquif.

The groundwater head time series of the observation wells are displayed for the past decade in Fig. 4, together with the monthly precipitation recorded at the coastal rain station of Quelfes (Fig. 3). Three wells (143, 471 and T24) are located in the upper Miocene sandy limestone aquifer in the centre of the study area and show identical behaviour with respect to water level oscillations. Well 375, which has the largest interannual and interseasonal amplitudes, is situated in the Jurassic limestones in the north (the groundwater head rose almost 60 m in the winter of 1995/1996). Up to the year 2000, groundwater head oscillations follow the seasonal variations in rainfall. The substitution of groundwater by surface water for urban water supply in 1999 apparently has no visible consequences. Groundwater

Depth

Results

Table 1 – Main results of groundwater analyses in 2003

4.

No.

processes that occur in the groundwater under natural conditions and as a result of the agricultural practices. PHREEQC performs aqueous geochemical calculations and modelling, including speciation, reaction-path, advective transport and inverse geochemical calculations.

Mg2+

Cl

HCO3

SO42

Fig. 3 – Spatial distribution of main crop types in the study area and location of the wells for former public extraction, groundwater head observation and groundwater sampling for chemical analysis in 2003; the location of the rain station of Quelfes, shifted 8 km east- and 4 km northwards, is also shown.

Cm Ml Ml Ml Cm Cm Ml Ml Ml Ml Ml

NO3

Na+/Cl

Ca2+/HCO3

SI Calcite

SI Dolom.

SI Gyps.

Log (PCO2)

agricultural water management 85 (2006) 121–132

126

agricultural water management 85 (2006) 121–132

heads drop significantly due to the dry year of 1998/1999, after which they recover again. On the contrary, the activation of the surface water irrigation district in 2000 has a clear impact. During the spring of 2001, heads continue to rise in all wells except 375, despite the lack of precipitation. Then quite remarkably the seasonal oscillations almost completely disappear and the high groundwater heads persist. In other words, the natural response to the dry season is almost entirely nullified. Even the deepest well (375) sees its oscillations greatly suppressed. Only the sequence of two extremely dry years, 2004 and 2005, caused a gradual drop of the groundwater heads, much less pronounced than in previous dry years (e.g. 1994/1995 and 1998/1999). Fig. 5 is compares the water table depths at the end of the summer of 1999 with those at around the same time of year in 2003, based on a dense monitoring network. The maps were created by subtracting the piezometric surfaces, obtained by kriging interpolation of the water table heights (in m above sea level), from the topographic surface. The largest differences (more than 15 m) are observed in the Miocene aquifer in the centre of the area, whereas in the Cretaceous marls in the south differences are 2–4 m. Noteworthy is the fact that in 2003 the water table at some places still reaches up to 1.5 m below the surface at the end of the summer. Fig. 4 shows that in the following months levels continue to rise in response to the high amount of rainfall. The main results of the chemical analyses, including some calculated chemical ratios and saturation indices are presented in Table 1. Several wells (T2, T9, T20, T22) were analysed before on two occasions, in January 1998 and in March 1999, whereas others only entered in one campaign, either in 1998 (T8, T15, T24, T39) or in 1999 (T81). The spring (T49) and well T25 were not sampled before, but T25 is located

near T30, a deeper well of the same aquifer, sampled in both former campaigns and therefore used for comparison. The complete data set of the older samples can be consulted in the work of Bonte (1999). Fig. 6 plots nitrate versus chloride concentrations for both old and new samples and reveals the existence of correlation between the two ions. However, concentrations are lower for many of the wells in 2003, below 1.5 mmol/l for nitrate (though frequently exceeding the drinking water guideline of 0.81 mmol/l), and below 6 mmol/l for chloride (except for well T22). The decrease is illustrated for three wells, namely T2, T8 and T20. Groundwater of well T22 is significantly reduced in NO3 , but maintains a high chloride level. Wells T15 and T39 located more to the east (Fig. 3) already had lower chloride contents in 1998/1999. Wells T9 and T25 have the lowest concentrations of both ions and do not show significant changes in time. An inverse trend is revealed by well T81, concentrations increasing with time. Regarding the sulphate concentrations (right plot of Fig. 6), the differences between old and new samples are more pronounced. Although the new samples still plot above the seawater mixing line (indicating the presence of an additional source of sulphate), their concentrations are always more than twice as low as the older samples of the corresponding wells. Values range from 0.14 mmol/l (T25) to 0.54 mmol/l (T2). Moreover, the new SO42 /Cl relationship seems approximately linear (when leaving out sample T22) and quite different from that of 1998/1999, as illustrated by the steep slopes of the lines connecting the samples T2, T8 and T20. The Na+/Cl ratio, plotted against chloride in the left plot of Fig. 7, is a good indicator of changes in groundwater chemistry, as it indicates cation exchange, a process found to occur whenever there is a disturbance in the chemical equilibrium

Fig. 4 – Groundwater head time series of six wells in the study area (see Fig. 3 for location); also indicated are: precipitation measured at Quelfes, estimated monthly irrigation and approximate dates of substitution of groundwater by surface water for urban supply and irrigation.

agricultural water management 85 (2006) 121–132

127

Fig. 5 – Water table contours at the end of the summers of 1999 and 2003.

Fig. 6 – Plots of NO3S vs. ClS (left) and SO42S vs. ClS (right) for wells sampled in 1998/1999 and 2003; also shown are the results of the hydrogeochemical simulation with PHREEQC.

Fig. 7 – Plots of Na+/ClS vs. ClS (left) and Ca2+/HCO3S vs. ClS (right) for wells sampled in 1998/1999 and 2003; also shown are the results of the hydrogeochemical simulation with PHREEQC.

128

agricultural water management 85 (2006) 121–132

between groundwater and the exchange complex. Values in rainwater and fresh groundwater are around the seawater ratio of 0.86 in coastal regions, indicated in the same plot. On the other hand, the exchange processes also influence the calcium concentrations and hence disturb the Ca2+/HCO3 equilibrium associated to calcite. Therefore, Ca2+/HCO3 ratios are plotted against chloride concentrations in the right graphic of Fig. 7. The samples that reveal a significant decrease of chloride in time (T2, T8 and T20) show a gradual increase of the Na+/Cl ratio and a decrease in the Ca2+/HCO3 ratio. On the other hand, the groundwater of well T22, which has a relatively high and constant salinity, shows little variations in any of the chemical ratios. The remaining samples of 2003 all have a Na+/Cl ratio above the seawater ratio and alkalinity nearly or more than twice as high as the Ca2+ concentration (in mmol/l). Most of them showed similar ratios in 1998/1999. The spring shows major ion concentrations comparable to those observed in groundwater samples located to the north and west, such as T15 and T20 (Table 1). Its nitrate and sulphate concentrations are among the lowest observed in the area. The main differences with other groundwater samples are a higher pH, extreme supersaturation with respect to calcite (and dolomite) and a lower calculated CO2 pressure (cf. Table 1). The obtained results reflect the observations made at three moments in time, which only permits the detection of longterm tendencies. For the purpose of analysing seasonal variations of salinity and nitrate levels in groundwater, Fig. 8 displays the monthly time series of NO3 and Cl concentrations of former water supply well 608/132. Unfortunately, this is the only well in the vicinity with available data. Both parameters reveal a decreasing trend since the introduction of surface water irrigation, though the lowering of the Cl concentration is more pronounced. In fact, the long-term trend is comparable to that of well T8 located nearby (see Fig. 3). The oscillations are irregular before 1999, reflecting the combined impact of the precipitation and groundwater extraction regimes. In 1999, a distinct increase in groundwater salinity occurs.

Fig. 8 – Time series of NO3S and ClS concentrations of former water supply well 608/132 (see Fig. 3 for location); also indicated are: precipitation measured at Quelfes and groundwater head time series of well 608/471.

5.

Discussion

5.1.

Aquifer dynamics

The sudden and almost complete attenuation of the seasonal groundwater head fluctuations after the year 2000 (Fig. 4) is the combined result of two factors associated to the activation of the surface water irrigation district. First of all, groundwater extractions and their impact on the groundwater heads in spring and summer are largely reduced. However, a natural lowering of the heads would still be expected to occur, in response to the lack of recharge in these seasons. This is avoided due to the introduction of an artificial source of recharge, namely the surface water irrigation return flow. The monthly distribution of irrigation quantities is shown in Fig. 4 and was estimated on the basis of average yearly and monthly crop water requirements for citrus trees in the Algarve (Toma´s, 2001) and the monthly precipitation totals. Only half the total calculated quantities are indicated, since approximately 50% of the study area is occupied by intensive agriculture (deduced from land use and orthophoto maps) and thus receives the artificial recharge. Irrigation return flow is estimated as 10–15% for drip irrigation systems such as those used in the study area (Beltra˜o, 1985; Keller and Bliesner, 2000). Beltra˜o (1985) performed field experiments in the region to study unavoidable water losses due to deep percolation in drip irrigation. If scheduling is poor, avoidable losses can be much higher. Keller and Bliesner (2000, pp. 471–472) refer to peak period and seasonal transmission ratios that represent the excess water that must be applied to offset the unavoidable deep percolation with drip irrigation. The authors indicate values for arid and humid regions, with different soil textures and rooting depths. Moreover, they refer that ‘‘the unavoidable excess depth of applied water is at least 10% on all parts of an area that is sufficiently irrigated to meet evapotranspiration demands’’. In addition, as the authors also refer in their book, it is a common and recommended practice to irrigate in excess of crop water requirements, especially in arid and semi-arid environments, in order to control soil water salinity and avoid salt accumulation. The increased discharge led to the reactivation of several springs, forming local discharge points of the aquifer. These springs previously were only active in periods with large amounts of rainfall, whereas now they even flow in the summer, as was confirmed during a field campaign in the summer of 2004, when the total discharge of four springs was 23 l/s, approximately four times the median aquifer yield (Almeida et al., 2000). Due to the extremely low amount of rainfall in the past year it is not yet possible to oversee the true consequences of the additional recharge component. Two possible scenarios can be imagined: (1) a new hydrodynamic equilibrium has already been reached and will not be much affected in the future; (2) the lack of precipitation has prevented a further rise of the water table, which in the Miocene aquifer is already near the surface in several areas (Fig. 5). In the latter, more probable scenario, a further rise could result in the water table entering the root zone (1–2 m) and causing damage to the citrus trees.

agricultural water management 85 (2006) 121–132

5.2.

Aquifer hydrochemistry

The increased discharge and additional recharge also have a noticeable effect on the hydrochemistry of the upper aquifer, as they induce groundwater flushing and dilution. The most distinct signs of flushing are revealed by wells T2, T20 and T8. Mixing with freshwater in the aquifer causes a drop in Cl , NO3 and SO42 and an increase in Ca2+ and HCO3 concentrations. As the equilibrium between water and exchange complex is disturbed, Na+ present on the complex is exchanged for Ca2+, the degree of which depends on the cation exchange capacity (CEC) of the soil and aquifer. The role of Mg2+ and K+ in this process is intermediate, because concentrations are often much smaller and both ions are adsorbed to the complex in favour of Na+, but not of Ca2+. At the same time the uptake of calcium causes the groundwater to become subsaturated with respect to calcite, which therefore dissolves (the mineral is a major constituent of both the Miocene sandy limestone aquifer and the Cretaceous marls). Consequently, the alkalinity increases relatively to calcium and the Ca2+/HCO3 ratio drops, as is observed for the three wells in Fig. 7. Distinct signs of flushing are also visible from 2001 onwards in the longer timeseries of the former public water supply well (Fig. 8). The nitrate level starts to drop at the end of 1998, when urban supply turns to surface water and extraction from the well ceases. No seasonal trends can be detected. Groundwater salinity however continues to rise until the year 2000, which can be explained by the fact that pumping in the surrounding wells continues for the purpose of irrigation. Recharge of the aquifers is extremely low in this period, as a result of the low amount of rainfall, so that increased and prolonged pumping activities start drawing brackish water from the Cretaceous marls in the south affected by ancient seawater intrusion (Bonte, 1999; Stigter and Carvalho Dill, 2001). The constant high Cl concentration of groundwater in well T22 indicates that mixing with brackish groundwater persists in this area and causes cation exchange of Ca2+ by Na+ on the complex. Groundwater velocities in the Cretaceous marls are extremely low, whereas the presence of the NW–SE trending fault just north of the well may form an additional barrier to groundwater inflow. Groundwater sampled at the spring shows typical signs of degassing of CO2, which occurs because the CO2 pressure of the atmosphere is much lower than that of soil and groundwater in aerobic conditions. A necessary consequence of degassing is precipitation of calcite, which explains the high saturation index indicating that precipitation was occurring at the time of sampling. The fact that in terms of composition the spring resembles well T20, could be an indication of the existence of groundwater flow along the NW–SE trending fault, possibly reaching the Ria Formosa lagoon along this flow path (Fig. 2). Besides the increase in groundwater discharge and the additional recharge, which generate a dilution effect, two additional factors need to be considered. Before 2001 local groundwater was extracted for irrigation, which, combined with return flow, induced a recycling process that gradually increased the groundwater salinity and nitrate concentration (nitrate behaving conservatively under aerobic conditions).

129

This salinisation mechanism is well described by Stigter et al. (1998, 2006) and is also found to occur under similar climatic conditions in other areas (e.g. Milnes and Renard, 2004; Paralta et al., 2000). At the same time the large amounts of nitrogen available in groundwater were never taken into account in fertilisation plans. With the introduction of surface water irrigation, the recycling process was interrupted and the salinisation potential of the return flow strongly diminished.

Table 2 – Description of the modelling process for the three simulated scenarios in PHREEQC Step description Scenario 1—Evolution of rainwater to fresh groundwater (a) Evapotranspiration of rainwater; recharge was calculated to be 18%, based on the Cl mass balance (b) Oxidation of ammonium (NH4+) present in rainwater to nitrate (NO3 ), in aerobic conditions, catalysed by bacteria (c) K+ fixation by soil and vegetation, to optimise the K+ concentration in fresh groundwater; to maintain charge balance, K+ is exchanged for Na+, which shows a slightly higher concentration than expected from seawater contribution only (d) Dissolution of gypsum and open system dissolution of calcite and dolomite, in equilibrium with a CO2 pressure of 10 1.3 atm, the maximum pressure found in groundwater of the area; calcite dissolution occurs till the point of saturation Scenario 2—Groundwater salinisation and nitrate contamination (irrigation with groundwater) (a) Definition of an exchanger, in equilibrium with fresh groundwater obtained from the previous simulation (b) Leaching of fertilisers: ammonium and calcium nitrate (NH4NO3, Ca(NO3)2); both applied in the area and representing slow and fast release of N, respectively; nitrification of NH4+ is considered to occur and leaching rate is optimised based on sample T20 (1999) (c) Evapotranspiration (ET) and irrigation return flow (IRF); this step is modelled simultaneously with step b, as IRF provides the vehicle for the leaching of fertilisers; ET is performed in six steps, with constant leaching rate and in equilibrium with calcite and the exchange complex; the first four steps reach the Cl concentration of sample T20 (1999), the last two simulate further salinisation Scenario 3—Flushing and dilution of contaminated groundwater (irrigation with surface water) (a) Definition of a new water type: the average composition of surface irrigation water, estimated on the basis of surface water quality data provided by the Instituto Nacional de A´gua (INAG) (b) Definition of surface water IRF: 10% regarded for drip irrigation systems (Beltra˜o, 1985; Keller and Bliesner, 2000); fertilisers are added to simulate leaching, the type and quantity consistent with step b of scenario 2; the final solution is brought into equilibrium with calcite and the CO2 pressure of scenario 1 (c) Mixing of IRF with fresh groundwater obtained from simulation 1, with a ratio 1:3, in equilibrium with calcite; the ratio roughly reflects that of artificial to natural recharge, considering that 50% of the total area is irrigated (d) Mixing of the obtained water type with contaminated groundwater (scenario 2), in equilibrium with calcite and the exchange complex retained from scenario 2; the mixing ratio is optimised to obtain the Cl concentration of sample T20 (2003) and is found to be 4:1

130

agricultural water management 85 (2006) 121–132

simulating ET in four steps (Table 2), shows a good fit with the majority of the samples analysed in 1998 and 1999. This was confirmed by plotting the other samples analysed in the same period (not shown in the graphic). Higher sulphate concentrations in groundwater occur where fertilisers such as ammonium sulphate are applied, as this anion is absorbed only in small quantities by the plants (Quelhas dos Santos, 1991) and therefore easily leached out of the soil to the water table. The model lines are dashed for the last two simulation steps (further ET, see Table 2) and indicate that salinisation and contamination could have increased further if the introduction of surface water had not stopped the process. Samples T2, T8 and T22 plot below the lines, indicating an additional source of chloride, possibly provided by mixing with brackish groundwater drawn from the Cretaceous marls in the south, as a result of intensive pumping. Cation exchange and calcite precipitation were induced by the groundwater recycling process in this scenario and are revealed by the lowering of the Na+/Cl ratio and the increase of the Ca2+/HCO3 ratio, as illustrated by the model lines in Fig. 7. In these plots, the simulations do not show a good fit with those samples that already showed signs of flushing (e.g. T15, T24, T39), most likely induced by the increase in recharge that occurred after the shutdown of the public extraction wells. The amount of exchanger was estimated to be 1 mequiv./100 g of dry soil, which reflects the low amount of clay present in the upper aquifers of Campina da Luz. Stigter et al. (1998) found much stronger indications of cation exchange (i.e. lower Na+/Cl and higher Ca2+/HCO3 ratios) in the upper formations of the area near Faro (Fig. 1) and determined the CEC to be almost 10 times higher. The final scenario simulates the change in groundwater chemistry since the introduction of surface water for irrigation. Modelled and observed changes correspond well for most ions (Table 3), the largest deviations occurring for sodium and sulphate. The corresponding lowering of the salinisation and contamination degree is shown by model line 2 in both graphics of Fig. 6. In the nitrate versus chloride plot, the line is superimposed by the dashed evolution line of well T20, as its observed concentrations are identical to the modelled ones in the last two simulations (cf. Table 3). The observed decrease in nitrate is less pronounced for wells T8 and T2, relatively to

Moreover, due to the low nitrogen content in surface water, the problem of excess fertilisation was greatly reduced. In other words, the agricultural practices were accidentally improved.

5.3.

PHREEQC model

In order to obtain a clear picture of the hydrogeochemical evolution from the beginning of intensive agricultural activity until the present-day situation, a hydrogeochemical model was built with PHREEQC. The following scenarios were simulated: (1) evolution of rainwater to fresh groundwater; (2) groundwater salinisation and nitrate contamination induced by fertiliser application and irrigation with groundwater and (3) flushing and dilution of contaminated groundwater, under influence of continued fertiliser application and irrigation with surface water. A short description of each scenario is provided in Table 2, whereas Table 3 provides the results of the simulations. The average composition of rainwater is based on samples collected and analysed by the Vrije Universiteit Amsterdam between 1979 and 1981. The contribution of seawater is calculated for every ion on the basis of their seawater ratio with chloride (conservative) and shown in Table 3. The residuals that exist after subtraction of the seawater contribution must be accounted for by other sources (Appelo and Postma, 1994). The additional sources are largest for calcium and bicarbonate, most likely associated to calcite containing dust. The presence of sulphate, nitrate and ammonium is of anthropogenic origin (mainly agriculture, some industry and traffic fumes). Fresh groundwater is represented by the oldest sample of well T30, which extracts groundwater from Jurassic limestones at 125 m depth. The results of the first scenario simulation correspond well to the observed concentrations, partly due to the relative simplicity of the model. The modelled NO3 concentration is 0.53 mmol/l, which demonstrates that the amount of N present in the precipitation of the study area can produce 33 mg/l of NO3 in groundwater. The simulation of the second scenario (intensive agriculture, irrigation with groundwater) is shown in Figs. 6 and 7. For nitrate, sulphate and chloride, the modelled evolution line

Table 3 – Concentrations of observed and modelled water types

Average rainwater Seawater contribution Other source

PHREEQC First scenario model Fresh gw (T30-1998) Second scenario model Polluted gw (T20-1999) Third scenario model Flushed gw (T20-2003)

pH

Na+

K+

Mg2+

Ca2+

6.70

0.297 0.296 0.001

0.051 0.006 0.044

0.058 0.034 0.024

0.349 0.007 0.342

0.345 0.345 0.000

0.780 0.001 0.779

6.91 6.95 7.22 7.41 7.22 7.32

1.88 1.88 4.56 4.70 3.36 3.72

0.06 0.03 0.10 0.12 0.09 0.23

0.82 1.19 1.29 1.39 1.05 0.95

3.41 3.18 3.92 4.40 2.93 2.87

1.92 1.92 5.64 5.64 3.37 3.38

7.36 7.43 6.16 5.90 5.99 5.90

Solutes in mmol/l, except for NH4+, in mmol/l; gw = groundwater.

Cl

HCO3

SO42

Na+/Cl

Ca+/HCO3

NO3

NH4+

0.052 0.018 0.035

0.052 0.000 0.052

42.5 0.00 42.5

0.86

0.45

0.39 0.42 1.14 1.31 0.52 0.34

0.53 0.61 2.02 2.05 1.09 1.09

0.00 1.50 0.00 1.66 0.00 0.00

0.98 0.98 0.81 0.83 1.00 1.10

0.46 0.43 0.64 0.75 0.49 0.49

agricultural water management 85 (2006) 121–132

chloride, which corresponds to the fact that the higher concentrations were probably influenced by an additional source of chloride. For sulphate, the observed decrease appears somewhat more pronounced than could be accounted for by the model. Most samples currently have a sulphate concentration near the seawater mixing line, which means that the contribution of other sources, such as dissolution of gypsum minerals or mineral fertilisers must be very low. Another possibility could be the occurrence of sulphate reduction, which only occurs in anaerobic environments in the presence of organic matter and catalysing bacteria (Appelo and Postma, 1994). Low sulphate concentrations, found mainly in the Cretaceous marls, were ascribed to the same process by Bonte (1999). However, in reducing conditions nitrate would be expected to disappear as well, even before sulphate, since it has a higher energy yield. This topic requires further investigation. The cation exchange processes that were thought to occur during flushing, as a result of groundwater freshening, are indeed observed in the hydrochemical simulation of PHREEQC, as shown by model line 2 in both plots of Fig. 7. The observed evolution lines of the wells T2, T8 and T20 are parallel to the model lines, thus showing a similar trend.

6.

Final considerations

The decision to construct a regional surface water irrigation district in order to provide the farmers in the region with highquality water, is a good example of how water management policies are sometimes implemented without considering all the potential consequences, advantages and disadvantages. In terms of water quality, the effect in the upper aquifers has been almost immediate, which proves that travel times in the vadose zone and aquifer must be short and that the role of N–S faults as preferential flow paths may be significant. The consequences can be considered to be beneficial, as groundwater freshening occurs and nitrate concentrations appear to decrease. The reduced nitrogen load is directly linked to the absence of nitrogen in surface water and thus a much lower chance of overfertilisation. Freshening occurs mainly due to the end of the groundwater extractions (and associated recycling process), the increased groundwater discharge and the additional recharge from return flow. It can be concluded that in semi-arid conditions the degree of nitrate contamination and salinisation depend largely on the irrigation source. Nevertheless, it is currently impossible to say if contamination levels will decrease much further, as fertilisers continue to be applied in large quantities and leaching still occurs. Irrigation return flow can be little mineralised, but extremely high in nitrate, as was found to occur in a surface water irrigation district of the Ebro River Basin in Spain (Causape´ et al., 2004). Besides, many households are still not connected to the municipal sewerage network, so that domestic wastewater leakage from septic tanks constitutes an additional source of nitrogen. The shift towards surface water irrigation also led to a sharp rise of the water table, a tendency that was only slightly inverted in 2005 due to the extremely low amount of precipitation. A further rise of the water table is expected in

131

the future, thereby possibly entering the root zone at some locations and damaging the citrus trees. In other words, the consequences of irrigating with surface water, beneficial as they may appear, could eventually become problematic and a more detailed analysis with the help of powerful tools such as groundwater flow simulation models would be recommendable. Problems could be avoided by using an integrated mixedsource irrigation system, using both groundwater and surface water. Another advantage of such a system would be avoiding the overexploitation of the water reservoirs, which are also used for urban water supply and in dry years such as 2005 show a rapid decline of their storage volume. Finally, the use of groundwater would imply the contemplation of the available nitrogen for fertilisation, following the Code of Good Agricultural Practices, which due to the application of the Nitrates Directive has become mandatory. Consequently, nitrate concentrations in the aquifer would be further reduced, minimizing the risk of nutrient-rich groundwater entering the Ria Formosa lagoon.

Acknowledgements The first author would like to thank the Fundac¸a˜o para a Cieˆncia e a Tecnologia for granting him a scholarship, Jose´ Carlos Tomas of the Agricultural Directorate for providing useful information and his valuable opinion, Jani Santos for performing the chemical analyses of the groundwater samples in 2003, Maria Jose´ Rodrigues for her help in preparing the sampling campaign, Joa˜o Vieira for his help during the sampling campaign and his bright ideas and Tony Appelo for looking at the PHREEQC model. The useful comments of the two reviewers of this article were also greatly appreciated.

references

Almeida, C., Mendonc¸a, J.J.L., Jesus, M.R., Gomes, A.J., 2000. Sistemas aquı´feros de Portugal Continental, Report INAG, Lisbon, Portugal. Available at http://snirh.inag.pt/ snirh.php?main_id=3&item=2.4&idioma=english. Appelo, C.A.J., Postma, D., 1994. Geochemistry, Groundwater and Pollution, 2nd ed. Balkema, Rotterdam, The Netherlands, 536 pp. Beltra˜o, J., 1985. A rega localizada, Report Universidade do Algarve. Faro, Portugal, 31 pp. Bonte, M., 1999. A chemical and isotopic study after the hydrogeochemical processes occurring under irrigated land in Campina da Luz, Algarve, Portugal. M.Sc. Thesis, Vrije Universiteit, Amsterdam, The Netherlands, 76 pp. Cameira, M.R., Fernando, R.M., Pereira, L.S., 2003. Monitoring water and NO3–N in irrigated maize fields in the Sorraia Watershed, Portugal. Agric. Water Manage. 60, 199–216. Causape´, J., Quı´lez, D., Aragu¨e´s, R., 2004. Assessment of irrigation and environmental quality at the hydrological basin level. II. Salt and nitrate loads in irrigation return flows. Agric. Water Manage. 70, 211–228. Chen, J., Tang, C., Sakura, Y., Yu, J., Fukushima, Y., 2005. Nitrate pollution from agriculture in different hydrogeological zones of the regional groundwater flow system in the North China Plain. Hydrogeol. J. 13, 481–492.

132

agricultural water management 85 (2006) 121–132

De Bruin, J., 1999. Report on groundwater research in Luz de Tavira, Algarve Portugal: Water balance, hierarchical clustering chemical data, Slingram electromagnetic survey. M.Sc. Thesis, Vrije Universiteit, Amsterdam, The Netherlands, 43 pp. Diez, J.A., Roman, R., Caballero, R., Caballero, A., 1997. Nitrate leaching from soils under a maize-wheat-maize sequence, two irrigation schedules and three types of fertilisers. Agric. Ecosyst. Environ. 65, 189–199. EC, 2002. Implementation of Council Directive 91/676/EEC concerning the protection of waters against pollution caused by nitrates from agricultural sources; synthesis from year 2000 Member States reports. OOPEC, Luxembourg, 44 pp. EEA, 2003. Europe’s water: An indicator-based assessment. Topic Report, EEA, Copenhagen, Denmark, 97 pp. Goodchild, R.G., 1998. EU Policies for the reduction of nitrogen in water: the example of the Nitrates Directive. Environ. Pollut. 102 (S1), 737–740. Jalali, M., 2005. Nitrates leaching from agricultural land in Hamadan, western Iran. Agric. Ecosyst. Environ. 110, 210–218. Keller, J., Bliesner, R.D., 2000. Sprinkle and Trickle Irrigation. Blackburn Press, Caldwell, NJ, USA, 652 pp. Liu, G.D., Wu, W.L., Zhang, J., 2005. Regional differentiation of non-point source pollution of agriculture-derived nitrate nitrogen in groundwater in northern China. Agric. Ecosyst. Environ. 107, 211–220. Milnes, E., Renard, P., 2004. The Problem of Salt Recycling and Seawater Intrusion in Coastal Irrigated Plains: An Example from the Kiti Aquifer (Southern Cyprus). Mota, A.M.C.da, Dias, A.M.M., 2000. Aproveitamento hidroagrı´cola do Sotavento Algarvio. IHERA, Lisbon, Portugal, 23 pp. Newton, A., Icely, J.D., Falca˜o, M., Nobre, A., Nunes, J.P., Ferreira, J.G., Vale, C., 2003. Evaluation of eutrophication in the Ria Formosa coastal lagoon, Portugal. Cont. Shelf Res. 23, 1945– 1961. Oenema, O., Van Liere, L., Plette, S., Prins, T., Van Zeijts, H., Schoumans, O., 2004. Environmental effects of manure policy options in The Netherlands. Water Sci. Technol. 49 (3), 101–108. Oenema, O., Van Liere, L., Schoumans, O., 2005. Effects of lowering nitrogen and phosphorus surpluses in agriculture on the quality of groundwater and surface water in The Netherlands. J. Hydrol. 304, 289–301. Paralta, E.A., Stigter, T.Y., Salgueiro, A., 2000. Caracterizac¸a˜o hidroquı´mica do complexo gabrodiorı´tico da regia˜o de Beja e modelac¸a˜o hidrogeoquı´mica PHREEQC da composic¸a˜o da a´gua sob influeˆncia clima´tica - resultados preliminares. ´ gua, APRH, Lisbon, Portugal In: Proc. 58 Congresso da A (CD-ROM).

Parkhurst, D.L., Appelo, C.A.J., 1999. User’s guide to PHREEQC (version 2.0)—A Computer Program for Speciation, Batchreaction, One-dimensional Transport, and Inverse Geochemical Calculations. Vol. Water-Resources Investigations Report 99-4259. US Geological Survey, Denver, Colorado, USA. Paz, J.M. de., Ramos, C., 2004. Simulation of nitrate leaching for different nitrogen fertilization rates in a region of Valencia (Spain) using a GIS–GLEAMS system. Agric. Ecosyst. Environ. 103 (1), 59–73. Quelhas dos Santos, J., 1991. Fertilizac¸a˜o: fundamentos da utilizac¸a˜o dos adubos e correctivos. Francisco Lyon de Castro. Europa-Ame´rica, Mem Martins, Portugal, 441. Rajmohan, N., Elango, L., 2005. Nutrient chemistry of groundwater in an intensively irrigated region of southern India. Environ. Geol. 47, 820–830. Ribeiro, L., Paralta, E., Nascimento, J., Amaro, S., Oliveira, E., Salgueiro, R., 2002a. A agricultura e a delimitac¸a˜o das zonas vulnera´veis aos nitratos de origem agrı´cola segundo a Directiva 91/676/CE. In: Proc. III Congreso Ibe´rico sobre Gestio´n e Planificacio´n del Agua, Universidad de Sevilla, Spain, pp. 508–513. Ribeiro, L., Paralta, E., Stigter, T., Carvalho Dill, A., 2002b. Avaliac¸a˜o por me´todos estoca´sticos da poluic¸a˜o das a´guas ´ &A Cieˆncia subterraˆneas por nitratos de origem agrı´cola. A ´ gua & Ambiente no. 40, pp. 1, 5–7. no. 6, separata of A Silva, M.O. da, 1984. Hidrogeologia do Algarve Oriental. Ph.D. Thesis, Universidade de Lisboa, Lisbon, Portugal, 260 pp. Stigter, T.Y., Carvalho Dill, A.M.M., 2001. Estudo geolo´gico e hidrogeoquı´mico das regio˜es abrangidas pelo projecto Interreg II: Efeitos do Uso Intensivo de Fertilizantes e ´ guas Produtos Fitossanita´rios na Qualidade do Solo e das A Subterraˆneas. Report Universidade do Algarve, Faro, Portugal, 67 pp. Stigter, T.Y., Van Ooijen, S.P.J., Post, V.E.A., Appelo, C.A.J., Carvalho Dill, A.M.M., 1998. A hydrogeological and hydrochemical explanation of the groundwater composition under irrigated land in a Mediterranean environment, Algarve, Portugal. J. Hydrol. 208, 262–279. Stigter, T.Y., Ribeiro, L., Carvalho Dill, A.M.M., 2006. Evaluation of an intrinsic and a specific vulnerability assessment method in comparison with groundwater salinisation and nitrate contamination levels in two agricultural regions in the south of Portugal. Hydrogeol. J. 14 (1–2), 79–99. Toma´s, J.C., 2001. A rega gota-a-gota de citrinos recomendac¸o˜es pra´ticas. Information brochure of the Direcc¸a˜o Regional da Agricultura do Algarve, Faro, Portugal. Wolf, J., Ro¨tter, R., Oenema, O., 2005. Nutrient emission models in environmental policy evaluation at different scales— experience from The Netherlands. Agric. Ecosyst. Environ. 105, 291–306.