Content area
Salt loading in irrigation return flows contributes to the salinization of the receiving water bodies, particularly when originated in salt-affected areas as frequently found in the middle Ebro River basin (Spain). We determined the salt loading in La Violada Gully from the total dissolved solids (TDS) and flows (Q) during the 1995 to 1998 hydrological years. Since this gully collects flows from various sources, an end-member mixing analysis (EMMA) was performed to quantify the drainage flow from La Violada Irrigation District (VID). Three flow components were identified in La Violada Gully: drainage waters from VID (Q^sub d^); tail-waters from irrigation ditches, spill-over, and seepage from the Monegros Canal (Q^sub o^); and ground water inflows (Q^sub g^) originating in the dryland watershed. Gypsum in the soils of VID was the main source for salts in La Violada Gully (flow-weighted mean TDS = 1720 mg L^sup -1^, dominated by sulfate and calcium). The contribution of Q^sub g^ to the total gully flow during the 1996 irrigation season was low (6.5% of the total flow). The 1995 to 1998 annual salt load average in La Violada Gully was 78628 Mg, 71% of which was exported during the irrigation season. The 1995 to 1998 irrigation season salt load average in Q^sub d^ was 43015 Mg (77% of the total load). Thus, irrigated agriculture in VID was the main source of salt loading in this gully, with a yield of 11.1 Mg of salts per hectare of irrigated land for the irrigation season. Efficient irrigation systems and irrigation management practices that reduce Q^sub d^ are key factors for controlling off-site salt pollution of these gypsum-rich irrigated areas. [PUBLICATION ABSTRACT]
Full text
ABSTRACT
Salt loading in irrigation return flows contributes to the salinization of the receiving water bodies, particularly when originated in salt-affected areas as frequently found in the middle Ebro River basin (Spain). We determined the salt loading in La Violada Gully from the total dissolved solids (TDS) and flows (Q) during the 1995 to 1998 hydrological years. Since this gully collects flows from various sources, an end-member mixing analysis (EMMA) was performed to quantify the drainage flow from La Violada Irrigation District (VID). Three flow components were identified in La Violada Gully: drainage waters from VID (Q^sub d^); tail-waters from irrigation ditches, spill-over, and seepage from the Monegros Canal (Q^sub o^); and ground water inflows (Q^sub g^) originating in the dryland watershed. Gypsum in the soils of VID was the main source for salts in La Violada Gully (flow-weighted mean TDS = 1720 mg L^sup -1^, dominated by sulfate and calcium). The contribution of Q^sub g^ to the total gully flow during the 1996 irrigation season was low (6.5% of the total flow). The 1995 to 1998 annual salt load average in La Violada Gully was 78628 Mg, 71% of which was exported during the irrigation season. The 1995 to 1998 irrigation season salt load average in Q^sub d^ was 43015 Mg (77% of the total load). Thus, irrigated agriculture in VID was the main source of salt loading in this gully, with a yield of 11.1 Mg of salts per hectare of irrigated land for the irrigation season. Efficient irrigation systems and irrigation management practices that reduce Q^sub d^ are key factors for controlling off-site salt pollution of these gypsum-rich irrigated areas.
Abbreviations: EC, electrical conductivity; EMMA, end-member mixing analysis; ET, evapotranspiration; Q, flow; Q^sub d^, subsurface drainage flow; Q^sub g^, ground water flow; Q^sub o^, surface flow; SAR, sodium adsorption ratio; TDS, total dissolved solids; VID, La Violada Irrigation District.
THE DIFFUSE CONTAMINATION of rivers and aquifers by dissolved salts drained from irrigated soils is recognized as an important and increasing environmental problem in many areas around the world (van Schilfgaarde, 1994). Salt loading in irrigation return flows is due primarily to evapo-concentration and weathering effects (El-Ashry et al., 1985; Aragüés and Tanji, 2003). The former always takes place under irrigation, as growing plants extract water through evapotranspiration and leave behind most of the dissolved salts, increasing their concentration in the soil water. Irrigation also adds to the salt load in irrigation return flows by leaching salts arising from weathered minerals occurring in the soil or deposited below.
Irrigation return flows consist of three components (Aragüés and Tanji, 2003): overflow or bypass waters (i.e., operational spills from distribution conveyances), tail-waters (i.e., surface runoff from the lower ends of irrigated fields), and subsurface drainage waters (i.e., the portion of the infiltrating water that flows through the soil and is collected by the subsurface drainage system). The proportions of these components determine the final quality of irrigation return flows: bypass waters are similar in quality to that of the irrigation water, tail-waters may pick up sediments and associated contaminants such as phosphorous and pesticides, and subsurface drainage waters may transport substantial amounts of salts and agrochemicals and are generally the primary source of pollution associated with irrigated agriculture.
Salt loading in irrigation return flows depends mainly on the salinity of the irrigation water, the minerals present in the soil and subsoil, and the management of irrigation water (i.e., the leaching fraction). Depending on these variables, keeping an adequate salt balance in the crop's root zone may lead to an unacceptable increase in the salinity of the receiving water bodies. Therefore, the quantification of salt loading in irrigation return flows is critical since the prediction of the resultant salinity in a body of water after mixing with the irrigation return flows requires knowledge of the mass of salts in each contributing body.
Soil salinity in the middle Ebro River basin (northeastern Spain; Fig. 1) has a geological and climatological origin (Alberto et al., 1986). About 310000 ha out of the present 800000 irrigated ha are affected by salinity or sodicity (Confederación Hidrográfica del Ebro, 2004a, 2004b), and their irrigation return flows have contributed to the salinization of water courses in this basin (Alberto and Aragüés, 1986; Quílez et al., 1992). We studied the off-site salt pollution originating in La Violada Irrigation District, a gypsum-rich area located in the middle Ebro River basin and representative of numerous irrigated areas in this basin.
The objectives of this study were to (i) characterize the chemical quality of La Violada Gully waters, the main drainage outlet of La Violada irrigation district, (ii) determine the various end-member flow components contributing to the total flow in La Violada Gully, and (iii) estimate the total salt loading exported through La Violada Gully and especially the salt loading originated in La Violada Irrigation District (the salts removed from the irrigated soils).
A companion paper focuses on the nitrogen export patterns through La Violada Gully in relation to the fertilization practices in La Violada Irrigation District (Isidoro et al., 2006).
Fig. 1. Location of La Violada Gully watershed upstream of the D-14 monitoring station. Drainage network, sampling points, and irrigation canals limiting La Violada Irrigation District.
MATERIALS AND METHODS
Description of the Study Area
La Violada Irrigation District (VID) belongs to the Monegros I irrigation scheme and is located in the middle Ebro River basin (northeastern Spain; 42°01' N, 0°35' W). The VID is drained by La Violada Gully, monitored at the gauging station D-14 (Fig. 1). The watershed of La Violada Gully upstream of D-14 has a total area of 19637 ha. The VID is located in the lower part of this watershed (roughly 400 m above sea level), whereas the upper part is a dryland area dedicated to winter crops (mainly wheat, Triticum aestivum L.) and rangeland (Fig. 1). The total area of VID is 5282 ha, 3866 of which are irrigated (average of the four study years) and the rest are non-irrigated agricultural land (834 ha), rangeland (166 ha), pine tree forest (109 ha), and non-productive land (307 ha).
The climate of VID is dry, subhumid, and mesothermic, with mean annual values for the 1965-1998 period of 469 mm (precipitation), 13.3°C (temperature), and 1124 mm (Hargreaves reference or potential evapotranspiration, ET^sub o^).
The upper dryland area of the watershed consists mainly of Tertiary calcareous rocks with some presence of gypsum while the heights to the west and south of the irrigated land are composed mainly of tabular gypsum rocks (Institute Tecnologico Geominero de España, 1995). The VID has some heights of gypsum, but the main part consists of Quaternary alluvial and colluvial deposits, with glacis and alluvial fans in the northeast. The soils in this northeastern VID are shallow, coarse-textured, stony, low in water-retention capacity (often lower than 50 mm determined to a depth of 120 cm or to a limiting layer; Faci et al., 2000), and sometimes have developed a petrocalcic horizon. The bottoms of the valleys along the Artasona and Valsalada Ditches are formed by alluvial silt, clay, and gravel deposits and have limited drainage. The soils developed in these valleys are deeper and almost stone-free with a higher water retention capacity (up to 100 mm) (Faci et al., 2000). The soils in VID are generally high in gypsum (>3%) and calcite (>30%), and the subsurface drainage waters are at or close to gypsum and calcite saturation.
Most of the VID area is flood irrigated with blocked-end plots, although some minor areas have pressurized systems. Traditional irrigated crops are alfalfa (Medicago saliva L., 824 ha, mean value of 1995 to 1998), corn (Zea mays L., 1924 ha), wheat (189 ha), barley (Hordeum vulgare L., 323 ha), and sunflower (Helianthus annuus L., 178 ha). Other small areas are dedicated to fruit trees (apple, Malus domestica Borkh., and olive, Olea europaea L., 52 ha), rice (Oryza saliva L., 43 ha), and pepper (Capsicum annuum L., 44 ha).
Surface irrigation efficiency in the district is low. Playán et al. (2000) found a mean on-farm application efficiency of 62% (ratio of the irrigation depth contributing to replenish the management allowable depletion to total irrigation depth). Isidoro et al. (2004) estimated a similar on-farm water consumptive use coefficient of 61% (ratio of the crop ET calculated through a daily water balance minus the effective precipitation to the irrigation depth minus the operational losses, estimated as the fraction of canal waters present in the gully water) and distribution efficiency of about 83 % (ratio of irrigation depth minus operational losses to irrigation depth). This inefficiency contributes to the salt loading of the gully waters through the gypsum-laden drainage waters. In recent years, the district's irrigation authority has begun the modernization of the irrigation system, with the shift to sprinkle irrigation in some areas and the construction of in-line distribution reservoirs.
The drainage network in VID consists of a dense system of open ditches (secondary ditches) that flow into the Valsalada and Artasona collectors. These two main ditches join upstream of D-14 forming La Violada Gully, the only surface outlet of La Violada watershed (Fig. 1). Many plots have also buried drains, especially in the lower areas near the two main ditches. The buried drains discharge into the system of open ditches. The secondary ditches collect the subsurface drainage water from the irrigated fields either directly or through the buried drains. Tail-waters from irrigated fields are small (because plots usually have blocked ends), but when they exist, they are also collected by the secondary ditches. Three natural gullies (Las Pilas, Azud, and Valdepozos) drain the upper dryland area of La Violada watershed and flow under the Monegros Canal into the drainage network (Fig. 1). Percolation to a regional aquifer system was considered nil due to the presence of an impervious stratum underlying the district (Torres, 1983; Faci et al., 1985).
La Violada Gully collects waters derived from operational spills from the Monegros Canal, seepage of irrigation Canals, bypass water from the irrigation ditches (usually irrigation ditches discharge to the drains when all the plots along a ditch have been irrigated until the gate at its mouth is closed), subsurface drainage waters originating in VID (collected by the subsurface and secondary drains), and surface and ground water inflows originated in the upper dryland watershed. Since the main scope of our study was to quantify the off-site salt-loading environmental impact derived from irrigation in VID, we identified and estimated these different flow components through an end-member mixing analysis (EMMA) (Mulholland, 1993; Durand and Juan Torres, 1996).
Water Quality Sampling
Field surveys for sampling drainage waters at 35 points in VID (labeled D or I in Fig. 1, including the D-14 monitoring station at the outlet of the district), irrigation water in Los Monegros Canal (CMO; Fig. 1), and surface inflows through Las Pilas, Azud, and Valdepozos gullies (labeled E; Fig. 1) were performed from December 1994 to March 1997, once a month during the non-irrigation season (October to March) and every 2 wk during the irrigation season (April to September). A spring close to Las Pilas Gully upstream of Los Monegros Canal (Fuente de los Tres Caños-F3C; Fig. 1) was also sampled in some surveys from July 1995 to February 1998. This spring was assumed to be representative of the ground water inflows into VID that originate in the upper dryland watershed.
The water samples were analyzed for electrical conductivity (EC, dS m^sup -1^ 25°C) using an A/S CDM83 conductivity meter (Radiometer, Brønshøj, Denmark), and chloride (Cl-), sulfate (SO^sub 4^^sup 2-^), and nitrate (NO^sub 3^^sup -^ as nitrate) concentration using a 2000isp ion chromatograph (Dionex, Sunnyvale, CA). The samples of the eight surveys from December 1994 to April 1995, three samples of the 1995 irrigation season, and two samples of the 1996 irrigation season were also analyzed for calcium (Ca^sup 2+^), magnesium (Mg^sup 2+)^, and sodium (Na+) using a Model 3030 atomic absorption spectrophotometer (PerkinElmer, Wellesley, MA). The ionic composition of the VID drainage waters (Q^sub d^) was characterized using the data of the secondary drains taken in 11 sampling dates (six in the 1994-1995 non-irrigation season and five in the 1995 and 1996 irrigation season) with full records of EC, cations, and anions. A few sampling dates, which showed a clear dilution (EC much lower than the mean EC), were discarded in this analysis. The secondary drains D-2, I-1, and I-5 receive water directly from Los Monegros Canal (leakage and direct releases through the gates) and exhibit a lower salinity, not representative of drainage water. Thus, these three secondary drains (D-2, I-1, and I-5), the D-3 drain that collects municipal wastewaters from Valsalada village, and the small drains that are tributaries of others were not used to characterize the quality of Q^sub d^.
The D-14 monitoring station located in La Violada Gully at the outlet of the study area belongs to the monitoring network of the Ebro River Basin Authority (Confederación Hidrográfica del Ebro). Water height records, rating tables, and mean daily flows during the 1995-1998 study period were provided by the Ebro River Basin Authority. An S-4040 automatic water sampler (Manning Environmental, Georgetown, TX) was installed at D-14 during this period. Two samples per day (at noon and midnight) were taken from April to November 1995 and one sample per day (usually at 1000 h) was taken for the rest of the period. Electrical conductivity, Cl-, SO^sub 4^^sup 2-^, and NO^sub 3^^sup -^ were determined in these samples.
Sixty-nine samples at D-14 (taken from May 1997 to August 1998) and six samples at the spring Fuente de los Tres Caños (taken from May 1997 to February 1998) were also analyzed for major cations and anions. In addition, 47 of the D-14 samples were analyzed for total dissolved solids (TDS) by weighing the dry residue after drying of the filtered samples in an oven at 105°C for 24 h. These samples were used to derive the relationship between TDS and EC necessary for the calculation of salt loads.
Finally, the D-14 waters were sampled in 10 intensive surveys every 2 h during 2 d (i.e., 24 samples in each survey) from April to October 1995 for analysis of EC and NO^sub 3^^sup -^. These samples were used to analyze the daily variations in salt load and the influence of the sampling hour in the estimation of total load (Isidoro et al., 2003). In 1995, the field surveys were always performed in one of the two days of these intensive surveys, so that the quality of Q^sub d^ and Q^sub g^ (considered constant in the 2-d period) can be used for flow separation within the 48-h periods.
Fig. 2. Electrical conductivity (EC) and Cl of La Violada Gully waters at the D-14 monitoring station (D-14), drainage waters (Q^sub d^) of La Violada Irrigation District (VID), ground water inflows into VID (Q^sub g^), and bypass water and operational losses from Monegros Canal (Q^sub o^) from December 1994 to April 1997.
End-Member Mixing Analysis
Isidoro et al. (2004) calculated the water budget of VID for the 1995 to 1998 hydrological years. The inputs considered were: irrigation water (including the operational losses), precipitation, direct water releases from the Monegros Canal into the drainage system, surface runoff, and municipal wastewaters. The outputs were: surface drainage outflow at D-14 (Q) and evapotranspiration (ET) in the whole surface of the study area. Evapotranspiration was calculated through a daily soil water balance performed for each main crop on each main soil type found in the district (Isidoro et al., 2004). Ground water outflows from VID were also considered to be small due to the narrow exit of the basin at D-14 and because the main flow in VID is directed toward the gully and not along it (Toth, 1963). The calculated water outputs exceeded the inputs by 23%. This unbalance was attributed to two unaccounted water inputs into VID: ground water inflows and canal seepage during the irrigation season and ground water inflows during the non-irrigation season.
Table 1. Chemical characteristics of the three end-members identified in La Violada Gully: Los Monegros Canal water (Q^sub o^), ground water inflows (Q^sub g^) into La Violada Irrigation District (VID), and drainage waters originated in VID (Q^sub d^); and for the water of La Violada Gully in the outlet of the district (D-14). Mean values for the 1995-1998 period (data for all the sampling dates with complete analysis).
Whereas the EC at D-14 may be explained by the mixing of canal (Q^sub o^) and drainage (Q^sub d^) waters, the occasionally higher Cl- concentrations in D-14 than in Q^sub d^ during the nonirrigation season (Fig. 2) may only be explained by the existence of another flow component with higher Cl- concentrations (Q^sub g^; Table 1). The waters in La Violada gully at D-14 were assumed to result from the mixing of three flow components: (i) irrigation water arising as bypass waters from irrigation ditches, operational spills, direct releases from the Monegros Canal, and canal seepages collected by lateral drains (Q^sub o^); (ii) subsurface drainage waters originating in VID (Q^sub d^); and (iii) ground water inflows into VID originating in the upper dryland watershed (Q^sub g^). Table 1 gives the mean chemical analysis of these components (Q^sub o^, Q^sub d^, and Q^sub g^), and Fig. 2 the EC and Cl- values of these components and of the waters sampled at D-14 from December 1994 to April 1997.
To perform an EMMA, EC and Cl of each end-member must be independently determined through sampling: Monegros Canal waters for Q^sub o^, Fuente de los Tres Caños spring waters for Q^sub g^, and VID drainage waters for Q^sub d^. As the spring was not sampled frequently enough throughout the non-irrigation season, the EMMA was not performed for the non-irrigation season. But the surface (canal releases or surface runoff) and subsurface components of flow on the non-irrigation season were estimated by hydrograph separation based on EC to state the salt contribution of canal releases and surface runoff (Isidoro et al., 2004).
1995 Irrigation Season
During the 1995 irrigation season two samples were taken in Fuente de los Tres Caños (25 July 25 and 8 August), so that Q^sub o^, Q^sub d^, and Q^sub g^ were obtained by EMMA only in those dates. Also, an EMMA based only on EC was performed to estimate Q^sub o^ and Q^sub d^ in D-14 assuming that Q^sub g^ was negligible by decomposing these two flow components for the mean of the nine 48-h surveys. The EC of Q^sub o^ was taken as the mean of the study period (0.38 dS m^sup -1^; Table 1), the EC of Q^sub d^ was taken as the mean EC of Q^sub d^ for all dates in the 1995 irrigation season (2.36 dS m^sup -1^), and the EC of the mixed water was the mean EC measured in the gully for that hour in the nine surveys. This analysis allowed us to establish the variations along the day of the diluted (Q^sub o^) and concentrated (Q^sub d^) waters.
1996 Irrigation Season
A record of Cl- in both Q^sub g^ and Q^sub d^ in all field surveys (about 15 d interval) was available for the period from 20 March to 16 October. Thus, the three end-member analysis could be performed, allowing for the estimation of Q^sub o^, Q^sub d^, and Q^sub g^ during the 1996 irrigation season. The EMMA was performed for each date with data and the proportions of each component were linearly interpolated between those dates to have daily estimates of Q^sub o^, Q^sub d^, and Q^sub g^. The EC of Q^sub o^ was taken as 0.38 dS m^sup -1^, and the EC values of Q^sub d^ and Q^sub g^ were those of the sampling date. The EC of the gully on that date was taken as the EC of the mixed water.
Salt Load Estimation and Components of the Salt Balance
The daily TDS were calculated from the daily EC through a linear regression equation fitted to 47 samples taken from May 1997 to February 1998. Daily salt-load estimates during the irrigation season were biased by the sampling hour (due to the daily cycles in flow and EC), so they were corrected by regression (Isidore et al., 2003). As there were no daily cycles out of the irrigation season, non-irrigation season daily load estimates were not corrected for bias.
The missing daily EC values during the irrigation season were obtained by linear interpolation between the previous and following days. Missing EC values during the non-irrigation season were also simply interpolated between adjacent EC values when only base flow was present. But, when two or more components contributed to the flow, EC was obtained from the flow components obtained by hydrograph separation into a surface flow (canal releases or surface runoff) and a subsurface flow as explained below. The subsurface flow could not be separated into Q^sub d^ and Q^sub g^ due to the lack of winter data for Q^sub g^.
The different flows in the system and their salt contribution were estimated for both seasons. Municipal wastewater was estimated as 80% of the water diverted for municipal uses, irrigation and precipitation were taken from the available information (Isidoro et al., 2004), and surface runoff and canal releases were determined from the hydrographs and the information given by the Ebro River Basin Authority.
Surface runoff and canal releases in the non-irrigation season were determined by separation of the hydrograph in two components based on the EC (Matsubayashi et al., 1993). The existence of a surface flow (canal releases or surface runoff) was deduced from the hydrograph. The release of water from Los Monegros Canal into the gully caused a "mesa" in the D-14 hydrograph (Isidoro et al., 2004) (the dates of the main releases were facilitated by the Ebro River Basin Authority, too), while surface runoff was revealed as a typical runoff peak. When both canal releases and surface runoff were present, flow was separated into surface and subsurface components, and canal releases and surface runoff were separated by drawing an arbitrary horizontal line through the assumed upper level of the mesa. During the irrigation season, canal releases were taken as the average values given by the Ebro River Basin Authority and surface runoff was taken as the rise of the hydrograph above the water level right before the rise (Isidoro et al., 2004).
During the irrigation season of 1996, the contribution of irrigated land to the salt load through D-14 was estimated subtracting from the gully salt load the contributions of canal releases, surface runoff, municipal wastewater, irrigation, precipitation, and ground water inflows (Q^sub g^) estimated by EMMA.
The salt inputs due to fertilization or dry deposition and the harvest outputs (Rhoades et al., 1992) were neglected along with the contribution of the dryland soils within the district.
RESULTSANDDISCUSSION
Quality of La Violada Gully Waters at D-14
Table 2. Chemical characteristics of La Violada Gully waters sampled at the D-14 monitoring station: mean, standard deviation (SD), maximum, minimum, and number of samples (n) for all the samples available in the whole study period (October 1994 to August 1998).
The ratio of TDS to EC (actually the coefficient of regression of the equation without a constant term) was 927 (mg L^sup -1^).(dS m^sup -1^)^sup -1^. Both coefficients of regression are well above the usual 640 (mg L^sup -1^).(dS/m)^sup -1^ given by USDA (1954) or 631 (mg L^sup -1^).(dS m^sup -1^)^sup -1^ found by Aragüés et al. (1986) for the rivers water of the Ebro River basin. This high TDS to EC ratio was due to the high concentration of the bivalent ions Ca^sup 2+^, Mg^sup 2+^, and SO^sub 4^^sup 2-^ in La Violada Gully water. The coefficients of variation (CV) of Ca^sup 2+^, Mg^sup 2+^, and SO^sub 4^^sup 2-^ are lower (about 20%) than those of Na+ and Cl- (about 30%) because Na+ and Cl- presented higher maximums relative to the mean in periods of high contribution of Q^sub g^. The variability of NO^sub 3^^sup -^ is even higher (CV = 38%) as it is affected by fertilization practices.
La Violada Gully waters are used for irrigation downstream of D-14. The salinity hazard of these waters is smaller than would be deducted from its EC (Rhoades et al., 1992). Their sodium adsorption ratio (SAR = Na/ [(Ca + Mg)/2]^sup 0.5^ with concentrations given in mmol^sub c^ L^sup -1^) is very low (0.53). Also, the relatively high NO^sub 3^^sup -^ concentration of these waters (39.1 mg L^sup -1^) reduces the need for nitrogen fertilization.
The high SO^sub 4^^sup 2-^ concentration of La Violada Gully waters may induce concrete corrosion, which is a relevant problem in La Violada Canal (Llamas, 1962). This problem is particularly significant for waters high in MgSO^sub 4^ and NO^sub 3^^sup -^ (Ayers and Westcot, 1985), such as those of La Violada Gully (Table 2).
End-Member Mixing Analysis
1995 Irrigation Season
The EMMA for the two dates with available data in the 1995 irrigation season (Fig. 3) showed that the EC-Cl- relationships of the D-14 waters can be explained by the mixing of Q^sub o^ and Q^sub d^. Most of the flow at D-14 was drainage originating in VID (Q^sub d^: 72% and 81% of the total flow in July and August, respectively), about one fifth was irrigation water (Q^sub o^: 25% and 18%, respectively), and the contribution of ground water inflows (Q^sub g^) was negligible (3% and 0.6%, respectively).
Fig. 3. Electrical conductivity (EC) and Cl- in 25 July and 8 Aug. 1995 for La Violada Gully water at the monitoring station (D-14) and its three end-members (ground water inflows, Q^sub g^; drainage waters from La Violada Irrigation District, Q^sub d^; and canal waters, Q^sub o^).
Thus, we ignored Q^sub g^ and separated the daily hydrographs into Q^sub o^ and Q^sub d^ based solely on their EC. The samples of the nine intensive surveys of the 1995 irrigation season (from 16-18 May 1995 to 19-21 Sept. 1995) were used for this separation (Fig. 4). The diluted component Q^sub o^ was regarded as bypass water entirely as there were no runoff inflows during the 1995 irrigation season (no apparent peaks in the hydrograph).
The mean EC values every 2 h of the nine 48-h surveys presented a daily oscillation with its peaks lagging 4 to 6 h after the minima of the instant flows (Fig. 4). If the daily EC patterns were due to a simple dilution of Q^sub d^ (subsurface flows with higher EC) by Q^sub o^ (surface or tail-water flows with lower EC), the maximum EC values should match the minimum flows. Instead, maximum EC values were delayed by about 4 h causing a clockwise hysteresis cycle in the daily EC-flow plot (Fig. 5). Simple dilution of Q^sub d^ with Q^sub o^ where one of the two components (Q^sub d^) was constant would result in a linear EC-flow relationship, showing no hysteresis.
Both Q^sub o^ and Q^sub d^ estimated for the mean of the 2-d surveys showed a clear daily wave (Fig. 4). While Q^sub o^ was well represented by a first-order harmonic, a second-order harmonic was significant for the Q^sub d^ daily cycle (Fig. 6). The high Q^sub o^ from 1200 to 1600 h raised the water level in the main ditches and in La Violada Gully at D-14 (hampering the drainage of the lower plots and tile drains) and thus, caused the decrease in Q^sub d^ observed during the afternoon (Fig. 6). In the afternoon hours, the water level along Valsalada Ditch was often higher than the outlets of the pipe drains into the ditch. The afternoon peaks in Q^sub o^ were caused by direct releases of water from the irrigation ditches to the drainage network after the completion of irrigation operations by farmers (bypass water). This limited drainage leads to salt-affected soils and high water tables in the more depressed fields of VID close to the main ditches (Torres, 1983).
Fig. 4. Stacked area chart showing surface (Q^sub o^) and subsurface (Q^sub d^) flows estimated by hydrograph separation at the D-14 monitoring station of La Violada Gully and electrical conductivity (EC) of the nine 48-h surveys performed in the 1995 irrigation season (from 16-18 May to 19-21 September).
Fig. 5. Mean bihourly values of electrical conductivity (EC) versus instant flow (Q^sub i^) for the nine intensive surveys performed in the 1995 irrigation season.
1996 Irrigation Season
The EC and Cl- of the three components in the samples of the 1996 irrigation season are presented in Fig. 7. Some D-14 samples could be well explained by the mixing of only two components Q^sub o^ and Q^sub d^, but others (particularly in the beginning of the irrigation season) showed a Cl- that required the presence of a Q^sub g^ component as high as 23%. The EMMA performed for the 1996 irrigation season showed that ground water inflows into VID (Q^sub g^) were relatively high in April and tended to decrease thereafter (Fig. 8). The high precipitation at the end of July and beginning of August did not increase Q^sub g^ substantially, indicating that precipitation was lower than ET^sub o^ in the dryland watershed. Calculations based on Eq. [1-3] showed that Q^sub g^ was 6.5% of the total flow at D-14 during the 1996 irrigation season, ranging from 23% in April to 1% in August.
Fig. 6. Harmonic series fitted to Q^sub d^ (open squares) (subsurface drainage flows) and Q^sub o^ (filled squares) (surface flows). Bars show the standard error of the differences between Q^sub d^ and Q^sub o^ instant values and their respective daily means.
Fig. 7. Electrical conductivity (EC) and Cl in the end-members considered: canal water (Q^sub o^), drainage water from La Violada Irrigation District (Q^sub d^), and ground water inflows (Q^sub g^); and in La Violada Gully water (D-14) used for the end-member mixing analysis (EMMA) during the irrigation season 1996.
The higher Q^sub g^ in the beginning of the season (April, 23%) is thought to come from the base flow originated by the high rains of the 1995-1996 winter (330.5 mm from January to April 1996). The proportion of Q^sub g^ reduced along the season as the dryland area was getting drained and delivering less water until it became almost nil in August. The rains in July and August did not increase the volume of Q^sub g^ appreciably (though they produced clear runoff peaks) possibly because the dryland area of the basin was too dry during the summer to generate any appreciable ground water flow (Fig. 8).
The proportions of the other flow components (especially Q^sub d^) were also different in the beginning of the irrigation season (in April Q^sub g^ = 23%, Q^sub d^ = 48%, Q^sub o^ = 29%) to the rest of the irrigation season (Q^sub g^ = 4%, Q^sub d^ = 72%, Q^sub o^ = 24%). Most of the irrigation in April was applied to help the emergence of corn. These irrigations replenished the initially drier soils, leading to less Q^sub d^ than later in the season.
Fig. 8. Stacked area chart showing surface flows (Q^sub o^), subsurface drainage flows (Q^sub d^), and ground water inflows (O^sub g^) along the 1996 irrigation season identified in La Violada Gully at the D-14 monitoring station through end-member mixing analysis (EMMA). The daily precipitation during this period is also shown.
Salt Exports through D-14
The annual salt load in La Violada Gully at D-14 during the 1995 to 1998 hydrological years ranged from 65135 Mg (1995) to 89596 Mg (1997), with an average of 78628 Mg, and with a relatively low CV of 13% (Table 3). The average salt load exported during the irrigation season (55887 Mg) was 71% of the annual load, and its CV was very low (4%).
The average salt yield of the irrigated land (i.e., salt load per hectare of irrigated land in VID) during the irrigation season was 14.5 Mg ha^sup -1^, with a range between 13.6 Mg ha^sup -1^ (1995) and 14.9 Mg ha^sup -1^ (1997). The average salt yield of the irrigated land for the hydrological year was 20.3 Mg ha^sup -1^, similar to the average salt yield of 18.8 Mg ha^sup -1^ reported by Faci et al. (1985) for VID in the 1982 and 1983 hydrological years. The salt yield in La Violada Irrigation District has remained stable during this 14-yr period, probably because the main source of salts is gypsum dissolution, and water management and the crops grown in VID have not changed. This salt yield of about 20 Mg ha^sup -1^ yr^sup -1^ for VID is almost 50% higher than the value of 13.5 Mg ha^sup -1^ yr^sup -1^ reported by Tedeschi et al. (2001) for a hydrological basin in Monegros II sprinkler irrigated with a high efficiency (SIPI = 92%) and with high-saline lutites present in the subsoil. The mean annual solute export (78628 Mg) meant a lumped salt yield of 4.0 Mg ha^sup -1^ for the whole watershed (Table 3), reflecting the lower contribution of the dryland area to salt load.
The mean monthly values of flow, salt load, and flow-weighted TDS of La Violada Gully waters at D-14 are shown in Fig. 9 for the 1995 to 1998 hydrological years. The TDS was rather constant along the study period (flow-weighted mean = 1720 mg L^sup -1^, CV = 16%), except between January to March 1995 when large releases from the Monegros Canal into the gully significantly reduced salt concentrations. If the TDS during this atypical period was deleted, the weighted mean TDS was 1751 mg L^sup -1^, with a lower CV of 9%.
Table 3. Salt loads exported with La Violada Gully waters at D-14 during the irrigation season, the non-irrigation season, and the total annual of hydrological years 1995 to 1998. Salt yields resulting for the irrigated area (3866 ha) of La Violada Irrigation District (VID) for the irrigation seasons and for the hydrological years and salt yield for the whole D-14 watershed (19637 ha) for the hydrological years.
The lowest TDS values were found during the irrigation seasons due to dilution of the gully waters by spills from the irrigation ditches, canal seepage, and operational losses, whereas the highest TDS values were found in winter, when the gully flow is dominated by base-flows of relatively high salt concentrations.
Based on the EMMA performed for the 1996 irrigation season, the salt input attributed to Q^sub o^ (CR and SR in Table 4: 207 Mg) was negligible, and the salt input attributed to Q^sub g^ (GI in Table 4: 4475 Mg) was only 8% of the salt mass exported through D-14 (Table 4: 56554 Mg). Salt inputs in precipitation and municipal wastewaters (MW) were also negligible, whereas salt inputs in irrigation (Table 4: 8380 Mg) were 15% of salt exports at D-14. The salt balance or difference between the total salt outputs and inputs in VID (salt balance in Table 4: 43281 Mg) was due to gypsum dissolution and leaching of dissolved salts with the drainage waters originated in VID. This meant a salt yield of 11.2 Mg ha^sup -1^. Therefore, 77% of the salts exported by La Violada Gully during the irrigation season originated in VID. During the non-irrigation season the salt contribution of irrigation, canal releases, surface runoff, and municipal wastewaters was negligible (6.3% of the salt load in the gully altogether) and the salt load of the gully was almost entirely of subsurface flow originating from VID and the dryland reaches of the basin, which could not be separated (Table 4).
The salt balance calculated in the 1996 irrigation season was extended to the rest of the years assuming that the ratio Q^sub g^ to Q was the same for each month of the season. For the 1995-1998 average irrigation seasons the major salt inputs to the district were irrigation (8204 Mg) and ground water inflows (4195 Mg), whereas the contribution of the rest of components was negligible due to their low salt concentrations and/or flows (Table 4). The average mass of salts originating from VID in the period 1995-1998 was 43015 Mg or 77% of the total load exported by La Violada Gully. Therefore, irrigated agriculture in VID was the main source of salts in this gully, yielding an irrigation season average value of 11.1 Mg of salts per hectare of irrigated land.
Fig. 9. Mean monthly values of flow (O), salt load (L), and flow-weighted total dissolved solids (TDS) of La Violada Gully waters at the D-14 monitoring station in the 1995 to 1998 hydrological years. The irrigation seasons are shaded.
This high salt yield was attributed to the poor management of water in VID (on-farm irrigation efficiency = 62%), which caused the dissolution of the gypsum extensively present in the soils of this irrigation district. Since the mass of salts originating from gypsum dissolution are proportional to the flow of water through the soil, the key factor for minimizing off-site salt pollution in VID is an increase in irrigation efficiency through better irrigation management and the change to more efficient pressurized irrigation systems.
Table 4. Salt mass balance (salt balance = outputs - inputs) for La Violada Irrigation District (VID) in the 1995 to 1998 irrigation seasons (IS) and salt inputs and outputs in the nonirrigation seasons (NIS). Inputs: irrigation (I), precipitation (P), canal releases (CR), surface runoff (SR), ground water inflows (GI), and municipal wastewaters (MW). Outputs: La Violada Gully waters at L)-14. The salt yield is the salt balance divided by the irrigated surface in VID (3866 ha).
CONCLUSIONS
La Violada Gully waters result from the mixing of three waters identified through end-member mixing analysis: bypass waters from the irrigation ditches and direct canal releases into the gully (Q^sub o^), drainage waters originating from the La Violada Irrigation District (VID) (Q^sub d^), and ground water inflows into VID (Q^sub g^) originating from the upper dryland watershed.
Since gypsum is preponderant in the soils of the study area, La Violada Gully waters were characterized by high SO^sub 4^^sup 2-^ and Ca^sup 2+^ concentrations and EC values close to gypsum saturation (mean EC = 1.9 dS m^sup -1^). The Q^sub o^ waters were low in salts (mean EC = 0.4 dS m^sup -1^), whereas the Q^sub d^ and Q^sub g^ waters had higher salt concentrations (mean EC = 2.4 and 2.7 dS m^sup -1^, respectively).
The EC of La Violada Gully waters presented a daily cyclic oscillation with a clear lag in relation to its instant flow, causing a clockwise hysteresis cycle in the daily EC-instant flow plots. This cycle in the instantaneous flow and its components Q^sub o^ and Q^sub d^, found in the 1995 irrigation season, suggested that the rise in the water level induced by high bypass flows to the main drains hindered the subsurface drainage in the afternoon. Reducing this direct discharge from the irrigation ditches could ameliorate the drainage conditions of the lower irrigated land, facilitating salt leaching. Reducing bypass from irrigation ditches also means better efficiency of the irrigation system. But, this bypass also reduces the overall salinity of La Violada Gully, so a decrease in the bypass water will increase the salinity in the gully.
The EMMA performed in the 1995 irrigation season showed that Q^sub d^ was the principal component of La Violada Gully flow (around 77% of the total flow), whereas the contribution of Q^sub o^ was around 22% and that of Q^sub g^ was negligible (around 1%). In contrast, the analysis performed in the 1996 irrigation season, where precipitation was higher than in 1995, showed that Q^sub g^ was 6.5 % of the total flow in La Violada Gully.
The 1995 to 1998 average annual salt load exported by La Violada Gully waters was 78 628 Mg. About 71% of the annual salt load was exported over the irrigation season of the four study years. The corresponding average salt yields (i.e., loads per hectare of irrigated land in VID) were 20.3 Mg ha^sup -1^ (hydrological year) and 14.5 Mg ha^sup -1^ (irrigation season). Due to the fairly constant TDS values of La Violada Gully waters (flowweighted mean TDS = 1720 mg L^sup -1^), the monthly salt loads and the monthly flows paralleled each other and the loads could be significantly estimated (P < 0.001) from the flows.
The salt mass balance performed in VID for the 1995 to 1998 irrigation season shows that the average mass of salts exported by the drainage waters of VID (Q^sub d^) was 77% of the total load in La Violada Gully. Therefore, irrigated agriculture in VID was the main source of salts in this gully, yielding an irrigation season average value of 11.1 Mg ha^sup -1^. This salt yield is higher than those reported in other areas of the Ebro River basin due to the low irrigation efficiency (IE = 62%) and the gypsumrich soils (>3%) typical of this irrigation district.
La Violada Gully water could be used to benefit from its particular properties. Its high Ca^sup 2+^ and Mg^sup 2+^ concentrations make it appropriate for the irrigation of sodicity-affected soils (an acute problem in some neighboring areas) to help their reclamation. The high NO^sub 3^^sup -^ of La Violada water could reduce the amount of N fertilizer needed on lands irrigated with this water downstream of the district.
ACKNOWLEDGMENTS
This study was funded by the Spanish Institute of Agricultural Research and Technology (INIA). The authors wish to thank the Almudévar Irrigation Council (Comunidad de Regantes de Almudévar) and the Ebro River Basin Authority (Confederación Hidrográfica del Ebro) for their cooperation. We also thank the thorough review work made by the editors and the reviewers. A Fulbright Grant and the financial sponsorship of the Spanish Ministry of Education supported Daniel Isidore.
REFERENCES
Alberto, F., and R. Aragüés. 1986. Curvas de tendencia salinidadtiempo en las aguas superficiales de la cuenca del Ebro. (In Spanish.) p. 237-251. In Sistema integrado del Ebro, estudio interdisciplinar. Convenio de cooperación cientifico-técnico Hispano-Norteamericano, Madrid.
Alberto, F., J. Machín, and R. Aragüés. 1986. La problemática general de la salinidad en la cuenca del Ebro. (In Spanish.) p. 221-236. In Sistema integrado del Ebro, estudio interdisciplinar. Convenio de cooperación cientifico-técnico Hispano-Norteamericano, Madrid.
Aragüés, R., F. Alberto, and D. Quílez. 1986. Relaciones entre los parámetros químicos de las aguas superficiales de la cuenca del Ebro. (In Spanish.) p. 253-277. In Sistema integrado del Ebro, estudio interdisciplinar. Convenio de cooperación cientifico-técnico Hispano-Norteamericano, Madrid.
Aragüés, R., and K.K. Tanji. 2003. Water quality of irrigation return flows, p. 502-506. In B.A. Stewart and T.A. Howell (ed.) Encyclopedia of water science. Marcel Dekker, New York.
Ayers, R.S., and D.W. Westcot. 1985. Water quality for agriculture. FAO Irrigation and Drainage Paper 29. Rev. 1. FAO, Rome.
Bellot, J., F. Golley, and M.T. Aguinaco. 1989. Environmental consequences of salts exports from an irrigated landscape in the Ebro River basin, Spain. Agric. Ecosyst. Environ. 27(1-4):131-138.
Confederación Hidrográfica del Ebro. 2004a. Regadíos y usos agrarios. Available at www.chebro.es/usos/regadios.htm (verified 3 Feb. 2006). Confederación Hidrográfica del Ebro, Zaragoza, Spain.
Confederación Hidrográfica del Ebro. 2004b. Calidad de las aguas. Available at http://oph.chebro.es/DOCUMENTACION/Calidad/CalidadDeAguas.html (verified 3 Feb. 2006). Confederación Hidrográfica del Ebro, Zaragoza, Spain.
Durand, P., and J.L. Juan Torres. 1996. Solute transfer in agricultural catchments: The interest and limits of mixing models. J. Hydrol. 181:1-22.
El-Ashry, M.T, J. van Schilfgaarde, and S. Schiffman. 1985. Salinity pollution from irrigated agriculture. J. Soil Water Conserv. 40:48-52.
Faci, J., R. Aragüés, F. Alberto, D. Quflez, J. Machin, and J.L. Arrue. 1985. Water and salt balance in an irrigated area of the Ebro River basin (Spain). Irrig. Sci. 6(l):29-37.
Faci, J.M., A. Bensaci, A. Slatni, and E. Palyán. 2000. A case study for irrigation modernization I. Characterization of the district and analysis of water delivery records. Agric. Water Manage. 42:313-334.
Instituto Tecnológico Geominero de España. 1995. Mapa geológico de España escala 1:50000. Almudévar. (In Spanish.) ITGE, Madrid.
Isidoro, D., D. Quílez, and R. Aragüés. 2003. Sampling strategies for the estimation of salt and nitrate loads in irrigation return flows: La Violada Gully (Spain) as a case study. J. Hydrol. 271(l-4):39-51.
Isidoro, D.. D. Quflez, and R. Aragüés. 2004. Water balance and irrigation performance analysis: La Violada Irrigation District (Spain) as a case study. Agric. Water Manage. 64:123-142.
Isidoro, D., D. Quílez, and R. Aragüés. 2006. Environmental impact of irrigation in La Violada district (Spain): II. Nitrogen fertilization and nitrate export patterns. J. Environ. Qual. XX:XXX-XXX (this issue).
Llamas, R. 1962. Estudio geológico-geotécnico de los terrenos yesíferos de la cuenca del Ebro y los problemas que plantean en los canales. (In Spanish.) Bol. Serv. Geol. Obras Públicas 12. Ministerio de Obras Públicas, Madrid.
Matsubayashi, V, G.T. Velasquez, and F. Takagi. 1993. Hydrograph separation and flow analysis by specific electrical conductivity. J. Hydrol. 152:179-199.
Mulholland, PJ. 1993. Hydrometric and stream chemistry evidence of three storm flowpaths in Walker Branch Watershed. J. Hydrol. 151:291-316.
Finder, G.F., and J.F. Jones. 1969. Determination of the ground-water component of the peak discharge from the chemistry of total runoff. Water Resour. Res. 5(2):438-445.
Playán, E., A. Slatni, R. Castillo, and J.M. Faci. 2000. A case study for irrigation modernization II. Scenario analysis. Agric. Water Manage. 42:335-354.
Quílez, D., R. Aragüés, and K.K. Tanji. 1992. Salinity of rivers: Transfer function-noise approach. J. Irrig. Drain. Eng. 118(3):343-359.
Rhoades, J.D., A. Kandiah, and A.M. Mashali. 1992. The use of saline waters for crop production. FAO Irrigation and Drainage Paper 48. FAO, Rome.
Tedeschi, A., A. Beltrán, and R. Aragüés. 2001. Irrigation management and hydrosalinity balance in a semi-arid area of the middle Ebro river basin (Spain). Agric. Water Manage. 49:31-50.
Torres, M. 1983. Balance hidrosalino de un polígono de riego en los Llanos de La Violada (Huesca). (In Spanish.) M.S. thesis. Instituto Agronómico Mediterráneo de Zaragoza (CIHEAM-IAMZ), Zaragoza, Spain.
Toth, J. 1963. A theoretical analysis of groundwater flow in small drainage basins. J. Geophys. Res. 68(16):4795-4812.
USDA. 1954. Diagnosis and improvement of saline and alkali soils. United States Salinity Laboratory, Agric. Handbook 60. USDA, U.S. Gov. Print. Office, Washington, DC.
van Schilfgaarde, J. 1994. Irrigation -A blessing or a curse. Agric. Water Manage. 25:203-219.
D. Isidoro,* D. Quílez, and R. Aragüés
Unidad de Suelos y Riegos, Centro de Investigatión y Tecnología Agroalimentaria (CITA-DGA) and Laboratorio de Agronomía y Medio Ambiente (DGA-CSIC), Apdo. 727, 50080-Zaragoza, Spain. Received 21 Feb. 2005. * Corresponding author ([email protected]).
Published in J. Environ. Quai. 35:766-775 (2006).
Technical Reports: Surface Water Quality
doi:10.2134/jeq2005.0064
© ASA, CSSA, SSSA
677 S. Segoe Rd., Madison, WI 53711 USA
Copyright American Society of Agronomy May/Jun 2006