Salinity Effects from Evaporation and Transpiration under Flood Irrigation

The four irrigation study sites selected for this study lie within the interdunal flats near the townships of Padthaway and Keith in Upper Southeast South Australia (Fig. 1). Pasture, clover, and lucerne crops are flood irrigated at these sites. The climate within the study areas can be characterized by warm-to-hot dry summers and cool-wet winters. The average annual maximum temperature is 22°C, with February being the hottest month at 29.8°C and July being the coldest month at 5.5°C. Annual potential evaporation is $1,600\mathrm{mm}/\mathrm{year}$ and $1,700\mathrm{mm}/\mathrm{year}$ for Padthaway and Keith, respectively.

Soil particle size (% sand, silt, and clay), bulk density, water content, and soil-water chloride was previously determined at each site from soil cores (see Harrington et al. 2004; Wohling 2007). Soil cores were collected at 50-cm-deep increments from either excavation pits or during drilling. Based on the particle-size distribution, the soil classification for the topsoil ranges from loam at Padthaway to sand in the Hundred of Stirling. Each site exhibits one or more calcrete layers that were encountered at depths of $<0.50\mathrm{m}$ at Padthaway and $<1\mathrm{m}$ at the Hundred of Stirling. The texture of most upper parts of the Padthaway Formation resembles a weathered marly clay.

Field measurements were made in four flood irrigation bays consisting of pasture, clover, and lucerne. Collection of water samples from irrigation water, soil-water (suction lysimeters), and groundwater took place during two irrigation events at each site, which occurred within the 2005/2006 irrigation season spanning from November 2005 to March 2006. The ponded water samples were collected at five evenly distributed places in the flood bay (stations labeled A to E, Fig. 2) to detect isotopic change during surface water movement. Controlled, Class A pan evaporation experiments were also conducted in parallel to monitor the evolution of chloride and $\delta {}^2\mathrm{H}$ and $\delta {}^{18}\mathrm{O}$ concentrations over time as a result of potential evaporation. Fig. 2 shows the sampling locations at a typical instrumentation site, consisting of capacitance probes, suction lysimeters, and piezometers.

All water samples were analyzed for $\delta {}^2\mathrm{H}$ and $\delta {}^{18}\mathrm{O}$ concentration, total dissolved solids (TDS) ($\mathrm{mg}/\mathrm{L}$), and chloride (${\mathrm{Cl}}^{-}$ $\mathrm{mg}/\mathrm{L}$). A measurement of TDS was used to give an overall assessment of the salinity, while ${\mathrm{Cl}}^{-}$ helps distinguish between increases in salinity caused by evapotranspiration and soil-water interaction. The measurement of $\delta {}^2\mathrm{H}$ and $\delta {}^{18}\mathrm{O}$ allows us to determine evaporation ($E$ only), while the buildup of ${\mathrm{Cl}}^{-}$ along with decrease in soil-water content enables an assessment of the total evapotranspiration flux ($E+T$).

Groundwater samples for $\delta {}^2\mathrm{H}$ and $\delta {}^{18}\mathrm{O}$ were analyzed by the Commonwealth Scientific and Industrial Research Organization (CSIRO) isotopic laboratory in Adelaide by using a Europa Scientific Ltd. GEO 20-20 dual inlet gas ratio mass spectrometer. Results are expressed as $\delta {}^{18}\mathrm{O}$ (${}^{18}\mathrm{O}/{}^{16}\mathrm{O}$) in per mil (‰) as a deviation from the Vienna Standard Mean Ocean Water ($V$-SMOW). The overall precisions of the $\delta {}^{18}\mathrm{O}$ and $\delta {}^2\mathrm{H}$ analysis are $\pm 0.1‰$ and $\pm 1‰$, respectively.

Capacitance probes ($C$-Probes) were installed to measure the dielectric constant of the soil and therefore the water content by the capacitance method. They represent the best experimental in situ techniques for monitoring root activity (plant water use) and soil drainage. While the water-content measurements provide an indication of advancing wetting fronts, they cannot resolve the difference between large and small drainage fluxes.

Holes were drilled (50 mm diameter) to depths of 3 m via air hammer techniques to accommodate the $C$-probe. Soil capacitance is measured via sensors positioned at nominal depths of 10, 20, 30, 50, 100, 150, 200, 250, and 300 cm within the vadose zone. Sensors were located both within the topsoil horizon and underlying Padthaway Formation. $C$-probes were used to determine lag times and the extent of water movement through the profile after an irrigation event and were used to map and monitor crop water uptake (root activity). The $C$-probes utilized a telemetry system to log and transmit data every 15 min. At the Padthaway irrigation sites, two $C$-probes were installed within the irrigation bays; and at the Hundred of Stirling irrigation sites, one $C$-probe was installed in the middle of the bay.

Suction lysimeters were installed to measure soil-water salinity (and chloride) and isotopic ratios of hydrogen ($\delta {}^2\mathrm{H}/\delta {}^1\mathrm{H}$) and oxygen ($\delta {}^{18}\mathrm{O}/\delta {}^{16}\mathrm{O}$) within the vadose zone over time. The chloride and isotopic concentrations measured from extracted soil-water reflect the chloride and isotopic concentrations of drainage water that would eventually recharge the unconfined aquifer.

At each site, four 100-mm-diameter holes were drilled via air hammer techniques within the unsaturated zone to nominal depths ranging from 0.3 to 4.0 m (depending upon rooting depth and soil structure) and equipped with suction lysimeters installed in the bottom 5 cm of the hole. The top lysimeter was located in the topsoil and the bottom three lysimeters were position within the unsaturated zone of the Padthaway Formation. The lysimeters were constructed by attaching a 15 cm porous ceramic cup to the end of 16-mm-diameter PVC conduit. These were placed in the hole with the ceramic cup and surrounded by diatomaceous earth to provide a good contact with the surrounding soil. A bentonite seal was placed above the diatomaceous earth and the hole was cemented to the surface. Two groups of four lysimeters were installed at each of the two flood irrigation sites (NAP4 and NAP5) in Padthaway to achieve average readings across the bay. Soil-water samples were extracted via a vacuum pump from suction lysimeters after irrigation at a time when the soil profile appeared to be approaching field capacity, as determined by the capacitance response. Because drainage can continue to occur for a number of days after irrigation, the changes in the isotopic signature of the wetting front as it moves through the vadose zone were monitored over time during the second round of sampling. Using the capacitance response as a guide, the suction lysimeters at NAP4 and NAP5 were subsequently sampled every 2 to 3 days after irrigation. Repeat sampling was not possible at sites PG and MTM because sufficient amounts of soil-water required for analysis could not be obtained because of the sandier soil, lower soil-water retention, and consequently lower water contents after irrigation.

Piezometers were installed 3 to 10 m below the water table in the middle of each bay. All piezometers were constructed from 50-mm-diameter Class 12 PVC pipe with slotted screens just below the water table. $\delta {}^2\mathrm{H}$ $\delta {}^{18}\mathrm{O}$, EC, and chloride concentrations were measured in groundwater, sampled from the piezometers 1–2 days after irrigation. A whaler pump was used to pump groundwater until three bore volumes had been purged.

The capacitance response from sensors located at various depths throughout the vadose zone show rapid drainage to the water table following irrigation (Fig. 3). The high capacitance responses from the sensor located in the topsoil at NAP4 is indicative of ponded irrigation water, which ponded for up to 24 h after the initial irrigation. The capacitance sensors show evidence of water loss either by root activity or evaporation to depths of 0.20 and 0.30 m. This depth corresponds to the depth of the topsoil and therefore the extent and bulk of the root zone. This is underlain by a calcrete-topped limestone. A change in the advancement rate of the wetting front was observed below this depth because water drains through the underlying limestone. Sensors positioned in the limestone indicate that postirrigation, drainage continued to occur for up to 60 h at NAP4 and 16 h at NAP5, which represents a time when the profile approaches field capacity. Because of the heterogeneous nature of the unsaturated zone, preferential flow through cracks and cavities (karstic features) may not be accounted for or detected using the capacitance response. The results confirm that water loss via evaporation and transpiration is likely to be constrained to the topsoil (upper 0.3 m) above the calcrete layer.

Vertical distribution of soil-water ${\mathrm{Cl}}^{-}$ measured from suction lysimeters postirrigation remained uniform with depth as a result of the high volume of irrigation water applied and high drainage (Fig. 4). However, long-term soil-water ${\mathrm{Cl}}^{-}$ data collected monthly at these sites from 2003 to 2006 show variations in the upper part of the profile over the longer term, which was attributed to evapotranspiration from the top portion of the soil profile.

Evaporation from the soil following irrigation at NAP4 and NAP5 was small and seemed only to affect the soil-water isotope values in the top 0.30 m ($\delta {}^{18}\mathrm{O}$ enrichment was 0.25‰ to 0.45‰ and $\delta {}^2\mathrm{H}$ enrichment was $<5‰$), whereas the isotope values below this depth remained steady over time (Fig. 4). The lack of isotopic enrichment below this depth suggests that (1) evaporation may be inhibited by a calcrete layer, commonly found at shallow depths (0.3 m), or (2) rapid infiltration of irrigation water via large cracks and channels coupled with a large reservoir of relatively immobile soil-water, owing to high marl content of the Padthaway Formation.

When there is no crop cover, such as during December irrigation at MTM, evaporation from saturated soil surfaces can be high from ponded water (Jensen et al. 1990). However, as most of the drainage occurred during the night (after 5:30 p.m.), evaporation from the soil was not apparent in the isotopic signatures of soil-water, which were collected the following day. At all other sites, evaporation from the soil was inhibited by dense crop cover and the calcrete layer at shallow depth (0.3 m).

Landon et al. (2000) showed that soil-water obtained from suction lysimeters may not be representative of drainage water (mobile/gravity), because differences in the isotopic values of soil-water collected using suction lysimeters, wick samplers, and core samples were found to occur because these methods collect different fractions of the total soil-water reservoir. They showed that wick samplers collect primarily mobile, gravity drainage water that is in excess of soil field capacity. Suction lysimeters collect a mixture of immobile water that is bound to the soil matrix at a tension of less than approximately 35 kPa and mobile water that is present in excess of field capacity at the time when suction is applied.

It can be postulated that because of the large volume of irrigation water applied here, the large amounts of soil-water encountered in the suction cup directly after irrigation consists mostly of irrigation water. When extracting soil-water from the suction lysimeters, only minor suction was required over a short time to obtain sufficient volumes of soil-water, suggesting that the soil-water extracted mostly consisted of mobile water (because mobile water is expected to be drawn into the cup before immobile connate water from the soil matrix).

$\delta {}^{18}\mathrm{O}$ versus salinity plots of soil-water, irrigation water, and groundwater are displayed in Fig. 5 for each site. $\delta {}^2\mathrm{H}$ versus salinity plots are not shown here, but showed a similar trend. Table 1 shows (1) the average increase in chloride concentration and enrichment of irrigation water relative to the source (irrigation bore) and (2) the average increase in chloride concentration and enrichment of soil-water relative to ponded irrigation waters. Table 1 and Fig. 5 show the comparison between the salinity effects as a result of evaporation (fractionating water loss) and transpiration from both irrigation water and soil-water. The percentage increase of chloride concentration as a result of fractionating water loss (evaporation) in irrigation water is small (0–5%) in comparison to the percentage increase of chloride concentration in soil-water (23%–117%). The low fractionating water loss detected in soil-water and large increase in chloride concentrations suggest transpiration is the dominant process across all sites.

The $\delta {}^{18}\mathrm{O}$ and $\delta {}^2\mathrm{H}$ composition of soil-water extracted from the suction lysimeters buried at each end of the irrigation bay at NAP4 and NAP5 are similar and plot close to the irrigation bore water (on the $x$-axis), signifying minor enrichment postirrigation (Fig. 5). This minor enrichment suggests that most of the irrigation drainage that recharges the aquifer has undergone a small amount of evaporation. Therefore, the isotopic enrichment of these evaporated waters was not reflected in isotopic signatures in soil-water.

During irrigation, the increase in chloride concentration of irrigation waters at NAP4 and NAP5 ranged from $30\mathrm{mg}/\mathrm{l}$ to $60\mathrm{mg}/\mathrm{l}$ (4.7%–9.5%) and $9\mathrm{mg}/\mathrm{l}$ to $35\mathrm{mg}/\mathrm{l}$ (0.75%–2.5%), respectively. However, the chloride concentration of soil-water extracted 1 to 3 days postirrigation was much more than that of the irrigation water, showing increases of $170\mathrm{mg}/\mathrm{l}$ and $189\mathrm{mg}/\mathrm{l}$ (26.5%–29.5%) and $274\mathrm{mg}/\mathrm{l}$ to $377\mathrm{mg}/\mathrm{l}$ (23%–31%) at NAP4 and NAP5, respectively. In both cases, the minor fractionating water loss ($<0.1‰$ for $\delta {}^{18}\mathrm{O}$) and the higher chloride concentration of soil-water suggest that transpiration was the dominant process at these sites.

The greater effect of evaporation on the open irrigation water, sampled during the December 2005 irrigation at MTM, is clearly evident by the greater spread of data points (exhibiting greater fractionation), which plot further toward the right across the $x$-axis than observed during the February irrigation, when crop cover was $\sim 95\%$ (Fig. 5). The corresponding increase in chloride concentration of irrigation water as a result of evaporation (fractionating water loss) was greater during the December irrigation ($129\mathrm{mg}/\mathrm{l}$, a 5% increase), than the February irrigation ($<1\%$), where chloride concentrations remained relatively unchanged. Because of the high crop cover in February, irrigation waters were not subjected to the same amount of evaporation as those measured in the evaporation pan over the same period (Fig. 5).

As observed across all sites, the isotopic enrichment of soil-water collected 1 to 3 days postirrigation at MTM was minor (0.5‰) and reflected partially evaporated irrigation water. The increase in chloride concentration as a result of fractionating water loss was minor (0 to $+129\mathrm{mg}/\mathrm{L}$, a 0–5% increase) compared to the increase in chloride concentration as result of transpiration ($+2,070$ to $3,070\mathrm{mg}/\mathrm{L}$, 79% to 118%).

The reduced influence of evaporation owing to dense crop cover was also confirmed by experiments conducted at irrigation site PG. Crop cover was close to maximum cover during both sampled irrigations, during which time the enrichment in $\delta {}^{18}\mathrm{O}$ ($<0.15‰$) and increase chloride concentrations (0 to $+30\mathrm{mg}/\mathrm{l}$, a 0–0.82% increase) of irrigation water were much less than the increases measured in evaporating pan water ($\delta {}^{18}\mathrm{O}$ enrichment was 0.66 ‰, ${\mathrm{Cl}}^{-}$ $+44\mathrm{mg}/\mathrm{l}$ to $+90\mathrm{mg}/\mathrm{l}$) over the same time period (Fig. 5).

During the January sampling event, some minor enrichment was detected in the soil-water ($<1‰$ $\delta {}^{18}\mathrm{O}$ and 1–3 ‰ $\delta {}^2\mathrm{H}$), which was equivalent to the enrichment of pan waters (1‰ $\delta {}^{18}\mathrm{O}$). The increase in chloride concentration of soil-water ($+1,968\mathrm{mg}/\mathrm{l}$ to $+2,100\mathrm{mg}/\mathrm{l}$, 54–58%), which was subject to both $E+T$, was much greater than the increase in chloride of pan water ($+90\mathrm{mg}/\mathrm{l}$), which was subject to $E$ only (Fig. 5). This suggests that the concentration of salt in the soil-water was dominated by transpiration at this site.

Wetting front movement along with soil isotopic and salinity values were measured 2 and 4 days following irrigation application at NAP4 and NAP5. The results are shown in Figs. 6(a, b): changes in water content versus soil-water chloride and Figs. 6(c, d): soil-water content versus soil-water $\delta {}^2\mathrm{H}/\delta {}^{18}\mathrm{O}$. Figs. 6(a, b) show that the isotopic composition of soil-water remains reasonably steady over time, while Figs. 6(c, d) show an increase in soil-water salinity with decreasing soil-water content over the same period. Therefore, this decrease in soil-water content and increase in soil-water salinity can only be explained by transpiration. The increase in soil-water chloride was mainly constrained to the top 0.30 m, signifying the extent and effect of evapotranspiration, which may be constrained by root activity and the calcrete layer commonly found at this depth. Below this depth, only minor changes in salinity and isotopic composition were detected over time (as shown in Fig. 4).

The salinity effect caused by the recycling of irrigation water has been assessed at each flood irrigation site. A net salinity effect to the unconfined aquifer can be calculated using drainage rate and drainage water salinity estimates. Unsaturated zone drainage rates ($D$ $\mathrm{mm}/\mathrm{year}$) were estimated using the water balance approach [Eq. (1)]where $P$ = precipitation (mm), $I$ = volume of irrigation water applied to the bay (mm), $D$ = drainage below the irrigation bay (mm), $ET$ = evapotranspiration (mm), and $\Delta S$ = change in soil-water content (mm), which because of the large volume of irrigation water applied, can be assumed to be negligible.

The salinity of drainage water under flood irrigation is assumably equivalent to that of the soil pore water salinity below the root zone, which is sampled at the two and three meter suction lysimeters. A salinity increase ($\Delta \mathrm{sal}$, $\mathrm{mg}/{\mathrm{L}}^{-1}$) because of the use of groundwater for irrigation is calculated as the difference between the estimated salinity of drainage water and the irrigation water applied (Harrington et al. 2004).

The net salinity effect to the aquifer ($\mathrm{t}/\mathrm{ha}/\mathrm{year}$) from the evaporation of irrigation water and evapoconcentration of soil-water is then given bywhere $I$ = volume of irrigation water applied to the bay, $D$ = drainage below the irrigation bay, $\Delta {\mathrm{sal}}_{sw}$ = net increase of salinity of drainage water (obtained from suction lysimeters below 0.5 m) minus salinity of irrigation water, and $\Delta {\mathrm{sal}}_{IW}$ = net increase in salinity of irrigation water during the ponding period. The salt balance and net salinity effects because of transpiration and evaporation of surface waters for each site are compared in Table 2.

An estimated 1.53 to $11.3\mathrm{t}/\mathrm{ha}$ of salt per irrigation was applied to flood irrigation bays (Table 2). The high salt loads applied at sites PG and MTM ($9.6–11.2\mathrm{t}/\mathrm{ha}$) was attributed to the higher salinity of irrigation water (4,800 to $6,400\mathrm{mg}/\mathrm{L}$, TDS).

Across all sites, the total salt output was slightly greater than the total salt input (Table 2). This may be because of the flushing (mobilization) of accumulated salt from the soil profile between irrigations over the short term. Over the longer term, however, it is reasonable to assume that the input of salt to the unsaturated zone via irrigation equals output via drainage.

The average salinity effect caused by evaporation of surface waters is minor (15%) in comparison to the salinity effect as a result of transpiration (85%). This is supported by Dincer et al. (1979), who showed that the contribution of transpiration from aquatic plants to water loss was greatest (71%) during summer, and Simpson et al. (1992), who showed that transpiration over an entire rice cropping season accounted for 60% of total losses to the atmosphere, with evaporation providing the remainder.

At NAP4 and NAP5, ponding occurred for up to 18 h postirrigation, and the salinity effect as a result of evaporation over the irrigation and ponding period ranged from 4 to 30% compared to the salinity effect as a result of transpiration (41 to 95%). In contrast, evaporation from rapidly draining soils at PG and MTM contributed only 2 to 18% of the effects compared to transpiration (81 to 97%).

An isotope study paralleling the same study sites (van den Akker et al. 2011) showed that evaporation from flood irrigation can amount to 6 mm day; however, when the crop was mature, evaporation was strongly limited by the dense canopy cover and can be 30% lower, and in some cases, negligible (i.e., $<1\mathrm{mm}$) when applied to rapid draining soils. This is supported by Fig. 5 and showed that isotopic signatures of soil-water collected postirrigation resemble partially evaporated irrigation waters, suggesting that soil-water did not undergo significant evaporation following irrigation. Transpiration of lucerne and pasture calculated by conventional methods (FAO56) can range from 4 to 6 mm per day (using crop coefficients of 0.8 and 0.9, respectively). Hence, over a 14 day irrigation cycle, water lost via transpiration can amount to $>56\mathrm{mm}$ between irrigations.

The cumulative water losses because of $E$ and $T$ and the corresponding salinity increase and soil moisture (capacitance) decrease in the topsoil (0.2 m) measured over a typical 14-day irrigation cycle following irrigation is shown on Fig. 7 for study sites NAP4 and NAP5. The results showed that 88% of water is lost by transpiration over the 14-day irrigation cycle. The water loss via transpiration and corresponding increase of salt concentration in the soil zone between irrigations is an ongoing process ($\mathrm{\text{amounting to}}+300\mathrm{mg}/\mathrm{l}$, TDS over a 4-day period post irrigation). The salt-water balance at these sites indicates that approximately 0.2 to $0.4\mathrm{t}/\mathrm{ha}$ of salt had accumulated between irrigations over a 14-day time frame. In contrast, the increase in salt concentration via evaporation was much less ($\mathrm{\text{amounting to}}+100\mathrm{mg}/\mathrm{l}$) because evaporation of irrigation waters occurred over a much shorter duration of 1 to 2 days, during the irrigation and ponding periods only.

At NAP5, the salinity effect because of evaporation of irrigation water over the duration of irrigation spanning 1.5 days was $0.12\mathrm{t}/\mathrm{ha}$, or $0.08\mathrm{t}/\mathrm{ha}/\mathrm{days}$; however, the salinity effect because of transpiration through concentration of soil-water salts over 14 days (between irrigations) was $0.4\mathrm{t}/\mathrm{ha}$ over 14 days, or $0.03\mathrm{t}/\mathrm{ha}/\mathrm{days}$ (Table 2). At PG, the salinity effect because of evaporation of irrigation water over duration of irrigation was $0.032\mathrm{t}/\mathrm{ha}/\mathrm{days}$, and the salinity effect because of transpiration through concentration of soil-water salts over 14 days (between irrigations) was $0.4–0.5\mathrm{t}/\mathrm{ha}$ over 14 days, or approximately $0.3\mathrm{t}/\mathrm{ha}/\mathrm{days}$ (Table 2).

The salinity effects because of evaporation at NAP4 were slightly more during the second irrigation than observed during the first irrigation. This was a result of irrigating at night during the second irrigation, thereby allowing water to pond and evaporate the following day. Salinity effects because of evaporation were also higher at MTM during the first irrigation when there was little crop cover (when $E>T$). Likewise, the salinity effect because of transpiration was $1\mathrm{t}/\mathrm{ha}$ greater during the second irrigation at MTM when the lucerne had reached maximum growth (when $T>E$), as shown in Table 2.

Because salinity (TDS) and ${\mathrm{Cl}}^{-}$ concentration are very strongly correlated $R^2=0.987$ (Fig. 8), and ${\mathrm{Cl}}^{-}$ is chemically inert and not involved in chemical reactions in the aquifer, the increase in salinity is not caused by mineral-water interactions within the aquifer and can only be explained from concentration through evaporation or transpiration. An increase in TDS through water-rock interaction (water-mineral reaction) would result in a nonlinear ${\mathrm{Cl}}^{-}$-TDS relationship. A TDS versus ${\mathrm{Cl}}^{-}$ plot of evaporated pan water has been included for comparison and showed a similar linear ${\mathrm{Cl}}^{-}$-TDS relationship as exhibited by soil-water.