Pesticide leaching at FAARDRUP
Chapter 6, PLAP - Monitoring results May 1999 - June 2002
6.1 Materials and methods
6.1.1 Site description and monitoring design
Faardrup is located in southern Zealand (Figure 1). The test field covers a cultivated area of 2.3 ha (150 x 160 m). The terrain slopes gently to the west by 1-3°. Based on three profiles
in the buffer zone bordering the field, the soil was classified as Haplic Vermudoll, Oxyaquic Hapludoll and Oxyaquic Argiudoll (Soil Survey Staff, 1999). The topsoil is characterized as sandy loam with 15% clay and 1.4% organic carbon (Table 1). Within the upper 1.5 m numerous desiccation cracks coated with clay are present. The test field contains glacial deposits dominated by sandy till to a depth of about 1.5 m overlying a clayey till. The geological description shows that small channels or basins filled with meltwater clay and sand occur both interbedded in the till and as a large structure crossing the test field (Figure 38). The calcareous matrix and the reduced matrix begin at 1.5 m and 4.2 m b.g.s., respectively (Table 1). The dominant direction of groundwater flow is towards the west in the upper part of the aquifer (Figure 38). During the monitoring period the groundwater table ranged from 1 to 2 and 2 to 3 m b.g.s. in the lower and upper parts of the area, respectively. During fieldwork within the 5 m deep test pit it was observed that most of the water entering the pit came from an intensely horizontally fractured zone in the till at a depth of 1.8-2.5 m. The intensely fractured zone could very well be hydraulically connected to the sand fill in the deep channel which might drain part of the percolation. The bromide tracer study showed that virtually none of the applied bromide reached the vertical monitoring well (M6) located in the sand-filled basin (Section 6.2.2), however, thus indicating that hydraulic contact with the surface in the "basin" does not differ from that in other parts of the test field and that the basin is a small pond filled with sediments from local sources. A brief description of the sampling procedure is provided in Appendix 2. The monitoring design and test site are described in detail in Lindhardt et al. (2001) and the analysis methods in Kjær et al . (2002).
6.1.2 Agricultural management
The field was sprayed with glyphosate on 11 August 1999 and sown with winter wheat (cv. Stakado) on 20 August. Potassium bromide tracer was applied on 5 October. Weeds were sprayed on 14 October using ioxynil and bromoxynil and again on 4 April using fluroxypyr. Fungicide spaying was carried out on 5 May and 31 May using propiconazole and fenpropimorph. The insecticide pirimicarb was applied on 19 June. The crop was harvested on 28 August yielding 92.7 hkg/ha of grain and 76.2 hkg/ha of straw (85% and 100% dry mater, respectively).
Figure 38. Overview of the Faardrup site. The innermost white area indicates the cultivated land, while the grey area indicates the surrounding buffer zone. The positions of the various installations are indicated, as is the direction of groundwater flow (by an arrow).
Figure 39. Geological description of the Faardrup site (Lindhardt et al. 2001).
On 4 October 2000 the field was sprayed with glyphosate and ploughed 12 days later. Sugar beet (cv. Havanna) was sown on 2 May 2001. The herbicides metamitron, phenmedipham, desmedipham and ethofumesate were sprayed on 21 May, 30 May and 15 June. FluazifopP-butyl was sprayed on 21 June to combat wild oats, and pirimicarb on 17 July to combat pests. The crop was harvested on October 24 yielding 147.9 hkg/ha of roots and 38.0 hkg/ha of tops (both 100% dry matter).
The field was ploughed on 30 October 2001. Due to the good weather conditions, spring barley (cv. Barke) was sown earlier than usual on 28 March 2002. When the barley had 2 leaves, weeds were sprayed with tribenuron methyl on the 7 May. Herbicide spraying was also carried out on 22 May using MCPA and on 25 May using flamprop-M-isopropyl. The barley was sprayed with a fungicide and a pesticide, propiconazole and dimethoate, respectively, on 4 June. The crop was harvested on 9 August yielding 65.6 hkg/ha of grain (85% dry matter), which was a high yield for that particular cultivar that year. Ten days later the field was ploughed. Management practice at the site is detailed in Appendix 3 (Table A3.5).
6.1.3 Model set-up and calibration
The MACRO model was applied to the Faardrup site covering the soil profile to a depth of 5 m b.g.s., always including the groundwater table. The model was used to simulate the water flow in the unsaturated zone during the full monitoring period September 1999-June 2002 and to establish an annual water balance.
The model was calibrated to the observed groundwater table measured in the piezometers located in the buffer zone and to time series of soil water content measured at three depths (25, 60 and 110 cm b.g.s.) from the two profiles S1 and S2 (see Figure 38). A simple calibration procedure was applied that only necessitated adjustment of the empirical BGRAD parameter regulating the boundary flow and the drain depth, which was determined by the groundwater level during drainage periods. All remaining parameters were based on measured data or literature/default values. For a detailed description of data acquisition, model set-up and calibration procedures, see Kjær et al. (2002).
Extending the modelling period to include the third monitoring year revealed problems with the general water balance. A thorough analysis was therefore performed of the measured time series of precipitation and drainage flow at Faardrup. This revealed that the precipitation data logging at Faardrup was influenced by electronic noise, possibly resulting in overestimation of the precipitation input. This noise is not present at any of the other five VAP sites. Until this electronic noise problem is resolved, precipitation measured at Flakkebjerg 3 km east of Faardrup will be used instead. The analysis also resulted in minor adjustments of the measured drainage flow compared to the previous reported data (Kjær et al., 2002).
6.2 Result and discussion
6.2.1 Soil water dynamics and water balances
The model simulations were generally consistent with the observed data, thus indicating a good model description of the overall soil water dynamics in the unsaturated zone. The dynamics and level of the measured groundwater table were well captured by the model, as was the dynamics of the soil water content in all three horizons (Figure 40D, E and F). The dynamics of the drainage pattern also seems to be reasonably well described by the model, although the duration of the drainage periods was not fully captured. This resulted in an underestimation of the drainage flow in all three years by 13 to 63 mm. The difference between measured and modelled drain flow was greatest for the first monitoring period (July 1999-June 2000).
The underestimation of drainage volume might be due to uncertainty in the measurements of precipitation. As described in Section 6.1.3, precipitation measurements from nearby Flakkebjerg were used instead of measured precipitation at Faardrup. The underestimation of drainage volume could thus be due to the use of precipitation input at Flakkebjerg, which might underestimate the precipitation occurring at the Faardrup site. Until the Faardrup data have been further analysed, the Flakkebjerg time series will be used in the model.
The model generally has problems in simulating periods with low drainage flow, probably because these periods are characterized by partial drainage of the Faardrup site due to the topographic slope of the field. During such periods it is likely that only the lowest part of the field contributes to the drainage flow. The one-dimensional model will not be able to match the drainage flow on a field-scale where the groundwater table is above the drain depth in only part of the field. An example is the initial drainage period in 1999 (September to November), when the groundwater table was 1.65 m b.g.s., but 17 mm drained from the field.
The three monitoring periods at Faardrup are characterized as normal to wet years, with precipitation being 2-29 % higher than normal (Table 12). During the first monitoring period there was a long period in spring/summer 2000 with very little precipitation. As a consequence, the soil was very dry during autumn 2000 and the groundwater table was below 3 m. During the second monitoring period the drainage period was short and late, starting as late as January 2001. The model simulation showed that percolation 1 m b.g.s. was very similar for the first and third period, with continuous percolation from September to May. Averaged over the year, no groundwater recharge occurred during the first monitoring period since recharge during the dry months of May/June 2000 (-110 mm) counterbalanced the recharge during the period October 1999--April 2000 (98 mm). Despite the high precipitation input, groundwater recharge decreased to a low level. This can be explained by a combination of a high actual evapotranspiration (due to a winter crop type having high transpiration during winter and spring), a dry period in the spring/early summer and high drainage flow (due to a high groundwater level). In the third monitoring period, estimated total groundwater recharge was 121 mm, mainly due to the wet autumn of 2001 (102 mm).
Figure 40. Soil water dynamics at Faardrup: Measured precipitation and simulated percolation 1 m b.g.s. at Faardrup (A), simulated and measured groundwater level (B), simulated and measured drainage flow (C), and simulated and measured soil water saturation (SW sat.) at three different soil depths (D, E, and F). The measured data in B derive from piezometers located in the buffer zone. The measured data in D, E and F derive from TDR probes installed at S1 and S2 (see Figure 38).
During the second monitoring period (July 2000-June 2001) precipitation was close to normal at Faardrup, precipitation input being only 2% above the yearly normal precipitation. Despite a fairly wet autumn, percolation did not start until mid November. It continued until July 2001, however. From March to July 2001, drainage flow was low. Because of the initially low groundwater table in autumn 2000 and the dry spring in 2001, drainage flow only totalled 50 mm. Despite the normal precipitation input, the simulated groundwater recharge was significantly higher than for the first and third monitoring periods, reflecting the low evapotranspiration and drainage flow during the second monitoring period.
Table 12. Annual water balance for Faardrup (mm/year). Precipitation is corrected to the soil surface according to the method of Allerup and Madsen (1979).
6.2.2 Bromide leaching
The bromide tracer was not detected 1 m b.g.s. until late December 1999, about three months after application (Figure 41A and B). The bromide concentration in the suction cups (1 m b.g.s.) peaked during spring 2000, reaching a maximum of 7.8 mg/l. Evidence of bromide leaching was also found in the analysis of the drainage water samples derived from 1 m b.g.s. (Figure 41C). The bromide breakthrough was similar to that detected in the suction cups located 1 m b.g.s. Still, the concentration during the leaching period 1999/2000 was lower. When interpreting the bromide concentration profiles of the suction cups it should be kept in mind that they were beneath the groundwater table during the winter seasons (as indicated in Figure 40B).
Total recovery during the 3-year monitoring period amounted to 3.6 kg/ha, indicating that only 18% of the applied tracer had leached into the drains. Although concentration levels decreased during 2002, elevated bromide concentrations were detected in both suction cups and drainage water at the end of the monitoring period. The results are thus consistent with those for Silstrup and Estrup, and indicate that part of the bromide is retained in the upper part of the soil profile, probably in the matrix. Bromide can therefore be expected to continue to leach for a long time to come.
The results also showed subsequent minor transport of bromide to a depth of 2 and 3.5 m b.g.s. (Figure 41A, B and D). Slightly elevated bromide concentrations were detected 2 m b.g.s in the suction cups as well as in a horizontal well 3.5 m b.g.s. The bromide concentration in the suction cups located 2 m b.g.s. never exceeded 1 mg/l during 1999/2000, but increased to approx. 2 mg/l during winter 2001/2002 at the same time as the concentration decreased at 1 m b.g.s. A small part of the applied bromide also reached the downstream monitoring wells. Although the concentration and detection frequency were very low, slightly elevated concentrations were detected during autumn 2001 in M4 and M5 and to a minor extent in M6 (Figure 42).
Figure 41. Bromide concentration at Faardrup. A and B refer to suction cups located at S1 and S2. The bromide concentration is also shown for drainage runoff (C) and the horizontal monitoring wells (D). The green vertical line indicates the date of bromide application.
Figure 42. Bromide concentration at Faardrup. The data derive from the vertical monitoring wells (M2-M7). Screen depth is indicated in m b.g.s. The green vertical line indicates the date of bromide application.
6.2.3 Pesticide leaching
Monitoring began at Faardrup in September 2000 and presently encompasses 18 pesticides and 14 degradation products. Pesticide application is shown together with precipitation and simulated percolation in Figure 43 and Table 13. It should be noted that precipitation is corrected to the soil surface according to Allerup and Madsen (1979), whereas percolation (1 m b.g.s.) refers to accumulated values as simulated with the MACRO model. It should also be noted that as tribenuron methyl (applied here as Express) degrades rapidly, the leaching risk is more associated with its degradation product, triazinamin-methyl. For the same reason it is the degradation product and not the parent compound that is monitored in the PLAP (Table 13).
With bromoxynil, ioxynil, fluroxypyr, fenpropimorph and propiconazole, which were applied to winter wheat in 1999, the leaching risk was found to be negligible at the Faardrup site. Apart from one sample containing less than 0.1 µg/l of fluroxypyr, fenpropimorph and propiconazole, none of these compounds or their degradation products listed in Table 13 have yet been detected during the two-year monitoring period. For further details, see Kjær et al . (2003).
The leaching risk of the pesticides used on sugar beet in 2001 and spring barley in 2002 will not be evaluated until the 2003 monitoring results become available, i.e. when 2 years of monitoring data have been collated. The preliminary findings are that:
Table 13. Pesticides analysed at Faardrup with the product used shown in parentheses. Degradation products are in italics. Precipitation and percolation are accumulated from date of first application (App. date) until 1 July 2002. 1 st month percolation refers to accumulated percolation within the first month after application. Cmean refers to weighted average concentration in the drainage runoff. The number of pesticide-positive samples is indicated in parentheses.
- Phenmedipham (0.01-0.02 µg/l) and MHPC (0.03-0.19 µg/l) were each detected in two samples.
- Pirimicarb and its degradation products were detected in several drainage water samples as well as in one groundwater sample, in all cases at concentrations below 0.1 µg/l. As Pirimor was applied on two separate occasions, it is not possible to relate the findings to any one specific application.
- Glyphosate was applied to the field both in August 1999 and in October 2000. Both applications were followed by moderate precipitation input, and percolation commenced more than 1.5 months after glyphosate application. The leaching risk of glyphosate is minor at Faardrup as it was only found on 2 occasions in 4 water samples from the drainage system (time-proportional and flow proportional) and in 3 samples from the groundwater monitoring wells. The concentration interval was 0.01-0.093 µg/l. The degradation product AMPA was found more frequently in the drainage water (10 samples), suction cups (4 samples) and in 2 groundwater samples. AMPA was first detected in a suction cup 1 m b.g.s. in April 2001, 5 months after it had last been applied. From May 2001 to January 2002, AMPA was frequently detected at low concentrations (0.02- 0.11 µg/l) in both time-proportional and flow-proportional drainage water samples. It was last detected in February in samples from the vertical monitoring well. Since glyphosate was applied on two separate occasions, it is not possible to relate the findings to one specific application. The more frequent detection of AMPA a relatively long time after glyphosate application indicates desorption in the uppermost part of the soil system.
- Metamitron, metamitron-desamino, ethofumesate and fluazifop (free acid) were found
to leach from the root zone, reaching both the drainage system and one monitoring well. All four compounds were detected in high concentrations in the drainage system in June/August 2001 after an intense precipitation event, indicating rapid macropore transport (Figure 44). Drainage runoff was very low (<1 mm) during this period, however. Despite the high concentrations, the total mass of the four compounds that leached out was small. The concentrations decreased after a short time, and were 0.01-0.05 µg/l during autumn 2001. When runoff eventually started to increase in January 2002 the concentrations of all four compounds were below the detection limit. As a consequence the average concentrations were low, ranging from 0.01 to 0.06 µg/l (Table 13). These four compounds were also very frequently detected in one of the monitoring wells, M5 (Figure 45 and Appendix 7). Thus ethofumesate, metamitron and metamitron-desamino were found in all 3 screens starting in August 2001, when the groundwater table was about 2 m b.g.s. At the same time bromide was detected at M5 in slightly elevated bromide concentrations, thus providing additional evidence that percolating water from the treated area had reached M5. During autumn 2001, the compounds were detected in concentration exceeding 0.1 µg/l in several samples from M5, but were not detected in any of the other monitoring wells (Appendix 7).
Figure 43. Pesticide application, precipitation (primary axis) together with simulated percolation (secondary axis) at Faardrup. Etho.: ethofumesate; Desm.: desmedipham; Phenm.: phenmedipham. Pesticides applied later than April 2002 are not included.
Evidence of rapid movement of surface-near water to M5 was also provided by the inorganic data (Appendix 8). Following the dry summer of 2000 chloride and nitrate concentrations decreased in M5, but remained more stable in other monitoring wells (M6 and M4). The summer of 2000 was characterized by an extremely low soil water saturation (May- September) and a groundwater table that fell to 3 m b.g.s. (Figure 40). The extreme low soil water content entails the possibility that deep desiccation fractures could penetrate from the root zone down towards the underlying till and hence enable rapid transport of near-surface water to the monitoring screens.
When interpreting the detection of pesticides in M5, however, it should be kept in mind that the lower filters of M5 were hydraulically interconnected. Thus purging of the secondlowest filter (3.5-4.5 m b.g.s.) affected the overlying screen (2.5-3.5 m b.g.s.) in terms of a decreasing groundwater table. M5 is located downstream of the test site in till interbedded with thin sandy till lenses. The hydraulic connection between the filters is probably attributable to these lenses of sandy till (Lindhardt et al., 2001).
In conclusion, pesticides and their degradation products were transported through the unsaturated zone and reached the uppermost screen of M5. The detection of pesticides in the deeper screens should be interpreted with caution, however, as this might possibly be caused by screen purging. To clarify this matter, additional purging tests are planned in 2003. Moreover, a tracer test will also be conducted near well M5 to determine whether the frequent detection of pesticides at this well is attributable to fracture transport.
Figure 44. Precipitation (A) together with concentration of metamitron (B), metamitron-desamino (C), ethofumesate (D) and fluazifop-P (free acid) in the drainage runoff at Faardrup. The green vertical lines indicate the date of application.
Figure 45. Measured concentration of bromide (A), metamitron (B), metamitron-desamino (C), ethofumesate (D) and fluazifop-P (free acid) (E) in the vertical monitoring well M5 at Faardrup. Well positions are indicated in Figure 38. The green vertical lines indicate the date of pesticide application. Bromide was applied in October 1999.
The risk of pesticide leaching at Faardrup can be summarized as follows:
The leaching risk of pesticides applied in 2001 and 2002 cannot be fully evaluated at present as the potential leaching period extends beyond the current monitoring period. The preliminary findings are that:
- With bromoxynil, ioxynil, fluroxypyr, fenpropimorph and propiconazole applied to winter wheat in 2000 the leaching risk was found to be negligible.
- Desmedipham was not detected, whereas two samples were found to contain phenmedipham (0.01-0.02 µg/l) and MHPC (0.03 - 0.19 µg/l).
- Pirimicarb and its degradation products were detected in several samples, although always at concentrations below 0.1 µg/l.
- Glyphosate was detected at low concentrations in a very small number of samples. The degradation product AMPA was frequently detected for a relatively long period following application, thus indicating minor desorption in the uppermost metre of the soil. Apart from one sample containing 0.11 µg/l of AMPA, the concentration was always below 0.1 µg/l.
- Metamitron, metamitron-desamino, ethofumesate and fluazifop (free acid) were frequently detected in both drainage water and one monitoring well. Their average concentrations in the drainage water ranged from 0.02 to 0.06 µg/l.