Pesticides Research No. 104 2006 – Establishment of a basis for administrative use of PestSurf

Establishment of a basis for administrative use of PestSurf

Summary

Background and Objectives

The Danish Environmental Protection Agency (DEPA) initiated in 1998 a project concerning development of a decision support system for administration of pesticides in relation to their effects on surface water. The decision support system for exposure was build around the models MIKE SHE and MIKE 11 and included a user interface and a number of scenarios for sandy and sandy loam conditions. The system is called PestSurf.

The experience with the use of PestSurf is limited and as data are available for whole catchments, there are numerous possibilites of extracting results at different scales and for different periods. This makes the system flexible, but not easily interpreted as an administrative tool. As an example, high concentrations may occur in small, almost dry ditches just after spraying, but this case may not represent the situation wanted as a basis for the administrative decision. The same situation is not found in other administrative tools, which are based on more standardised but hypothetical situations. The objective of this project is to clarify the possibilities and limitations of the PestSurf system through use by calculation of concentrations for 12 pesticides (alpha-cypermethrin, bentazon, bromoxynil, fluazinam, malathion, metamitron, pendimethalin, propiconazol, prosulfocarb, rimsulfuron, terbutylazin and tribenuron-methyl) in 54 scenarios.

Pesticide concentrations were calculated based on direct spray, drift, three FOCUS Surface Water scenarios (sandy D3-ditch, sandy loam D4-pond and D4-stream) and four scenarios in PestSurf (sandy pond and stream, sandy loam pond and stream). The method generating the highest concentrations is “direct spray” as expected, except for one combination of PestSurf scenario and pesticide. The FOCUS scenarios always generate lower concentrations than the drift calculations. For four compounds, the PestSurf sandy pond generates higher concentrations than the corresponding drift calculation (fluazinam, pendimethalin autumn appl, terbutylazin and tribenuron methyl). For the Pestsurf sandy loam pond, all PestSurf simulations generates values that are lower than the drift calculations.

For the PestSurf sandy stream and sandy loam stream, PestSurf generates higher concentrations than the drift calculations for all pesticides. The highest concentrations in the simulations are always generated in the PestSurf sandy loam stream.

The ditch is not entirely comparable to the two sandy scenarios in PestSurf, while the sandy loam pond and stream must be judged to be comparable.

The maximum concentrations occurring in comparable water bodies may not be generated by the same type of process. The highest concentration in the sandy ditch is always caused by drift. This is also the case for the sandy stream. However, for compounds with a low sorption, transport to groundwater may take place in this scenario. Bentazon, Rimsulfuron and Tribenuron methyl have low Koc-values. The first two are transported to groundwater in this scenario, while Tribenuron methyl with the lowest DT50-value of the three seems to degrade.

For the PestSurf sandy pond, which is in direct contact with shallow groundwater, transport in groundwater is the dominating mechanism. Drift plays a role if the degradation is very fast, such as for malathion. For bromoxynil, the relatively high sorption and the fast degradation rate interacts to make drift an important source. For alphacypermethrin, the sorption and small dose play a role. As this compound degrades slowly, concentrations will tend to build up in groundwater over time. For the FOCUS SW-scenarios, groundwater does not transport pesticide. For autumn applications, the small water depth plays a role for the calculation of concentrations.

The FOCUS D4 stream receives all its major contributions from drift. This is also the case for 11-12 of the 14 PestSurf scenarios. Two scenarios are strongly affected by pesticide in groundwater contributions, and in one scenario, the maximum concentration is caused by a 20-year rainfall event. However, the drainage concentrations are significant in a number of the simulations, and because the simulation results indicate that the assumptions on which the drift calculation is based should be revised, the main conclusions concerning drain flow are described below.

For the PestSurf sandy loam stream scenario, drift is always important in case of malathion, where the fast breakdown counteracts leaching. For the upper part of the catchment, drift plays a role for alphacypermetrin and to some extent for bromoxynil, but drainage contributions are important. For several of the compounds, the highest concentration, particularly in the upstream part of the catchment, is generated by an extreme (20-year return period) precipitation event. The event is not important to malathion due to the fast degradation, and to prosulfoarb due to the spraying date, which is about a week after the extreme event. For bromoxynil it seems to be a combination of the high degradation rate and the time of spraying (april) or the autumn-application (october) that makes the extreme event unimportant. Similarly, tribenuron has a low DT50, is sprayed in april and is not dominated by the extreme event. FOCUS SW does not contain an extreme event of similar severity, and this should be taken into account in the comparison. The return period of the event (1 in 20 years) should be taken into account when judging the relevance of the event for administrative purposes.

In the lower part of the catchment, a contribution through groundwater becomes important for all substances with a DT50 value greater than 56 days in soil. Pendimethalin stands out with very high concentrations occurring at the lower end. As described in Styczen (2004a), it is believed that the buildup in groundwater is exaggerated in the model due to the description chosen for the macropores, and very high concentrations at the lower end of the catchment should therefore not be used as the only reason to discard a compound. However, occurrence of ”old” pesticides in baseflow is a known phenomena, and the spraying of 85-90 % of the catchment area eight successive years will cause a buildup of groundwater concentrations if the substance leaches in important amounts. FOCUS SW assumes that baseflow is free of pesticide and only contributes to dilution of the concentrations in the water body.

To avoid the described problem, values for administrative use should only be extracted in the upper part of the sandy loam stream (OvreLillebaek), between 500 m (where the stream is permanent) and 1700 m from the upstream end.

For the sandy loam pond, there seems to be good correspondence between the two models (FOCUS SW D4-pond and PestSurf sandy loam pond). In nine of the fourteen cases, both scenarios are dominated by drift. In three cases, drainage plays a role for both models. Only in two of the fourteen cases is there a discrepancy between the causes of the highest concentrations, which FOCUS SW considers drainage and PestSurf drift. In both cases drainage is also a source in PestSurf, and becaise the PestSurf pond is smaller, and therefore more exposed to drift, the difference is justified. The explanation is greater exposure and for the autumn applications a smaller water depth. If the concentration is due to drainage, the two models produce comparable results.

The FOCUS SW D3-ditch is compared to the sandy pond and sandy stream. The pond is dominated by entirely different mechanisms than the ditch and there is a poor correspondence between results generated. With respect to the stream, there is an almost linear correlation, but the concentration levels generated by PestSurf are a factor 4.3 larger. For the stream, the area sprayed is about three times the area sprayed for the ditch and the water depth in the upper part is considerably less than 30 cm, but the flow is much larger and part of the stream is protected by buffer zones.

Required improvements

During the project, an error was identified in the addition of drift to the waterbodies. The error was corrected, but for the streams, the generated concentrations became unrealistically high. It is recommende that the assumptions regarding drift be changed to represent field conditions more realistically.The proposed changes are described in Chapter 5.

During the process of data analysis, errors were found in the programme with respect to calculation of adsorption of substance to sediment in the stream. These errors are corrected. The maximum time step used for data storage for the stream has been changed from 3 days to 1 day. This makes the data extraction for PestSurf and FOCUS SW more comparable, and provides a better basis for calculation of key values. Further, it is recommended that an extraction routine is build that automatically generates the tables required for administrative use. Some simple macros were made for the present project that could be refined for this purpose.

The PestSurf result presentation templates are based on result presentation from pre-defined locations. The selected points did not always catch the maximum concentrations in the streams. A new version of the presentation templates should be made with data extraction in the points of maximum concentration identified as part of the present project. The templates do not contain points from the tributaries to the two streams. It is not recommended to change this, as the tributaries run dry for periods of the year. The same goes for the upstream end of the sandy loam catchment.

At present there are still no clear guidelines as to the interpretation of the spatial data, and how this should be included in an administrative evaluation of a compound. How much of the stream should be affected before it is important? And are all river stretches of equal importance? The very conservative method used in the report is to analyse the maximum values occurring in the catchment without any weight on the spatial and temporal occurrence or on where in the catchment they occur.

It would be possible to generate a 3-D-graph showing length of stream sections on the x-axis, time on the y-axis and concentrations (greater than or equal to) on the the z-axis. The resulting graph would be a “mountain-shaped” graph that for each cross section of the z-axis would show the time that a concentration level was “equal to or exceeded” for a given length of stream.

It is recommended that the Danish EPA at least decides on a “time criteria” like what is used for groundwater. If very high concentrations only occur during the 20-year-event, it must be considered equal to a 95 %-fractile according to the present climate. A similar consideration could be made for the very dry year (1996), but the simulations show that the summer period of the “normal year” was just as dry, and this event does therefore not really justify as an exception.

To avoid that a single point with a very high concentration becomes decisive for the simulation, a similar criterion could be implemented for the spatial distribution. The models contain 133 and 72 calculation points, of which 96 and 40 are present in the main streams, and 15 of the 40 lie between the 500-m-point and 1700 m point. A 95 %-criteria would be equal to 4-5 points in the sandy catchment and <1 point (60 m) in the sandy loam catchment.

It should be noticed that neither conditions such as the extreme drainage event, the groundwater contribution, the varying water level in the stream, the varying water level in the ponds nor the pond sizes identified as relevant under Danish conditions are represented in FOCUS SW. Water depths tend to be considerably lower than the 30 cm during summer and autumn in Danish 1. order streams, and the water depth in ponds also drop during dry periods.

The presently used description of colloid transport in macropores leads to an exaggerated flux of pesticide to groundwater. An ongoing project is working on an improvement of the description. It is recommended to include the improved description in the model when it becomes available.

The present version of the model was built in the 2002- version of MIKE SHE/MIKE 11. The models have been through rather strong revision since then, including an ongoing reprogramming exercise of the advection/dispersion scheme of MIKE 11. It is recommended to change the setup into the new version as and when other changes are to be implemented. Because the old version is not supported at code level anymore, it is extremely difficult to make fundamental changes in the present version. If the Danish EPA decides to continue with the PestSurf system, it is recommended to upgrade it as described in Chapter 6.