methomyl concentration for a time lag of about 33 h, and that even 3 weeks after
application, several pesticide peaks were detected in the stream water. During this
rather dry period many peaks were detected, with a significant delay (up to 24 h)
after the preceding rain event. For some of the peaks, a rain event of 0.1 mm was
sufficient to initiate pesticide transport from the hillslope to the stream. During the
30 day experiment, between 6.4 % and 11.4 % of the applied mass of methomyl was
recovered in the stream water; however, recovery of the stronger sorbing pesticide
chlorothalonil was distinctly less (data not shown) – only 1.6–3 % of the applied
pesticide mass was recovered.
In a follow-up field-scale study (Duffner et al. 2012), the experiment of Kahl
et al. (2008) was repeated for atrazine at the same hillslope site. This time; however,
the pesticide transport experiment was combined with a standard method used
in hydrology to identify hydrological flow components: a three-component
hydrograph separation. This study allowed assigning concentration peaks to specific hydrological flow components. Atrazine concentration peaks associated with
discharge peaks could be attributed to surface run-off, while concentration peaks on
the falling limb and during the later recession phase could be attributed to preferential interflow. Interflow is soil water that moves laterally (downslope) instead of
seeping vertically into the groundwater. Interflow was observed many times at the
hillslope site, flowing from the riparian zone into the stream. The appearance of
interflow depended strongly on soil water conditions, and was only active if a
certain soil water threshold was exceeded.
The characteristic input patterns observed at the field-scale were also observed at
the watershed scale. Sangchan et al. (2012) monitored seven pesticides frequently
applied by local farmers at the Uplands Program’s headwater and outlet gauging
stations, these being atrazine, chlorothalonil, chlorpyrifos, cypermethrin, dichlorvos, dimethoate, a- and b-endosulfan. Water sampling was performed in high
temporal resolution; every ten minutes one water sample of a fixed volume was
taken, and six samples were mixed to one composite sample resulting in an hourly
resolution. With the exception of dichlorvos, all pesticides were detected at both
stations, with pesticide peaks detected at the same time as the discharge peak, at the
falling limb of the discharge curve and at smaller peaks during the later phase of
recession. In addition, a fourth input pattern was found for nearly all the pesticides;
low, but more or less continuous concentrations on a baseline level after recession.
This pattern was probably related to some long-term storage in the underground
area and suggests that groundwater was already contaminated with these pesticides.
This conceptual understanding of pesticide transport, as derived from the above
mentioned field experiments, was translated into a simple mathematical model
(Kahl et al. 2010). For simulating pesticide transport from hillslopes to adjacent
surface waters, a two-domain water reservoir model was set up. In the first domain
of the model, water flows slowly through the soil matrix, while in the second
domain it flows quickly along fractures, resulting in rapid preferential flow. For
the study, surface run-off was modeled using a simple rainfall-infiltration intensity
relation. The model did not match exactly the single observed pesticide peaks, but
the shape and timing of the simulated peaks agreed well with the observed patterns.
3 Water and Matter Flows in Mountainous Watersheds of Southeast Asia:. . .
143
application, several pesticide peaks were detected in the stream water. During this
rather dry period many peaks were detected, with a significant delay (up to 24 h)
after the preceding rain event. For some of the peaks, a rain event of 0.1 mm was
sufficient to initiate pesticide transport from the hillslope to the stream. During the
30 day experiment, between 6.4 % and 11.4 % of the applied mass of methomyl was
recovered in the stream water; however, recovery of the stronger sorbing pesticide
chlorothalonil was distinctly less (data not shown) – only 1.6–3 % of the applied
pesticide mass was recovered.
In a follow-up field-scale study (Duffner et al. 2012), the experiment of Kahl
et al. (2008) was repeated for atrazine at the same hillslope site. This time; however,
the pesticide transport experiment was combined with a standard method used
in hydrology to identify hydrological flow components: a three-component
hydrograph separation. This study allowed assigning concentration peaks to specific hydrological flow components. Atrazine concentration peaks associated with
discharge peaks could be attributed to surface run-off, while concentration peaks on
the falling limb and during the later recession phase could be attributed to preferential interflow. Interflow is soil water that moves laterally (downslope) instead of
seeping vertically into the groundwater. Interflow was observed many times at the
hillslope site, flowing from the riparian zone into the stream. The appearance of
interflow depended strongly on soil water conditions, and was only active if a
certain soil water threshold was exceeded.
The characteristic input patterns observed at the field-scale were also observed at
the watershed scale. Sangchan et al. (2012) monitored seven pesticides frequently
applied by local farmers at the Uplands Program’s headwater and outlet gauging
stations, these being atrazine, chlorothalonil, chlorpyrifos, cypermethrin, dichlorvos, dimethoate, a- and b-endosulfan. Water sampling was performed in high
temporal resolution; every ten minutes one water sample of a fixed volume was
taken, and six samples were mixed to one composite sample resulting in an hourly
resolution. With the exception of dichlorvos, all pesticides were detected at both
stations, with pesticide peaks detected at the same time as the discharge peak, at the
falling limb of the discharge curve and at smaller peaks during the later phase of
recession. In addition, a fourth input pattern was found for nearly all the pesticides;
low, but more or less continuous concentrations on a baseline level after recession.
This pattern was probably related to some long-term storage in the underground
area and suggests that groundwater was already contaminated with these pesticides.
This conceptual understanding of pesticide transport, as derived from the above
mentioned field experiments, was translated into a simple mathematical model
(Kahl et al. 2010). For simulating pesticide transport from hillslopes to adjacent
surface waters, a two-domain water reservoir model was set up. In the first domain
of the model, water flows slowly through the soil matrix, while in the second
domain it flows quickly along fractures, resulting in rapid preferential flow. For
the study, surface run-off was modeled using a simple rainfall-infiltration intensity
relation. The model did not match exactly the single observed pesticide peaks, but
the shape and timing of the simulated peaks agreed well with the observed patterns.
3 Water and Matter Flows in Mountainous Watersheds of Southeast Asia:. . .
143