Journal of Environmental Management 217 (2018) 980e990

Contents lists available at ScienceDirect

Journal of Environmental Management

journal homepage: www.elsevier.com/locate/jenvman

Research article

Coupling HYDRUS-1D with ArcGIS to estimate pesticide accumulation

and leaching risk on a regional basis
Ruediger Anlauf a, *, Jenny Schaefer a, 1, Puangrat Kajitvichyanukul b
a
Osnabrueck University of Applied Sciences, Faculty of Agricultural Sciences and Landscape Architecture, 49009, Osnabrueck, Germany
b
Naresuan University, Faculty of Engineering, 99 Moo 9, T. Tha-Pho, Mueang, Phitsanulok, 65000, Thailand

a r t i c l e i n f o a b s t r a c t

Article history: HYDRUS-1D is a well-established reliable instrument to simulate water and pesticide transport in soils. It
Received 30 September 2017 is, however, a point-specific model which is usually used for site-specific simulations. Aim of the
Received in revised form investigation was the development of pesticide accumulation and leaching risk maps for regions
18 March 2018
combining HYDRUS-1D as a model for pesticide fate with regional data in a geographical information
Accepted 22 March 2018
Available online 24 April 2018
system (GIS). It was realized in form of a python tool in ArcGIS. Necessary high resolution local soil
information, however, is very often not available. Therefore, worldwide interpolated 250-m-grid soil data
(SoilGrids.org) were successfully incorporated to the system. The functionality of the system is shown by
Keywords:
Pesticide accumulation
examples from Thailand, where example regions that differ in soil properties and climatic conditions
Pesticide leaching were exposed in the model system to pesticides with different properties. A practical application of the
Herbicide system will be the identification of areas where measures to optimize pesticide use should be imple-
Modeling mented with priority.
HYDRUS-1D © 2018 Elsevier Ltd. All rights reserved.
SoilGrids

1. Introduction problem (Rigotto et al., 2014). According to Riwthong et al. (2015),

farmers in the upland areas of northern Thailand spray pesticides
Land use intensification is usually accompanied by a reduction up to 16 times a year and use about 22 kg of active ingredients per
in the use of traditional methods of pest management and an in- hectare. Inappropriate pesticide handling, improper use of personal
crease in the use of synthetic insecticides, fungicides and herbicides protective equipment as well as inadequate understanding of the
(Riwthong et al., 2015). As a consequence, pesticide poisoning is an toxicity of the chemicals (Panuwet et al., 2008) put farmers in
important health problem in Thailand as a result of the intensive Thailand at high risk of pesticide poisoning. Panuwet et al. (2008)
use of hazardous pesticides (Tawatsin et al., 2015). Similar problems used urinary concentrations of pesticide metabolites to assess the
exist in several low and middle-income countries, such as Mexico, farmers' exposure to pesticides. Each year, thousands of cases of
Brazil, Uruguay, Cameroon or Malaysia (Schreinemachers et al., pesticide intoxication are reported with morbidity rates between
2011). Fast agricultural development results in an increase in 76.4 and 96.6 per 100000 inhabitants (Tawatsin et al., 2015).
agricultural productivity and inevitably in a shift from extensive to Apart from the significant human health risk, there are adverse
more intensive types of land use (Riwthong et al., 2015). Traditional consequences also for the environment. Pimentel (1995) estimated
methods of pest management are replaced by agrochemicals and that less than 0.1% of the pesticide applied to the crop actually
the intensity and amount of pesticide use grows rapidly reaches the target pest. The rest enters the environment and
(Schreinemachers et al., 2011; Riwthong et al., 2015). In Brazil, for adversely affects beneficial biota, contaminates soil, water and the
example, the pesticide market experienced a rapid expansion of atmosphere and, eventually, ends up again in contaminating the
190% over the last decade while pesticides are a major public health human population. The costs in Thailand of externalities like health
hazards, residues in food and water or resistance build-up are high,
but since pesticides still have a high reputation in their relevance
* Corresponding author. for securing sufficient agricultural production, Thailand's crop
E-mail addresses: [email protected] (R. Anlauf), j.schaefer.sako@gmail. protection policy has been and is still rather supportive for pesti-
com (J. Schaefer), [email protected] (P. Kajitvichyanukul). cide use (Jungbluth, 1996; Tirado et al., 2008; Schreinemachers
1
Present address: Beak Consultants GmbH, Am St.-Niclas-Schacht 13, 09599 et al., 2011). Thus, several studies conclude with the urgent
Freiberg, Germany.

https://doi.org/10.1016/j.jenvman.2018.03.099
0301-4797/© 2018 Elsevier Ltd. All rights reserved.
 R. Anlauf et al. / Journal of Environmental Management 217 (2018) 980e990 981

necessity of an integrated pest management in Thailand, the pro- pollution problems (Sim
 unek et al., 2012, 2013; Giannouli and
motion of good agricultural practice and of raising awareness Antonopoulos, 2015; Shrestha and Datta, 2015; Bernard et al.,
among farmers regarding the risk they are exposing themselves to 2016). There are several models available ranging from relatively
(Riwthong et al., 2015), and, furthermore, the promotion and simple analytical and semi-analytical solutions, to more complex
development of non-synthetic methods of pest control and the numerical approaches. Most models famous for predicting the fate
restriction of highly hazardous pesticides (Schreinemachers et al., and transport of agrochemicals are based on the convection-
2011; Tawatsin et al., 2015). A system to estimate pesticide accu- dispersion equation, such as HYDRUS-1D, which considers con-
mulation and leaching risk which can be used to identify highly vection, dispersion, adsorption and degradation (Ladu and Zhang,
endangered areas in a region, could supply predictable data of the

2011; Sim unek et al., 2012). HYDRUS-1D is a finite element model
pesticide's risk potential and thus play an important role in for simulating water, heat and solute movement in one-
developing an integrated pest management. The results can be

dimensional variably saturated media (Sim unek et al., 2012). It
used as a decision support for a sustainable pest management.
numerically solves the convection-dispersion equation for solute
HYDRUS-1D has established itself as a reliable instrument for
transport and the Richards equation for variably-saturated water
simulating pesticide transport in soils (Ladu and Zhang, 2011;
flow and can consider processes such as root water uptake, infil-


Sim unek et al., 2013; Leiva et al., 2017; Ma rquez et al., 2017). tration, evaporation, soil water storage, capillary rise, deep
HYDRUS-1D is public domain and, thus, freely available. Also, the

drainage, groundwater recharge and surface runoff (Sim unek et al.,
FORTRAN source code is freely available (PC-Progress, 2017).
2012). The solute transport equation accounts for advection and
HYDRUS-1D is a point specific model generally used to describe
dispersion in the liquid phase and diffusion in the gaseous phase
processes in soil columns or at homogeneous sites. However,
and considers linear equilibrium reactions between the liquid and
because HYDRUS-1D is a site-specific model, there are very few
the gaseous phase, zero-order production, and first-order degra-
examples where HYDRUS-1D was used to simulate the variability of

dation reactions (Sim unek et al., 2012). HYDRUS-1D may also
regions (Assefa and Woodbury, 2013). The effect of the variability of
consider different physical and chemical non-equilibrium transport
regions with respect to pesticide leaching and accumulation risk is
situations to account for e.g. preferential flow situations. Although
important to find out priority areas for water and soil protection.
non-equilibrium situations are particularly interesting in case of
In the risk assessment system developed in this investigation,
heterogeneous soils, the governing parameters are difficult to
HYDRUS-1D was coupled with ArcGIS to make it applicable to
obtain on a regional basis. Therefore, in this investigation only
larger regions. This combined system unites the advantageous
equilibrium conditions were considered.
properties of HYDRUS-1D, as a proper model for pesticide fate, and
HYDRUS-1D is widely used to evaluate the fate and transport of
the possibility of using a geographic information system (GIS). The
chemicals in the unsaturated zone between the soil surface and the
system has a user-friendly user interface which consists of
groundwater table (Sim
 unek et al., 2013). Several authors used
parameter inputs for soil, land management information, climate
information and pesticide application amounts. It offers the pos- HYDRUS-1D to successfully simulate pesticide transport in soils
sibility to select between different pesticides, climatic regions, and (e.g. Pang et al., 1999; Ko€ hne et al., 2006; Dousset et al., 2007;
irrigation options. Results of the system are maps showing the Papiernik et al., 2007; Abdel-Nasser et al., 2011; Ladu and Zhang,
accumulation and leaching risk of selected pesticides. As particu- 2011; Giannouli and Antonopoulos, 2015; Shrestha and Datta, 2015;
larly soil information is sometimes very difficult to obtain, the Bernard et al., 2016; Noshadi et al., 2017). Ko € hne et al. (2009)
system uses easily available soil data. concluded that HYDRUS-1D is one of the leading model systems
Paraquat and atrazine were used as examples to show the for simulating pesticide transport in agricultural structured field
applicability of the tool to estimate accumulation and leaching risk soils.
based on the simulation of a 10-year period with long-term average
climate data. The results are presented as maps with accumulation 2.2. Soil information
or leaching risk classes aiming to show differences between
different map units. No effort was taken to compare the simulated HYDRUS-1D needs soil information to simulate transport of
concentrations with measured concentrations to show actual pesticides in soils. Necessary soil information is needed for the
contaminations. Such site-specific simulations are a completely layers or horizons at each site to be calculated, or at least values for
different approach where much more information is needed, such the topsoil (0e30 cm) and subsoil (30e200 cm) as there are usually
as time and site-specific pesticide application amounts. This data is big differences, mainly with respect to organic carbon content. In
simply not available for regions. Due to the same reason, no cali- addition, the soil texture is needed (or the clay, silt, and sand
bration or validation of the model results was performed. This was content), the bulk density and the soil organic carbon content
not possible, and not necessary, to estimate differences in potential (Corg). The rooting depth defines the lower boundary of the soil;
contamination risk. pesticide leaching was assumed to take place once pesticides leave
Thus, the objectives of the investigation were i) to technically the rooting depth. Important soil physical parameters can be
achieve a coupling of HYDRUS-1D and ArcGIS with the help of a calculated with soil transfer functions from other parameters, such
python script, ii) to test the applicability of the system with avail- as available water capacity and hydraulic conductivity (estimated
able data from different regions in Thailand, and iii) to test whether from texture), or linear adsorption coefficient Kd (estimated from
worldwide available SoilGrids soil data are useable in the system. Corg) (reviewed e.g. by Guber and Pachepsky, 2010).
Detailed soil information at larger scales (at least 1:25.000,
depending on the size of the region under consideration and the
2. Materials and methods necessary accuracy of the simulation) is difficult to obtain at many
places. The reason is either simple non-existence of large scale soil
2.1. HYDRUS-1D data, or formal and administrative problems making soil data not
available to researchers. Therefore, online available soil data on a
Numerical models simulating and predicting contaminant 250-m grid (SoilsGrids) were used (Hengl et al., 2014, 2017;
transport processes in soils and groundwater are increasingly used Shangguan et al., 2017; SoilGrids, 2017). SoilGrids is an automated
in research and engineering projects addressing subsurface soil mapping system based on global soil profile and environmental
982 R. Anlauf et al. / Journal of Environmental Management 217 (2018) 980e990

covariate data using automated soil mapping based on machine the dimensionless Henry coefficient (KAW), the degradation coef-
learning algorithms (SoilGrids, 2017). SoilGrids data are available ficient in water (t1/2, d
1), the diffusion coefficients in water and air
worldwide at 250 m spatial resolution in 7 depths from 0 to 200 cm. (DPL and DPG, cm2 d
1), the dispersion length (Disp, cm) and the
Gridded data for clay, silt and sand content, bulk density, Corg, soil acceptable daily intake (ADIsoil, mg kg
1soil). Table 1 lists all 16
depth and stone content were downloaded for whole Thailand as pesticides and their properties that have been selected for the risk
example region. assessment tool.
Due to limitations in calculation time, HYDRUS-1D cannot run All properties were obtained either directly from the Pesticide
for every 250-m grid cell in larger regions. Therefore, the SoilGrids Properties DataBase (PPDB) (AERU, 2016) or by parameter estima-
data must be aggregated to space units having homogeneous tion: Necessary properties taken from the PPDB were the dimen-
properties. As soil texture (particularly clay content) and soil depths sionless Henry constant, the partition coefficient octanol to water
are very important soil parameters determining pesticide accu- (log KOW, -), the dissipation half-life (DT50, d), the molar mass (MP, g
mulation and leaching, these parameters were used to group the mol
1) and the acceptable daily intake ADI (mg kg
1body weight
data into new units. This was done with a python script in ArcGIS in d
1).
the following sequence: According to Trapp and Matthies (1998) the KOC was estimated
by the equation of Karickhoff (1981) as KOC ¼ 0.411 ∙ KOW. This
C The clay content of the first 4 SoilGrids layers (0e30 cm equation is a proper method for pesticides which do not dissociate
depths) was averaged for each grid cell to get an average under the soil's pH values and which are adsorbed mostly by
topsoil clay content; organic matter. However, the KOC was not estimated in the case of
C The topsoil clay content was reclassified into three groups paraquat, since it has a strong affinity for adsorption by not only
(lowest third, medium third and highest third); organic matter but also soil particles like clay minerals (Kamrin,
C The rooting depth was classified into three classes: shallow 1997). Instead, a KOC value of 100 was used, according to AERU
(<110 cm) medium (110e150 cm) and deep (>150 cm); (2016).
C The reclassified topsoil clay content and rooting depth were Equation (1) was used to estimate the diffusion coefficient in
intersected in ArcGIS liquid phase (DPL).
C Zones of adjoining grid cells were constructed having the (Trapp and Matthies, 1998):
same reclassified topsoil clay content and rooting depth;
these zones were considered as homogenous soil units
C For each of these homogeneous zones the average values of sffiffiffiffiffiffiffi
clay, silt, sand content, bulk density, Corg, soil depth and stone M0
content were calculated for all 7 depths down to 200 cm
DPL ¼ D0 $ (1)
MP
C For each zone, the texture was calculated according to the
FAO classification system (FAO, 2006) for all 7 depths based with.
on clay, silt and sand content.
DO ¼ Diffusion coefficient of oxygen in water ¼ 1.728e
4 m2 d
1
MO ¼ molar mass of oxygen ¼ 32 g mol
1
2.3. Pesticide application data MP ¼ molar mass of the pesticide (g mol
1)

Pesticide properties necessary for the simulation of transport Similarly, the diffusion coefficient in gaseous phase (DPG) was
are soil organic carbon-water partition coefficient (KOC, cm3 mg
1), estimated (Eq. (2), Trapp and Matthies, 1998):

Table 1
The 16 most important pesticides in Thailand and its properties used for the calculations.

CAS_RN Name of pesticide KOCa KAWb T1/2a DPLc DPGc Dispd ADIsoilb

1)
1
1)
1
1
3
(cm mg (
) (d ) (cm d2 2
(cm d ) (cm d ) (mg kg
1soil)

834-12-8 Ametryn 1.75E-01 1.70E-07 1.87E-02 0.6486 6249.7 5 15

1912-24-9 Atrazine 2.06E-01 1.20E-07 9.24E-03 0.6656 6413.3 5 20
133-06-2 Captan 1.30E-01 2.85E-07 8.66E-01 0.5638 5432.4 5 100
2921-88-2 Chlorpyrifos 2.06Eþ01 2.80E-04 1.39E-02 0.5218 5028.1 5 1
94-75-7 2,4-D 6.22E-05 1.40E-09 1.58E-01 0.6575 6335.1 5 50
1071-83-6 Glyphosate 2.59E-07 6.60E-19 4.62E-02 0.7517 7242.9 5 300
8018-01-7 Mancozeb 8.79E-03 1.76E-10 6.93Eþ00 0.5935 5718.3 5 50
16752-77-5 Methomyl 5.06E-04 7.50E-09 9.90E-02 0.7675 7395.2 5 2.5
1910-42-5 Paraquat dichloride 1.00Eþ02 1.70E-12 1.90E-03 0.6096 5873.4 5 4
15263-53-3 Cartap Hydrochloride 4.61E-05 5.62E-20 2.31E-01 0.6345 6113.8 5 100
63-25-2 Carbaryl 9.42E-02 4.50E-04 4.33E-02 0.6891 6639.8 5 7.5
52315-07-8 Cypermethrin 8.20Eþ01 3.70E-06 1.16E-02 0.4791 4616.2 5 50
55285-14-8 Carbosulfan 1.08Eþ04 1.83E-05 3.30E-02 0.5011 4828.8 5 5
10605-21-7 Carbendazim 1.24E-02 8.82E-07 1.73E-02 0.7069 6811.4 5 20
12071-83-9 Propineb 2.26E-04 1.90E-06 2.31E-01 0.5742 5532.7 5 7
20427-59-2 Copper II hydroxide 1.13E-03 7.92E-11 6.93E-05 0.9897 9535.7 5 150

CAS_RN: CAS Registry Number, KOC: Soil organic carbon-water partition coefficient, KAW: Dimensionless Henry coefficient, t1/2: Degradation coefficient in water, DPL: Pes-
ticide's diffusion coefficient in water, DPG: Pesticide's diffusion coefficient in air, Disp: Dispersion length, ADIsoil: Acceptable daily intake calculated from ADI.
a
Kotz et al., 2008.
b
AERU, 2016.
c
Trapp and Matthies, 1998.
d
FOCUS, 2009.
 R. Anlauf et al. / Journal of Environmental Management 217 (2018) 980e990 983

2.5. Assessment of leaching risk

sffiffiffiffiffiffiffiffiffi
MW
DPG ¼ DW $ (2) To assess leaching risk, legal thresholds for drinking water
MP
quality in Germany were used. The threshold is uniform for nearly
all pesticides and is 0.1 mg L
1 for a single substance, while
with. 0.5 mg L
1 is the limit for the sum of all substances, and 1.0 mg L
1 is
the specified action value (Drinking Water Ordinance, 2016; EU-
DW ¼ Diffusion coefficient of water vapor in air ¼ 2.22 m2 d
1 Pesticides Database, 2018). These thresholds were used to eval-
MW ¼ Molar mass of water ¼ 18 g mol
1 uate the leaching risk based on the calculated bottom concentra-
MP ¼ Molar mass of the pesticide (g mol
1) tions in liquid phase during a ten years simulation period with long
term monthly climate averages (Table 3).
The degradation coefficient in water can be related to the DT50 While the risk category gives only a qualitative view of the
as follows (Eq. (3), Kotz et al., 2008): leaching, the number of days where the threshold value of
0.1 mg L
1 is exceeded gives further information. Combination of
0:693 leaching concentration and number of threshold exceedance days
t1 ¼ (3)
2 DT50 give a better impression of how serious the risk concerning
leaching to the groundwater might be.
Finally, in accordance with FOCUS (2009) the dispersion length
was set uniformly to 5 cm.
Farmers in Thailand do not follow any rule regarding pesticide
application amounts or cycles. Very often they create their own
pesticide mixture and apply as much as they need, some follow the 2.6. Assessment of accumulation risk
instructions on the pesticide's label. For the calculations the
assumption was made, in which the pesticides are applied only The acceptable daily intake (ADI, mg kg
1body weight d
1) is a
once a year in early spring (10th calendar week), with application measure of the amount of pesticide, that can be ingested on a daily
amounts and concentrations as they are declared on the pesticide's basis and over a lifetime without any risk to health. It is based on
label. Table 2 shows the application amounts used for the results from international toxicological studies, which follow Eu-
calculations. ropean regulations (EU-Pesticides Database, 2018). Since it was not
possible to find any thresholds for pesticide concentrations in soils
in German regulations or any law in Thailand, the total pesticide
concentration in the soil in relation to the ADI seems to be a good
approach to at least get an idea of the dimension of pesticide
accumulation in the upper soil.
2.4. Climate data
As the ADI taken from AERU (2016) is related to one day and one
kilogram of body weight, it was converted to ADISoil which is
Climate data needed for the simulation are long term averages
related to one kilogram of soil assuming a body weight 10 kg (child)
of precipitation, number of rainy days and average evaporation for
and an amount of soil swallowed of 10 g d
1 (Eq. (4)).
every month of the year. The user may provide own climate data
but may also select between six different climate sets of six regions
in Thailand. The choice of the climatic regions is based upon the Table 3
division by the Thai Meteorological Department (2014). Since it was Leaching risk assessment classes based on 10 years simulation with long-term
difficult to find one source with all necessary information the monthly climate averages.

climate sets are a data collection coming from different Web pages No risk no exceedance of any threshold value
or books (mainly long-term averages 1982e2012). Precipitation Low risk exceeding at least one day 0.1 mg L
1
was obtained from Climate-Data.org (2012), monthly evaporation Medium risk exceeding at least one day 0.5 mg L
1
High risk exceeding at least one day 1.0 mg L
1
was taken from Muller (1982).

Table 2
Application amounts of the pesticides included in the risk assessment system.

CAS_RN Name of pesticide Application in Calendar week Amount of water Amount of pesticide Concentration of active substance (%)
(L rai
1a) (mL L
1 of water)

1910-42-5 Paraquat dichloride 10 80 7.5 27.6

2921-88-2 Chlorpyrifos 10 80 4.5 40
834-12-8 Ametryn 10 80 7.5 80
94-75-7 2,4-D 10 80 5.25 85
1071-83-6 Glyphosate 10 80 6.25 48
52315-07-8 Cypermethrin 10 80 1.5 35
1912-24-9 Atrazine 10 80 2.34 80
133-06-2 Captan 10 560 2.5 50
8018-01-7 Mancozeb 10 560 4 80
16752-77-5 Methomyl 10 560 1.75 40
63-25-2 Carbaryl 10 80 3 85
10605-21-7 Carbendazim 10 560 1 50
12071-83-9 Propineb 10 80 1.5 70
20427-59-2 Copper II hydroxide 10 80 1 77
55285-14-8 Carbosulfan 10 80 2 20
15263-53-3 Cartap Hydrochloride 10 80 1.5 50
a
1 rai ¼ 1600 m2 (rai: common unit of area in Thailand).
984 R. Anlauf et al. / Journal of Environmental Management 217 (2018) 980e990

ADI$BW$1000
ADISoil ¼ (4)
Soilswallowed

with.

ADISoil ¼ ADI in soil (mg kg
1)

ADI ¼ ADI (mg kg
1body weight d
1)
BW ¼ Body weight of a child, assumed as 10 kg
Soilswallowed ¼ Amount of soil swallowed per day, assumed as
10 g

The ADI-factor is used to demonstrate pesticide accumulation.

Its value is obtained by calculating the total pesticide concentration
in the upper 30 cm after ten years and dividing it by the ADISoil (Eq.
(5)).

Ctot10a
ADI
factor ¼ (5)
ADIsoil

with.

ADI-factor ¼ factor relating total soil concentration and ADI (
)

Ctot10a ¼ total soil concentration after 10 years (mg kg
1)
ADIsoil ¼ ADI in soil (mg kg
1)

An ADI-factor value below 1 would denote a soil which could be Fig. 1. Workflow of the python tool.
swallowed without exceeding the ADI, while a value larger than 1
would result in exceedance of the ADI (under the assumptions
described above). The ADI-factor was used to evaluate the accu- the version 4.16.0080 of HYDRUS-1D was used in combination with
mulation risk according to Table 4. ArcGIS 10.3.0.

2.8. Example regions

2.7. Workflow of the program
4 example areas of 100 km  100 km in Thailand with different
Coupling of HYDRUS-1D and ArcGIS was done by creating a
climate and different soils were selected (Fig. 2), i.e. Northern
python toolbox, which is a very flexible way of creating a tool in
Thailand (Chiang Mai), Northeast Thailand (Khon Khaen), Central
ArcGIS. The toolbox contains a python tool which runs HYDRUS-1D
Thailand (Phitsanulok) and Central-South Thailand (Bangkok).
in the background, reads its output files and returns necessary in-
Some climate information is given in Table 5 with precipitation
formation about pesticide concentrations so that ArcGIS can display
ranging between 1162 mm a
1 in the Northeast and 1430 mm a
1
it as a map.
in Bangkok. Fig. 3 shows as an example the soil property maps of
The general flow of the tool is as follows (Fig. 1): As a first step,
the example region Central Thailand (Phitsanulok) for which risk
the user provides all necessary information through the tool's
maps are presented in the results and discussion section. In the
graphical user interface (GUI), including soil, land use, climate in-
western part of the region deep soils are dominant with depths of
formation, area of interest, selection of pesticides, etc. The data is
more than 2 m. In the eastern part, more shallow soils with depths
provided either in the form of shape files (spatial information) or
of mainly between 180 and 200 cm occur (Fig. 3a). The texture of
EXCEL files (tabular information).
the deep soil region is mainly clay loam with a bulk density of
During tool execution, the user input is read, the input files for
>1.35 g cm
3, whereas the texture of the shallower part is loamy
HYDRUS-1D are written, HYDRUS-1D is run in the background and
with bulk densities between 1.30 and 1.35 g cm
3 (Fig. 3b and c).
the output files of HYDRUS-1D are read. All necessary pesticide
The western part shows Corg values of about of 0.012e0.018 g g
1
concentrations are obtained and returned to ArcGIS. Finally, the
whereas the eastern part has higher Corg values of 0.022 -
attribute table and the symbology of the shapefile with the calcu-
>0.028 g g
1 (Fig. 3d). Input and risk maps for the other three re-
lation results are updated, so that the calculated pesticide leaching
gions are given as supplemental files.
and accumulation risk maps can be displayed. For the calculations,

3. Results and discussion

Table 4
Accumulation risk classes.
The functionality of the risk assessment system was tested for
ADI-factor Accumulation Risk Class the example regions calculating the leaching and accumulation risk
0 1 (no accumulation risk) for the two herbicides atrazine and paraquat. Atrazine and paraquat
>0e0.25 2 (very low accumulation risk)
are on the PANs International List of Highly Hazardous Pesticides
>0.25e0.5 3 (very low accumulation risk)
>0.5e0.75 4 (very low accumulation risk) (PAN, 2016), are prohibited in the European Union (EU) (Bethsass
>0.75e1.0 5 (low accumulation risk) and Colangelo, 2006; Watts, 2011), but still approved for use in
>1.0e5.0 6 (medium accumulation risk) Thailand (Tawatsin et al., 2015). Atrazine is used for broadleaf and
>5.0e10.0 7 (high accumulation risk) grassy weed control in crops such as corn, sugarcane, pineapple,
>10.0 8 (very high accumulation risk)
sorghum, asparagus or conifer restoration planting, but also on
 R. Anlauf et al. / Journal of Environmental Management 217 (2018) 980e990 985

Fig. 2. Example regions in Thailand.

Table 5 sediments with half-lives of up to 20 years and 820 years, respec-

Climate data from the example regions. tively (Watts, 2011). It can be degraded by ultraviolet light, sunlight
North Northeast Central Bangkok and microorganisms (Kamrin, 1997).
Mean annual precipitation (mm) 1183 1162 1358 1430
The accumulation risk for paraquat is shown in Fig. 4a: The map
Rainy days per year (d) 102 118 93 96 is clearly divided into a western region with ADI-factors of less than
Mean annual evaporation (mm) 1517 1444 1482 1806 unity (0.75e1) which means that risk of exceeding the ADI is low.
The eastern region has ADI factors of 1e5 meaning that exceedance
of the ADI is expected. The accumulation risk for atrazine is shown
non-cropped industrial lands and on fallow lands (Kamrin, 1997; in Fig. 4b: The risk of accumulation of atrazine is clearly very much
Bethsass and Colangelo, 2006). According to AERU (2016), atra- lower than for paraquat. ADI-factors for atrazine range from 0 to
zine is moderately toxic to mammals and animals and is generally 0.25. Obviously, the maps mirror the different properties of para-
not persistent in the field nor in aquatic systems. However, several quat (high adsorption coefficient) and atrazine (low adsorption
studies show that it has a high mobility and persistence in water coefficient), but also the differences in soil properties: as paraquat
and causes ubiquitous and unpreventable water contamination, is mainly adsorbed at soil organic matter, the differences in soil Corg
which is why it was banned in the EU (Kamrin, 1997; Bethsass and in the western and eastern part of the example area reflect the
Colangelo, 2006). It is mainly degraded by chemical hydrolysis and adsorption of paraquat. Because the adsorption of atrazine is very
soil microorganisms (Kamrin, 1997)Paraquat is a highly acutely low, differences in soil Corg will have very little effect on its
toxic herbicide but widely used on more than 100 crops such as adsorption.
rice, banana, oil palm, tea, alfalfa, beans, vineyards, potatoes and The leaching potential for paraquat and atrazine is shown in
also non-cropped areas such as roadways or paths (Watts, 2011; Fig. 5 a and b, respectively. The leaching potential for paraquat is
AERU, 2016). It is highly toxic via ingestion, inhalation or absorp- negligible (no leaching risk, Fig. 5a) and, thus, an exceedance of the
tion through damaged skin, extremely biologically active and toxic drinking water threshold in leachate will not take place at any day
to plants and animals and extremely persistent in soil and (not shown in the maps). There is mostly no leaching risk for
986 R. Anlauf et al. / Journal of Environmental Management 217 (2018) 980e990

Fig. 3. Soil properties for the example region Central Thailand (Phitsanulok): a) Profile depth (cm), b) soil texture, c) bulk density (g cm
3), d) organic carbon (Corg; g g
1).

atrazine in the eastern part of the map (Fig. 5b). Possible reasons are the eastern part of the map is also visible in the map of the number
the mostly high Corg values. However, there is some variability in of days of exceedance of the dinking threshold from zero days to
this eastern part of the maps with respect to soil depth, texture, >1050 days (out of 10 years ¼ 3650 days; Fig. 5c). The leaching risk
bulk density and soil Corg (see Fig. 3aed). This results in smaller for atrazine in the much more homogeneous western part of the
areas with low, medium and high leaching risks. This variability in map is higher, but still in the risk class “low” (Fig. 5b and c).
 R. Anlauf et al. / Journal of Environmental Management 217 (2018) 980e990 987

Fig. 4. Accumulation risk (based on 10-year simulation with long term climatic averages): ADI_Factor for the example region Central Thailand (Phitsanulok): a) Paraquat accu-
mulation risk, b) Atrazine accumulation risk.

Each of the 4 example regions (North, Northeast, Center, the model and may increase leaching and accumulation risk at
Bangkok) has different soil and climatic conditions. Maps of the lower positions and decrease it in higher positions. There is also no
leaching and accumulation risk for the regions North, Northeast comparison of the soil data used (SoilGrids) with large scale site
and Bangkok for paraquat and atrazine are given as supplemental specific soil information, because such information is simply not
files. Average values of leaching and accumulation risk are given in available in the example region. Inaccuracies in the soil information
Table 6 to illustrate the differences between the different regions. In may have an important effect on the results. Therefore, future
all regions there is no leaching risk for paraquat, because it is research to improve the tool should concentrate on the quality of
strongly adsorbed at the soil particles. This high adsorption is re- the pesticide application data and soil information.
flected in the average area weighted ADI-factors for paraquat However, objective of the investigation was not to simulate
ranging from 0.95 in the Bangkok region to 1.07 in the Northeast. concentrations in the drainage water or total amounts in the soil,
These average values are all around 1 meaning that the risk of which would need much more data which are at the moment not
paraquat accumulation in the soil and exceeding ADI values is high. available on a regional basis. Main objective was to keep the model
The average leaching risk for atrazine (Table 6) is between 1.01 simple and to give hints on priority areas where a general higher
and 1.52 (with 1 ¼ no leaching risk and 2 ¼ low leaching risk) and accumulation and leaching risk can be expected.
with an average number of days of exceedance of the drinking
water threshold from 6 to 236 days. The average accumulation risk
of atrazine is low with ADI-factors between 0.11 and 0.21. These 4. Applicability of the leaching and accumulation risk maps
values demonstrate the variability between the 4 different example
regions. However, these average values are of minor importance Problems related to pesticide use in Thailand and many other
because they do not reflect the important variability within the countries show the necessity to improve the situation for the health
regions. Practical advantages of the leaching and accumulations benefit of farmers and for the environment. The maps created show
risk calculations can only be drawn from the maps. the high variability of the pesticide leaching and accumulation risk
The leaching and accumulation risk maps show differences in the different example regions. Therefore, an improvement of the
between areas as potential risks divided into risk classes. An situation requires site (and region) specific procedures based on
essential limitation of this approach is the quality of the input data. priority zones.
Particularly the real pesticide application by farmers with respect An application example is the approach to combat contamina-
to amount and number is virtually unknown. The real amounts of tion of the groundwater in drinking water catchments in the
pesticide leached to the groundwater and accumulated in the upper 1980ies with agrochemicals in Germany and other countries in
soil, respectively, may be much higher than calculated in the model. Central Europe which was based on priority areas. The potential
The reason is that only one application of pesticides is assumed, groundwater contamination risk was highly variable in different
whereas this number might be much higher in reality. Also, surface areas of the country depending mainly on precipitation and soil
flow of water from higher to lower positions is not considered in properties. Therefore, a site and region-specific approach was
established (Anlauf, 2003). Special cooperative groups in priority
988 R. Anlauf et al. / Journal of Environmental Management 217 (2018) 980e990

Fig. 5. Leaching risk (based on 10-year simulation with long term climatic averages) for the example region. a) paraquat leaching risk, b) atrazine leaching risk, c) exceedance of
drinking water threshold for atrazine.

areas were founded after 1990 with representatives of local adequate to the specific situation. Measures were a combination of
farmers, advisory services, government administration and local information, ban of some substances, economic incentives for
scientists (Anlauf, 2003; Kastens and Newig, 2008; Adem et al., farmers in prority areas, compensation payments, help to establish
2017). These groups decided about measures to be taken being organic farming, amongst others. Also, special industry
 R. Anlauf et al. / Journal of Environmental Management 217 (2018) 980e990 989

Table 6
Average values of leaching risk, days of exceedance of the drinking water standards, and ADI-factor for the 4 example regions in Thailand.

Site Average Leaching risk classa Days exceeded (d) ADI-factor (
)

North Atrazine 1.02 9 0.14

Paraquat 1.00 0 1.08
Northeast Atrazine 1.27 236 0.21
Paraquat 1.00 0 1.07
Central Atrazine 1.52 119 0.13
Paraquat 1.00 0 0.98
Bangkok Atrazine 1.01 6 0.11
Paraquat 1.00 0 0.95
a
1 (no leaching risk) e 8 (very high leaching risk).

independent advisory service for water protection zones was restrictions may concentrate on these priority areas.
founded to give advice to farmers with a focus on groundwater
protection measures. In the two decades after 1990, the pollution of Funding sources
the groundwater improved tremendously, although until today not
all problems are solved. Thus, site specific measures in priority This work was supported by the European Union (ERASMUS-
areas seem to be an adequate method to improve the situation. Plus) and Osnabrueck University of Applied Sciences.
Similar successful approaches are reported from other parts of the
world (Nel et al., 2016; Safford et al., 2017). As a prerequisite, Conflicts of interests
however, it is necessary to identify priority areas to start with,
because time and money will most probably be insufficient to begin None.
simultaneously everywhere. This is where leaching and accumu-
lation risk maps are a necessary help to identify these priority areas. Appendix A. Supplementary data
Future research to improve the accumulation and leaching risk
maps must focus on the following topics: SoilsGrids soil informa- Supplementary data related to this article can be found at
tion was used because it seems to be the best soil information easily https://doi.org/10.1016/j.jenvman.2018.03.099.
available in Thailand and in many other parts of the world. This soil
information, however, is interpolated from many sources of very
References
different quality (SoilGrids, 2017). Whether the quality of the
simulation may be improved by large scale on-site verified soil data Abdel-Nasser, G., Al-Turki, A.M., Al-Wabel, M.I., El-Saeid, M.H., 2011. Behaviour of
should be investigated at least for example areas. HYDRUS-1D has atrazine and malathion pesticides in soil: simulation of transport process using
been effectively used in many detailed investigations worldwide numerical and analytical models. Res. J. Environ. Sci. 5, 221e235. https://doi.
org/10.3923/rjes.2011.221.235.
(Ladu and Zhang, 2011; Sim
 unek et al., 2013; Leiva et al., 2017; Adem, E.B., Geneletti, D., Albert, C., 2017. Boundary work for implementing adaptive
Marquez et al., 2017). However, detailed measurements at management: a water sector application. Sci. Total Environ. 593, 274e285.
https://doi.org/10.1016/j.scitotenv.2017.03.121.
example sites in Thailand with different soil, land use and climatic AERU, 2016. Pesticide Properties DataBase. Agriculture & Environment Research
conditions are necessary to validate the simulated results and, thus, Unit (AERU) of the University of Herfordshire, UK. http://sitem.herts.ac.uk/aeru/
improve the possible acceptance of the risk maps. Another impor- ppdb/en/atoz.htm. Accessed: 2016-January0-5.
Anlauf, R., 2003. Groundwater protection strategy in Niedersachsen, Germany. In:
tant point for future research is the further development of the
N€aringen, Greppa (Ed.), Environmental Goals in Professional Vegetable and
model system towards connectivity between map areas. As Potato Production (Alnarp, Sweden).
mentioned before, HYDRUS-1D is a point specific model which Assefa, K.A., Woodbury, A.D., 2013. Transient, spatially varied groundwater recharge
modeling. Water Resour. Res. 49, 4593e4606. https://doi.org/10.1002/wrcr.
does not simulate lateral surface fluxes of water and pesticides from
20332.
higher to lower topographic positions. This may have an important Bernard, P.Y., Benoît, M., Roger-Estrade, J., Plantureux, S., 2016. Using biophysical
influence on pesticide accumulation and leaching risk. Further, models to manage nitrogen pollution from agricultural sources: utopic or
detailed information on pesticide application amounts and appli- realistic approach for non-scientist users? Case study of a drinking water
catchment area in Lorraine, France. J. Environ. Manage. 183, 260e274. https://
cation numbers is completely missing. Farmers will very often doi.org/10.1016/j.jenvman.2016.08.050.
apply pesticides more often and with higher amounts than Bethsass, J., Colangelo, A., 2006. European Union bans atrazine, while the United
assumed in the simulations. This may lead to higher leaching and States negotiates continued use. Int. J. Occup. Environ. Health 12 (3), 260e267.
https://doi.org/10.1179/oeh.2006.12.3.260.
accumulation risks than calculated in the examples. An advisory Climate-Dataorg, 2012. Climate Data for Cities Worldwide. http://en.climate-data.
system should be established to determine such data and, if org/. Accessed: 2016-January0-5.
possible, to develop measures to decrease pesticide input. Drinking Water Ordinance, 2016. Ordinance on the Quality of Water Intended for
Human Consumption. https://www.bundesgesundheitsministerium.de/
fileadmin/Dateien/3_Downloads/E/Englische_Dateien/Drinking_Water_
Ordinance.pdf. Accessed: 2018-February-5.
5. Conclusions Dousset, S., Thevenot, M., Pot, V., Sim
unek, J., Andreux, F., 2007. Evaluating equi-
librium and non-equilibrium transport of bromide and isoproturon in disturbed
and undisturbed soil columns. J. Contam. Hydrol. 94, 261e276. https://doi.org/
It is technically possible to couple HYDRUS-1D with Arc-GIS to 10.1016/j.jconhyd.2007.07.002.
create leaching and accumulation risk maps. The maps created for EU-Pesticides Database, 2018. http://ec.europa.eu/food/plant/pesticides/eu-
example regions in Thailand show high variability of results pesticides-database/public/?event¼homepage&language¼EN. Accessed 2018-
February-5.
depending on soil, climate and pesticide properties. If local soil data
FAO, 2006. Guidelines for Soil Description. Food and Agriculture Organization of the
are not available, SoilGrids-250 m raster data may be used. The United Nations, Rome, Italy. ftp://ftp.fao.org/agl/agll/docs/guidel_soil_descr.pdf.
quality and reliability of these soil data must be further investi- (Accessed 20 September 2017).
gated. The accumulation and leaching risk maps can help to identify FOCUS, 2009. Assessing Potential for Movement of Active Substances and Their
Metabolites to Ground Water in the EU. Technical report, report of the FOCUS
areas with a high leaching and/or accumulation risk for specific Ground Water Work Group, EC Document Reference Sanco/13144/2010, p. 604,
pesticides. Special advisory service and site-specific land use version 1. http://eusoils.jrc.ec.europa.eu/public_path/projects_data/focus/gw/
990 R. Anlauf et al. / Journal of Environmental Management 217 (2018) 980e990

NewDocs/FOCUS2009FinalreportGW.pdf. (Accessed 5 February 2018). Pang, L., Close, M.E., Watt, J.P.C., Vincent, K.W., 1999. Simulation of picloram, atra-
Giannouli, D.D., Antonopoulos, V.Z., 2015. Evaluation of two pesticide leaching zine and simazine leaching through two New Zealand soils and into ground-
models in an irrigated field cropped with corn. J. Environ. Manage 150, water using HYDRUS-2D. J. Contam. Hydrol. 44, 19e46. https://doi.org/10.1016/
508e515. https://doi.org/10.1016/j.jenvman.2014.12.044. S0169-7722(00)00091-7.
Guber, A.K., Pachepsky, Y.A., 2010. Multimodeling with Pedotransfer Functions. Panuwet, P., Prapamontol, T., Chantara, S., Thavornyuthikarn, P., Montesano, M.,
Documentation and User Manual for PTF Calculator (CalcPTF). USDA-ARS, Whitehead, R., Barr, D., 2008. Concentrations of urinary pesticide metabolites in
Beltsville, Version 2.0. https://www.ars.usda.gov/ARSUserFiles/80420525/ small-scale farmers in Chiang Mai Province. Thail. Sci. Total Environ. 407 (1),
EnvironmentalTransport/CalcPTFFiles/PTF_Manual.version_2.0.pdf. (Accessed 655e668. https://doi.org/10.1016/j.scitotenv.2008.08.044.
20 September 2017). Papiernik, S.K., Yates, S.R., Koskinen, W.C., Barber, B., 2007. Processes affecting the
Hengl, T., de Jesus, J.M., MacMillan, R.A., Batjes, N.H., Heuvelink, G.B.M., 2014. dissipation of the herbicide isoxaflutole and its diketonitrile metabolite in
SoilGrids1km - global soil Information based on automated mapping. PLoS One agricultural soils under field conditions. J. Agric. Food Chem. 55 (21),
9 (8), e105992. https://doi.org/10.1371/journal.pone.0105992. 8630e8639. https://doi.org/10.1021/jf071256s.
Hengl, T., de Jesus, J.M., Heuvelink, G.B.M., Ruiperez Gonzalez, M., Kilibarda, M., PC-Progress, 2017. Source Code for HYDRUS-1D. https://www.pc-progress.com/
Blagoti c, A., 2017. SoilGrids250m: global gridded soil information based on Downloads/Pgm_Hydrus1D/H1D_Src.zip. (Accessed 24 September 2017).
machine learning. PLoS ONE 12 (2), e0169748. https://doi.org/10.1371/journal. Pimentel, D., 1995. Amounts of pesticides reaching target pests: environmental
pone.0169748. impacts and ethics. J. Agric. Environ. Ethics 8 (1), 17e29. https://doi.org/10.
Jungbluth, F., 1996. . Crop protection Policy in Thailand - Economic and Political 1007/bf02286399.
Factors Influencing Pesticide Use. Pesticide Policy Project, Publication Series No. Rigotto, R., Vasconcelos, D., Rocha, M., 2014. Pesticide use in Brazil and problems for
5 of the Institute of Horticultural Economics Hannover and Deutsche Gesell- public health. Cad. Saude Publica 30 (7), 1360e1362. https://doi.org/10.1590/
schaft für Technische Zusammenarbeit (GTZ) GmbH, Eschborn, Germany, p. 90. 0102-311xpe020714.
https://www.ifgb.uni-hannover.de/fileadmin/eagr/EUE_files/PPP_Publicat/ Riwthong, S., Schreinemachers, P., Grovermann, C., Berger, T., 2015. Land use
Series/PPP05.pdf. (Accessed 5 February 2018). intensification, commercialization and changes in pest management of small-
Kamrin, M., 1997. Pesticide Profiles: Toxicity, Environmental Impact, and Fate. CRC holder upland agriculture in Thailand. Environ. Sci. Policy 45, 11e19. https://doi.
Press, Boca Raton, USA, p. 704. org/10.1016/j.envsci.2014.09.003.
Karickhoff, S., 1981. Semi-empirical estimation of sorption of hydrophobic pollut- Safford, H.D., Sawyer, S.C., Kocher, S., Hiers, K.J., Cross, M., 2017. Linking knowledge
ants on natural sediment and soils. Chemosphere 10, 833e846. https://doi.org/ to action: the role of boundary spanners in translating ecology. Front. Ecol.
10.1016/0045-6535(81)90083-7. Environ. 15, 560e568. https://doi.org/10.1002/fee.1731.
Kastens, B., Newig, J., 2008. Will participation foster the successful implementation Schreinemachers, P., Sringarm, S., Sirijinda, A., 2011. The role of synthetic pesticides
of the water framework directive? The case of agricultural groundwater pro- in the intensification of highland agriculture in Thailand. Crop Prot. 30 (11),
tection in northwest Germany. Local Environ. 13, 27e41. https://doi.org/10. 1430e1437. https://doi.org/10.1016/j.cropro.2011.07.011.
1080/13549830701581713. Shangguan, W., Hengl, T., de Jesus, J.M., Yuan, H., Dai, Y., 2017. Mapping the global
Ko€hne, J., Ko

€ hne, S., Sim unek, J., 2006. Multi-process herbicide transport in struc- depth to bedrock for land surface modeling. J. Adv. Model. Earth Syst. 9, 65e88.
tured soil columns: experiments and model analysis. J. Contam. Hydrol. 85, https://doi.org/10.1002/2016MS000686.
1e32. https://doi.org/10.1016/j.jconhyd.2006.01.001. Shrestha, S., Datta, A., 2015. Field measurements for evaluating the RZWQM and
Ko€hne, J., Ko

€ hne, S., Sim unek, J., 2009. A review of model applications for structured PESTFADE models for the tropical zone of Thailand. J. Environ. Manage 147,
soils:b) Pesticide transport. J. Contam. Hydrol. 104, 36e60. https://doi.org/10. 286e296. https://doi.org/10.1016/j.jenvman.2014.09.017.
1016/j.jconhyd.2008.10.003.

Sim

unek, J., Sejna, M., Saito, H., Sakai, M., van Genuchten, M., 2012. The HYDRUS-1D
Kotz, J., Treichel, P., Townsend, J., 2008. Chemistry and chemical reactivity, seventh Software Package for Simulating the One-dimensional Movement of Water,
ed., 2. Brooks Cole, South Melbourne, Australia, p. 795. Heat, and Multiple Solutes in Variably Saturated media. HYDRUS Software Se-
Ladu, J.L.C., Zhang, D., 2011. Modeling atrazine transport in soil columns with ries 3. Technical report. Dept. of Environmental Sciences, University of Cali-
HYDRUS-1D. Water Sci. Eng. 4, 258e269. https://doi.org/10.3882/j.issn.1674- fornia, Riverside, USA, p. 338, Version 4.15. https://www.pc-progress.com//
2370.2011.03.003. Downloads/Pgm_Hydrus1D/HYDRUS1D-4.15.pdf. (Accessed 12 February 2018).
Leiva, J.A., Nkedi-Kizza, P., Morgan, K.T., Kadyampakeni, D.M., 2017. Imidacloprid

Sim

unek, J., Jacques, D., Langergraber, D., Bradford, S., Sejna, M., van Genuchten, M.,
transport and sorption nonequilibrium in single and multilayered columns of 2013. Numerical modeling of contaminant transport using HYDRUS and its
Immokalee fine sand. PLoS One 12 (8), e0183767. https://doi.org/10.1371/ specialized modules. J. Indian Inst. Sci. 93 (2), 265e284. http://journal.library.
journal.pone.0183767. iisc.ernet.in/index.php/iisc/article/download/1224/2474. (Accessed 5 February
Marquez, D., Faúndez, C., Aballay, E., Haberland, J., Kremer, C., 2017. Assessing the 2018).
vertical movement of a nematicide in a sandy loam soil and its correspondence SoilGrids, 2017. An Automated System for Global Soil Mapping. http://soilgrids.org.
using a numerical model (HYDRUS 1D). J. Plant Nutr. Soil Sci. 17 (1), 167e179. (Accessed 1 March 2017).
https://doi.org/10.4067/S0718-95162017005000013. Tawatsin, A., Thavara, U., Siriyasatien, P., 2015. Pesticides used in Thailand and toxic
Müller, M., 1982. Selected Climatic Data for a Global Set of Standard Stations for effects to human health. Med. Res. Arch. 3, 1e10. https://doi.org/10.18103/mra.
Vegetation Science. Springer Netherlands, The Hague, The Netherlands, p. 334. v0i3.176.
https://doi.org/10.1007/978-94-009-8040-2. Thai Meteorological Departement, 2014. Weather. https://www.tmd.go.th/en/.
Nel, J.L., Roux, D.J., Driver, A., Hill, L., Maherry, A.C., Snaddon, K., Petersen, C.R., Accessed: 2016-January0-5.
Smith-Adao, L.B., Van Deventer, H., Reyers, B., 2016. Knowledge co-production Tirado, R., Englande, A., Promakasikorn, L., Novotny, V., 2008. Use of Agrochemicals
and boundary work to promote implementation of conservation plans. Con- in Thailand and its Consequences for the Environment. Technical Note 03/2008.
serv. Biol. 30, 176e188. https://doi.org/10.1111/cobi.12560. Greenpeace Research Laboratories, p. 19. http://www.greenpeace.to/
Noshadi, M., Foroutani, A., Sepashkah, A., 2017. Analysis of clodinafob-propargyl publications/GPSEA_agrochemical-use-in-thailand.pdf. (Accessed 5 February
herbicide transport in soil profile under vetiver cultivation using HYDRUS-1D 2018).
and modified PRZM-3 models. Toxicol. Open Access 3 (120). https://doi.org/ Trapp, S., Matthies, M., 1998. Chemodynamics and Environmental Modeling. An
10.4172/2476-2067.1000120. Introduction. Springer, Berlin, Germany, p. 282. https://doi.org/10.1007/978-3-
PAN, 2016. PAN international List of Highly Hazardous Pesticides (PAN List of HHPs). 642-80429-8.
Technical report. Pesticide Action Network International (PAN). http://www. Watts, M., 2011. Paraquat. Pesticide Action Network Asia & The Pacific, p. 43. http://
pan-germany.org/download/PAN_HHP_List_161212_F.pdf. (Accessed 5 wssroc.agron.ntu.edu.tw/note/Paraquat.pdf. (Accessed 5 February 2018).
February 2018).