United States
Environmental Protection
Agency
Environmental Research
Laboratory
At he nsG A 30613
Research and Development
EPA-600/S3-84-055 May 1984
vyEPA Project Summary
Field-to-Stream Transport of
Agricultural Chemicals and
Sediment in an Iowa Watershed:
Part II. Data Base for Model
Testing (1979 - 1980)
HP. Johnson and J.L. Baker
In a continuation of a previous
project, data were collected on the
field-to-stream transport of sediment,
nutrients, and pesticides in an agricul-
tural watershed. These data contribute
to an improved qualitative understand-
ing of the field-to-stream processes
involved and provide a quantitative
base for testing mathematical models
that predict hydrology, erosion, and
sediment and chemical transport.
During the study reported here
(1979-1980), data were collected for
small corn, soybean and pasture fields;
for two larger mixed-cover sub-water-
sheds; and at three drainage stream
sites. In 1979, annual rainfall (1009
mm) was well above the long-term
average (823 mm), with several intense
rainstorms occurring in June and July.
As a result, stream flow (445 mm) was
more than twice the normal amount,
and sediment losses from the row-crop
field sites were very high (average of
63.3 t/ha). Soil loss from the watershed
as a whole was 7.6 t/ha in 1979. In
1980, precipitation (744 mm) and
stream flow (182 mm) were slightly
below normal; soil loss from the
watershed was 3.8 t/ha. In December
1979, P was fall-applied in the field
sites without incorporation; as a result,
PO4-P concentrations in snowmelt and
rainfall-runoff were over 1 mg/L until
the fertilizer was soil-incorporated
using tillage.
Flow from the watershed was roughly
half subsurface flow and half surface
runoff, with about half of the surface
runoff being snowmelt. During extended
high flows between surface runoff
events, in-stream IMOa-N concentrations
were high and very similar to those in
flow from shallow subsurface tile
drains. The percentage of stream- flow
derived from subsurface drainage could
be estimated, at any given time, from
knowledge of NOa-N concentrations in
in-stream, surface and subsurface flow.
NOa-N losses from the whole watershed
in stream flow averaged 25 kg/ha,
equal to 28% of the N applied as
fertilizer.
The severe runoff-erosion events in
1979 resulted in field runoff losses of
herbicide as high as 7.2% (for metribuzin)
most of which was associated with the
water phase for the four herbicides
studied (alachlor, propachlor, cyanazine,
and metribuzin). The maximum loss
from the whole watershed in 1979 was
2.0% (for metribuzin). In 1980, the
maximum field loss was 2.8% (for
cyanazine); for the whole watershed,
maximum loss was 1.9% (for alachlor).
This Project Summary was developed
by EPA's Environmental Research
Laboratory, Athens. GA, to announce
key findings of the research project that
/s fully documented in a separate report
of the same title (see Project Report
ordering information at back).
-------
Introduction
In an effort to achieve national water
quality goals, water pollution control
activities have been directed increasingly
at agricultural nonpoint sources. This
resulted from the knowledge that control
of point municipal and industrial sources
alone would not allow the goals to be
reached, particularly in predominantly
agricultural areas such as Iowa. In
addition, the increasing role of agriculture
in our national economy and international
trade has resulted in more intensive
agricultural management to increase
production. Consequently, more land,
which is usually less suitablefor cropping
because of poor soils or higher slopes, is
being put into production. Also, chemical
inputs are being increased to produce
higher yields on currently cropped land.
The study watershed illustrates these
last two factors. Between 1970 and
1980, the percentage of the study
watershed in row-crops increased from
55 to 80%; land in pasture, hay, grass,
oats, government set-aside, and wood
lots was reduced. Herbicides were
applied to 40% of the watershed in 1970
and to 80% in 1980; nitrogen fertilizer
use increased 2.3 times in this period,
due to the increased area of row-crops,
increased percentage of row-crop area
treated, and increased application rates.
Although increased erosion and agri-
cultural chemical losses are unintended
side effects of the highly productive
agricultural systems, research has
demonstrated that management practices
can be used to help control these
undesirable effects. The concept of Best
Management Practices (BMPs) was
developed as the primary means of
controlling agricultural nonpoint sources
of pollution. These practices are to be
effective and technically feasible, and
socially and economically acceptable.
Practices such as the use of conservation
tillage and the installation of terraces and
grassed waterways decrease sediment
loss and sediment associated pollutants.
Others, such as soil incorporation of
chemicals, decrease chemical interaction
with overland flow and thereby decrease
chemical concentrations and losses in
surface runoff.
It is neither physically nor economically
practical to field test every potential BMP
for all agricultural chemicals and for all
possible combinations of weather and
field conditions. Therefore, work has
been undertaken to develop mathematical
models (from knowledge of physical and
chemical processes) that are capable of
predicting BMP effectiveness for different
sets of conditions. In the development of
these models, transport processes in the
field and possible chemical transforma-
tions and their impact on concentrations
and losses must be understood. In
addition, once a model has been devel-
oped, field data are necessary to test its
validity.
In 1976, Iowa State University began
the collection of field data in the Four Mile
Creek watershed in Tama County, Iowa.
Results for 1976 through 1978 were
presented in a report entitled "Field-to-
Stream Transport of Agricultural Chem-
icals and Sediment in an Iowa Watershed,
Part I: Data Base for Modeling (1976-
1978)," EPA-600/3-82-032. This report
(Part II) presents the 1979 and 1980 field
data on runoff and sediment and chemical
losses for three small, single cover fields
(including soil sampling data); two mixed
cover, intra-basin sub-watersheds; and
three stream stations. Watershed inven-
tory and weather data are included,
together with data on sediment sizes,
sediment deposition, and the stream
channel as a sediment source.
As a culmination of this project, a
national conference was held in Ames,
Iowa, in 1981 to gather and disseminate
information on the state-of-the-art with
respect to agricultural nonpoint source
pollution problems and their management.
Twenty-five papers were presented
covering work from various universities,
government agencies, and practicing
engineering groups.
Results
The inventory data for Four Mile Creek
watershed, presented in Table 1, show
that land planted to row-crops (corn and
soybeans) accounted for 80% of the
watershed area for 1979 and 1980, an
increase of about 3% from the 1976 to
1978 study period, and an increase of
25% from 1970. The percentage of corn
fertilized increased from 88% in 1970 to
99% in 1980, for soybeans the proportions
were 4 and 28%, respectively. Application
of N on the whole watershed increased a
factor of 2.3 times in those ten years, as a
result of increased percentage of corn
fertilized, increased area planted to corn
(39 to 56%), and increased application
rate (123 to 178 kg/ha). Although
application rates of P on corn and
soybeans (and K on soybeans) decreased,
the area fertilized increased substantially,
so application of P to the whole watershed
increased 1.5 times (1.8 times for K).
Herbicide use on corn and soybeans
increased from 1970 (70 and 75%,
respectively) to the point in 1980 when
Table 1. Four Mile Creek Watershed In-
ventory
1970" 1979 1980
Corn !% area)
fertilized (%>
N (kg/ ha)
PsOs (kg/ ha)
herbicide (%)
insecticide f%)
Soybeans (% area}
fertilized (%)
P20$ Ikg/ha)
herbicide <%)
39
88
123
71
71
54
16
4
76
75
50
98
181
61
98
78
30
26
52
99
56
99
178
62
99
70
24
28
57
100
*Values have been revised since Part I report.
over 99% of the row-crop area received
herbicide treatment. Five herbicides,
alachlor, atrazine, butylate, cyanazine
and 2,4-D, represented at least 90% by
weight of herbicides used on corn. For
soybeans, the five herbicides, alachlor,
bentazon, chloramben, metribuzin and
trifluralin, represented at least 90% by
weight of herbicides used. Insecticide
use increased from 54% of the corn area
treated in 1970 to 70% in 1980; soybeans
received no insecticide. Five insecticides,
carbofuran, chlorpyrifos, fonofos, phorate
and terbufos, represented over 95% by
weight of the insecticide used.
With respect to tillage, the biggest
change from 1976 to 1980 came with
substitution of use of a disk or chisel for
use of the moldboard plow for primary
tillage. In 1976, 51, 38 and 11% of the
cropland (corn, soybeans, oats, hay and
pasture) were moldboard plowed, disked
and chisel plowed, respectively. In 1980,
the corresponding figures were 16, 54
and 28% (there was 1% buffalo-till and
less than 1% no-till). In 1976,0.5% of the
cropland was terraced; in 1980, 3% was
terraced. Contouring increased from 6%
of the row-cropped land in 1976 to 19% in
1980.
As shown in Table 2, precipitation in
the watershed during the study period
varied significantly from the average
yearly precipitation of about 823 mm for
the area. In 1979, precipitation in the
watershed was 186 mm above the average
and, in 1980, 79 mm below average. Not
only was the rainfall amount in 1979
above average, rainfall intensities at
individual rain gages within the watershed
registered four particularly severe events
in June and July (13 events total) with
return intervals from 5 to 100 years for
different durations. This rainfall, coupled
with a soil profile well filled with moisture
in the fall of 1978, resulted in large
amounts of runoff. About 45% of the total
stream flow in the 5-year study (1976 to
1980) occurred in 1979. Although rainfall
was below average in 1980, there were
-------
Table 2. Nutrients and Sediment in Precipitation, Surface Runoff, Tile, and Creek Flow
Precipitation
Runoff
Corn:
Site 1
Site 2"
Soybeans:
Site 2
Site T
Pasture:
Site 3
Tile drainage
Intra basin
Site 7
284 ha
Site 8
149 ha
Creek
Site 6
345 ha
Site 5
3575 ha
Site 4
5055 ha
Year
1979
1980
1979
1980
1979
1980
1979
1980
1979
1980
1979
1980
1979
1980
1979
1980
1979
1980
1979
1980
Amount
mm
1009
744
251.5
119.6
199.3
88.4
66.1
45.3
111.4
92.4
137.4
74.0
394.3
143.1
422.7
179.2
444.6
182.4
Nh
ppm
0.59
0.70
0.34
0.52
0.14
0.52
0.31
047
0.12
0.08
0.52
1.02
0.22
0.36
0.20
0.63
0.54
0.70
0.52
0.51
kg/ha
5.97
5.24
0.86
0.62
0.28
0.46
0.2}
0.21
0.57
0.94
0.30
0.26
0.72
0.90
2.29
1.25
2.33
0.93
NOz-N
ppm kg /ha
0.6
0.8
2.2
1.3
1.0
1.3
1.0
1.1
12.3
11.1
3.5
3.4
2.1
1.5
8.8
6.1
8.9
7.1
8.0
6.3
6.3
5.7
5.7
1.6
2.0
1.1
.6
.5
3.9
3.1
2.8
1. 1
30.7
8.8
37.7
12.7
35.4
11.5
PO
ppm
0035
0.063
0.096
0.723
0.120
1.512
0.787
0.930
0.090
0.082
0.671
0.808
0.293
0.328
0.115
0.318
0.248
0.209
0.155
0.141
t-P
kg/ha
0.357
0.467
0.242
0.865
0.240
1.336
0.520
0.421
-
0.748
0.746
0403
0.243
0.402
0.456
1.048
0.375
0.689
0.258
Cl
ppm kg/ha
1.6
0.8
4.0
10.2
3.6
20.4
3.7
8.2
16.1
19.0
8.8
7.6
4.8
7.4
11.2
13.5
14.2
14.0
12.0
13.2
16.2
6.1
10.1
12.2
72
18.1
2.4
3.7
9.8
7.0
6.7
5.5
39.1
19.3
59.9
25.0
53.3
24.1
TDS
ppm kg /ha
7
11
69
88
91
133
83
63
316
331
135
148
96
97
230
214
267
221
220
247
74
82
175
105
180
118
55
29
-
150
137
132
72
804
306
1129
396
977
450
Sediment
ppm kg/ha
.
.
20424
9245
37771
2458
64
30
1100
6034
13769
8828
4328
2046
1693
2075
1712
2062
51369
11061
75272
2172
42
14
1225
5576
18914
6534
15120
2954
7156
3718
7612
3760
'Sites 1 and 2 were fall fertilized before the 1980 growing season; fertilizer was incorporated in the spring by chisel plowing on site 1 and disking on
site 2.
eight runoff events in that year. In late
1979, the soil profile was wetter than in
the fall of 1978. This, coupled with the
significant rainfall events that occurred
during the 1980 growing season, resulted
in runoff from the watershed.
For all but the four extreme events of
1979, the field that had been in corn the
previous year and had been spring-
plowed had the least runoff (or in some
cases no runoff when there was runoff
from the other row-cropped site). In
addition, the plowed field (in soybeans)
was cultivated once in June and once in
July each year, but the corn field was not
cultivated. For the four extreme events in
June and July 1979, runoff volumes
from sites 1 and 2 were nearly identical,
seemingly independent of previous or
recent tillage, crop or crop canopy, or
watershed topography. For these events,
runoff ranged from 20 to 59% of precipi-
tation.
The portion of stream flow during storm
events that was subsurface flow was
determined by an interpolation technique
between the time of beginning of runoff
and the time runoff was calculated to
have ended. Because there was such a
large difference between NOa-N, Cl and
TDS concentrations in stream flow which
was all subsurface drainage (very similar
to concentrations in the tile drainage
water) and in surface runoff, knowledge
of the concentrations in the total stream
flow at any given time could be used to
estimate the portion of stream flow
attributable to subsurface drainage and
(or) surface runoff at that time.
The four severe events in 1979 caused
severe erosion and sediment transport.
As also evidenced in 1976 to 1978,1979
flow-weighted sediment concentrations
for the larger events were generally
greater for the soybean cropped field
(moldboard plowed before planting) than
for the corn field (disked before planting).
In 1980, when the soybean cropped field
was chisel plowed, 33% residue cover
remained after planting, and sediment
concentrations were less than those for
the corn field, which had been disked and
had only 8% residue cover after planting.
Much less rainfall-runoff occurred from
the chisel plowed field, resulting in soil
losses only one-fifth of those from the
disked field. For all events analyzed, the
sediment load decreased as sediment
moved from field (sites 1 and 2) to intra-
basin (sites 7 and 8) to stream (site 4).
Annual nutrient loss data and nutrient
amounts deposited with precipitation for
1979 and 1980 are presented in Table 2
together with annual flow-weighted
concentrations and arithmetic average
concentrations of nutrients in tile drainage
water. During snowmelt, NH4-N concen-
trations in runoff from the row-cropped
fields were somewhat higher than
concentrations later in the growing
season. One of the differences between
snowmelt runoff and later rainfall runoff
4s the degree of contact with the soil. NHa-
N concentrations in runoff from the corn
field in 1979 and 1980 were highest in
runoff for the first events following N
fertilizer application, although the ferti-
lizer had been incorporated by disking.
Annual NH4-N losses from the single
cover fields and pasture were al! less than
1 kg/ha, and much less than the 5 to 6
kg/ha deposited with precipitation. Total
watershed losses were at most 2.3
kg/ha, the majority of which occurred
during snowmelt in 1979.
During snowmelt, NOs-N concentra-
tions in runoff from the three single cover
fields and pasture were very similar to
concentrations in the snowfall itself. As
evidenced by the high NOu-N concentra-
tions in tile drainage water, the leachabil-
ity of NOa-N can result in large losses. The
very close match between NCb-N concen-
trations in subsurface stream flow and in
the tile drainage water, would indicate
that during the sustained high flow be-
tween closely spaced rainfall-runoff
events, most of the stream flow consisted
of subsurface drainage from tile drains.
Annual NOa-N losses in surface runoff
averaged 2.6 kg/ha from the row-
cropped fields and less than 1 kg/ha from
-------
the pasture. This was less than the 6
kg/ha deposited with precipitation.
Annual NOa-N losses with stream flow
were much larger because they included
leaching losses.
During snowmelt in 1979, POi-P
concentrations in runoff from the row-
cropped fields were similar «0.1 ppm)
to those for rainfall-runoff events later in
the growing season after tillage and
planting. This indicates that the unincor-
porated corn and soybean residue was
releasing little, if any, P0.4-P to snowmelt
runoff. P04-P levels in snowmelt runoff
from the pasture were high (nearly 1
ppm), however, because of dead and
decaying grass, animal wastes, and
previously applied P fertilizer on the soil
surface. The high in-stream levels of P0<-
P during snowmelt probably also resulted
from these surface sources of PO<-P. The
higher concentrations in field runoff in
1980 resulted from P application in
December 1979 without soil incorpora-
tion. Losses were only about 1 kg/ha,
however, with winter rains and snowmelt
in 1980.
In general, nutrient concentrations in
sediment increased as sediment concen-
trations in runoff decreased. This would
be expected if chemical activity of
sediment increased as sediment size
decreased (greater surface area per unit
mass) and sediment size decreased as
sediment concentrations decreased. The
equation:
nutrient concentration =
a (sediment concentration)""
where a and b are empirical parameters,
fitted the nutrient and sediment concen-
tration data quite well. For the row-
cropped fields, total N and P losses were
dominated by the losses associated with
sediment.
Table 3 shows the percentages of
applied herbicides lost from the row-crop
fields and the whole watershed on an
annual basis for 1979 and 1980. For
1979, there were three particularly
severe events for which runoff from the
row-cropped fields exceeded 34 mm, and
a fourth for which runoff exceeded 12
mm. As is almost always the case, the
first significant storm after herbicide
application resulted in the largest single-
event field losses. From 63 to 93% of the
measured annual losses from sites 1 and
2 occurred during this one storm. For the
whole watershed, from 46 to 74% of the
annual losses occurred during this event.
Because of the severe events in 1979,
field runoff losses were much higher
than they were in 1976 to 1978, when
losses of the four herbicides studied
never exceeded 1% of the amounts
applied. The greatest recorded loss in
1979 was for metribuzin, when 7.2% of
that applied was lost from the field and
2.0% from the whole watershed. The
trend of lower storm losses from the
watershed as a whole than those mea-
sured at the field borders, which was
evident in the 1976 to 1978 data, was
also evident in 1979.
In 1980, storms in a 5-day period in late
spring resulted in most of the measured
herbicide runoff losses from both the field
and whole watershed. These events also
accounted for most of the rainfall-runoff
that occurred in 1980.
The May-June surface runoff amounts
for the whole watershed for 1979 and
1980 were both about 42 mm, and
annual herbicide losses were similar
(although the relative amounts of alachlor
and metribuzin lost were reversed). The
May-June surface runoff amounts for the
disked corn field (site 1 in 1979, site 2 in
1980) were also similar (51 mm in 1979,
48 mm in 1980), with cyanazine losses
somewhat less for 1980, and propachlor
losses somewhat greater. For the soybean
field, which was moldboard plowed (site
2) in 1979, but chisel plowed (site 1) in
1980, May-June runoff was much less in
1980 (14 mm) than in 1 979(48 mm), and
therefore alachlor and metribuzin field
losses were also less in 1980. The
herbicides were so much affected by
decreased runoff that the trend of larger
field losses than whole watershed losses
was reversed for alachlor and was
marginal for metribuzin. It appears that
for 1980, chisel plowing of site 1
decreased herbicide losses by decreasing
runoff as well as by decreasing erosion).
For all cases a majority of the annual
losses occurred with water.
Observations and Conclusions
For smaller, less severe runoff
events, the fields with the most
recent or more intensive tillage had
significantly less runoff.
TableS.
Year
1979
1980
Percentage of Applied Herbicides Lost
Site Alachlor Metribuzin
field
4 mi watershed
field
4 mi watershed
27
1.1
0.6
1.9
7.2
2.0
1.1
0.8
Propachlor
0.6
0.6
0.7
0.4
Cyanazine
5.6
1.7
2.8
1.6
For the severe runoff events of 1979
(runoff total 120mm), runoff amounts
from both row crop fields for each
event were nearly identical, despite
differences in timing and degree of
tillage and in crop canopy.
Surface runoff from the pasture was
mostly snowmelt; runoff from rainfall
occurred only during intense rain
storms.
Flow from the whole watershed was
roughly half subsurface flow and
half surface runoff. Half or more of
the surface runoff was snowmelt
runoff.
During extended high flows between
surface runoff events, high instream
IMOs-N concentrations were very
similar to those in monitored shallow
subsurface tile drains, implying that
much of the stream flow at these
times was from the drains. During
low flow winter conditions, in-
stream NOs-N concentrations were
much lower.
For the study conditions of this
watershed, the difference between
concentrations of NOa-N, Cl and
IDS in subsurface flow (measured
in tile drainage water) and concen-
trations in field surface runoff
indicates that these data could be
used to predict the percentage of
surface runoff in stream flow.
Generally, sediment concentrations
in runoff from the soybean field
(corn residue incorporated by mold-
board plowing) were appreciably
greater than for the corn field
(disked soybean residue); however,
the opposite was true when a chisel
plow was used to incorporate the
corn residue.
For all rainfall-runoff events analyzed,
as runoff flowed from field to intra-
basin station to the main channel,
the sediment load decreased on a
unit-area basis. For snowmelt,
however, field losses were less than
stream losses on the unit-area
basis.
In the long term, the stream channel
is becoming deeper and wider,
thereby providing a source of sedi-
ment.
For rainfall-runoff from row-cropped
fields not recently fertilized, NHi-N
concentrations in runoff were less
than in precipitation because of
extraction by adsorption to the soil.
Because of the large volumes and
the high NOa-N concentrations of
subsurface drainage, NOs-N losses
from the watershed in stream flow
-------
(1979 and 1980) averaged 25 kg/ha,
28% of the applied N.
Concentrations of PCu-P in winter
surface runoff andsnowmeltfollow-
ing fall P application to the row-
cropped fields without incorporation,
were higher by a factor of 10 to 15
times than when P had not been
applied, although losses were only
about 1 kg/ha.
Total N and P concentrations in
sediment in runoff samples increased
as the sediment concentration
decreased. The equation: nutrient
concentration = a (sediment concen-
tration)"0, where a and b are empirical
parameters, fitted the data quite
well.
The severe runoff-erosion events in
1979 resulted in herbicide field
runoff losses as high as 7.2%, most
of which was with water for all four
herbicides studied; maximum loss in
1980 was 2.8%. Losses from the
whole watershed were less: a
maximum of 2.0% in 1979 and 1.9%
in 1980.
1 Chisel plowing, by reducing runoff
and erosion, in 1980 reduced herbi-
cide field runoff losses of alachlor to
below those for the whole watershed
(on a percent of applied basis);
whereas in the previous four years
of moldboard plowing, field alachlor
losses were greater.
1 Concern for (and modeling of)
pollutants transported with subsur-
face drainage needs to be empha-
sized, along with that for surface
runoff, in cases where volume of
subsurface drainage is significant.
i The factors important in determining
the effect of recent tillage, including
cultivation, on runoff volumes, and
how this effect declines with time
(or precipitation), need to be defined.
i Factors determining sediment deposi-
tion within an agricultural watershed,
which is important in determining
the sediment delivery, need to be
quantified.
> The possibility that a cycle exists
whereby sediment is deposited in
water courses during lesser events
to be eroded during high flows, such
as snowmelt, needs further study.
» Additional analyses of chemical-
sediment partitioning and enrich-
ment relative to sediment particle
size should be performed.
> Better management systems for the
increasing amounts of nitrogen
applied to crops need to be developed
and implemented to decrease the
environmental, as well as economic
and energy concerns associated
with NOa-N leaching losses.
The problems of chemical application
with conservation tillage (e.g.,
incorporation of nutrients without
incorporation of residue, or possible
runoff and volatilization losses of
herbicides applied to crop residues)
need to be solved to obtain the
greatest benefits from this increas-
ingly accepted practice.
H. P. Johnson and J. L. Baker are with Iowa State University. Ames. IA 50011.
C. N. Smith is the EPA Project Officer (see below).
The complete report, entitled "Field-to-Stream Transport of Agricultural Chem-
icals and Sediment in an Iowa Watershed: Part II. Data Base for Model Testing
(1979-1980)."(Order No. PB84-1 77419; Cost: $34.00. subject to change) will
be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield. VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Environmental Research Laboratory
U.S. Environmental Protection Agency
College Station Road
Athens, GA 30613
-------
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use $300
/i'UGJC'&S !':.'",,£ !' S
li3LV&-f r---- :9-v? V T*
/» /- ; "-- - - ', 4
\. r-. -» .;-'. > (.-.:..'£.- . ' X
^v>^"..-i ;,V-,-' .. .V ., ^-_.. -, t v :^-( , - ^
U.S. GOVERNMENT PRINTING OFFICE: 1984-759-102/965
------- |