Effects of Watershed Scale on Agrochemical ... - ACS Publications

Effects of Watershed Scale on Agrochemical. Concentration Patterns in Midwestern Streams. David B. Baker and R. Peter Richards. Water Quality Laborato...
0 downloads 0 Views 2MB Size
Chapter 3

Effects of Watershed Scale on Agrochemical Concentration Patterns in Midwestern Streams

Downloaded by UNIV MASSACHUSETTS AMHERST on September 20, 2012 | http://pubs.acs.org Publication Date: April 3, 2000 | doi: 10.1021/bk-2000-0751.ch003

David B. Baker and R. Peter Richards Water Quality Laboratory, Heidelberg College, Tiffin, OH 44883

For selected Lake Erie Basin tributaries, detailed studies of nutrient and sediment runoff have been underway since 1974 and of pesticide runoff since 1983. The monitoring stations subtend watersheds ranging in size from 11 km to 16,400 km and having similar land use and soils. Examination of the agrochemical concentration and loading patterns at these stations reveals systematic changes related to watershed size (scale). As watershed size increases, peak storm event concentrations decrease while the durations of mid-range concentrations lengthen. The extent of these scale effects is parameter specific, being most evident for suspended solids. We hypothesize that these scale effects are attributable to the pathways and timing of pollutant movement from fields into streams, coupled with dilution associated with routing of runoff water into and through the stream system from differing positions in the watershed. These scale effects need to be considered when comparing concentration and loading data from different watersheds and when designing sampling programs. 2

2

The concept of "watersheds" can be applied at spatial scales ranging from less than 1 m boxes in laboratory settings to major continental river basins, such as the Mississippi with a drainage area of 3.2 million km . Studies of runoff of water, suspended sediments, nutrients, and pesticides have occurred throughout the above range of watershed sizes. Examination of the resulting data reveals systematic shifts in a variety of hydrologie, concentration, and loading characteristics that are related to watershed size. These shifts can be considered scale effects. Many factors in addition to watershed size contribute to observed differences in runoff characteristics among watersheds (/). Watershed shape and topography influence hydrology. Watershed land use has major impacts on runoff characteristics. Land use may vary greatly among watersheds, in relation to variations in soil and water resources and other geographic factors. Small agricultural watersheds, such as research plots or individual fields, generally have a single land use, crop, and set of management practices. As the size of agricultural watersheds increases, the diversity of crops and sets of management practices within its boundaries also increases. As watershed size increases still further, tributaries draining watersheds having predominantly urban or forested land uses may join those draining agricultural 2

2

46

© 2000 American Chemical Society

In Agrochemical Fate and Movement; Steinheimer, T., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

47 watersheds, resulting in changing hydrologie, concentration and loading characteristics. These and many other factors influence runoff characteristics, and consequently confound efforts to examine the effects of watershed scale (size) per se. In northwestern Ohio, our laboratory has monitored a set of watersheds ranging in size from 11.3 km to 16,395 km for up to 23 years. These watersheds have very similar land uses (Table 1) and have generally similar soils. Point sources contribute very little to the total watershed outputs, even for the large watersheds. Consequently, they do allow an evaluation of scale effects with minimal confounding by other variables which influence runoff characteristics and stream concentrations. In this paper, we describe scale effects on hydrology, concentration durations, and loading patterns of pollutants derived from agricultural land use. The magnitude of scale effects varies systematically among parameters. We hypothesize that these variations are derived from the differences among parameters in export pathways and timing as pollutants are delivered from the land into stream systems. These effects interact with increasing dilution of local runoff by increases in channel storage as streams become larger. This dilution effect is magnified by the downstream movement of the flood front as a kinematic wave. Most of the scale effects described below have been noted in various reports and papers published by our laboratory (2-5). These same scale effects, especially as they relate to pesticide concentrations, have been noted in USGS publications (6). This paper focuses on scale effects, expanding our previous analyses and offering hypotheses regarding the generation of these scale effects in river transport.

Downloaded by UNIV MASSACHUSETTS AMHERST on September 20, 2012 | http://pubs.acs.org Publication Date: April 3, 2000 | doi: 10.1021/bk-2000-0751.ch003

2

2

Methods All of the sampling stations identified in Table I are located at or near USGS continuous stream gauging stations. Submersible pumps located just above the stream bottom pump water continuously from the streams into the sampling stations. Refrigerated automatic samplers containing 24 polypropylene bottles and located in the gage houses are used to collect three samples per day, with the sampler pumps taking water supplied to the gage house by the submersible pumps. At weekly intervals, sample bases are changed and samples are returned to the laboratory for analyses. During runoff periods, all three samples for each day are analyzed, while during non-runoff periods, a single sample per day is analyzed. During the pesticide runoff season (April 15 - August 15) additional automatic samplers containing glass bottles are used to collect samples for pesticide analyses. During the "non-runoff" season, grab sampling techniques are used to collect pesticide samples twice per month. The analytical program for each sample includes suspended solids, total phosphorus, soluble reactive phosphorus, nitrate plus nitrite, nitrite, ammonia, total Kjeldahl nitrogen, sulfate, chloride, silica and conductivity. Automated, colorimetric procedures are used for the analyses of the above nutrients (8). From 1983 through 1992, pesticides were analyzed using dual column GC with nitrogen-phosphorus detectors. Beginning in 1993, pesticides were analyzed by GC/MS. In 1993, the laboratory shifted from liquid-liquid extraction to solid phase extraction. Pesticide analytical procedures have generally followed the methods outlined in successive versions of EPA Method 507. More details on the sampling and analytical procedures can be found elsewhere (3,5). Results and Discussion Storm Hydrographe. The effects of watershed size on unit area hydrographs are illustrated in Figure 1. For Rock Creek, the peak runoff rates during storms, in mm per acre, are much higher than for the Maumee River. The duration of the individual storms is also much shorter for Rock Creek than for the Maumee River. In Rock Creek flows return to base flovj between successive storms more often than for the Maumee River. MÎÏÏëflâOI CfaâQUttt SotiÉtjf

Library 1155 Fate 16tfiSUfcW. In Agrochemical and Movement; Steinheimer, T., et al.;

ACS Symposium Series; American D Chemical Society: Washington, DC, 2000. Washington, £ 20036

48 Table I. Monitoring program for agricultural watersheds in the Lake Erie Basin USGS Basin Nutrient Pesticide Period Land use Tributary Station Area (percent) samples samples of Number (km ) record C Ρ F W Ο analyzed analyzed 1,093 9,532 MaumeeR. 04193500 16395 1976-1978 76 3 8 4 9 1982-1997 Sandusky R. 04198000 3240 1975-1997 80 2 9 2 7 10,689 1,214 Honey Cr. 04197100 386 1,328 1976-1997 83 1 10 1 6 11,393 1,186 RockCr. 04197170 88 8,928 1983-1997 81 2 12 1 4 6,636 720 LostCr. 04195440 11.3 1982-1993 83 0 11 1 5 Land use categories indicate percent of basin in : C, cropland; P, pasture; F, forest; W, water/wetland; O, other. Data from (7).

Downloaded by UNIV MASSACHUSETTS AMHERST on September 20, 2012 | http://pubs.acs.org Publication Date: April 3, 2000 | doi: 10.1021/bk-2000-0751.ch003

2

1

For the Maumee River, the Sandusky River, Honey Creek and Rock Creek, we have analyzed sets of individual storm runoff events through the 1995 water year. The resulting data base included 226 storms for the Maumee River, 237 for the Sandusky River, 261 for Honey Creek, and 158 for Rock Creek. Data for Lost Creek were not analyzed in this manner. In Figure 2, various storm characteristics are compared for the four rivers. Figure 2a illustrates changes in hydrograph shape, as described by the ratio of peak flow to storm volume. As the watershed size decreases, this ratio increases. Figure 2b shows the duration of storm events in relation to watershed size. Storm durations decrease as watershed size decreases. Figure 2c shows the distribution of the ratios of peak flow to basin area (peak unit area discharge). As watershed size decreases the peak unit area discharges increase. The above hydrologie scale effects are well known and described in great detail in the hydrological literature (9-11). Pesticide Concentration Patterns. Although it has the smallest watershed, Lost Creek was not used for these comparisons because its period of record is shorter than that of the other stations, half the pesticide samples were concentrated in the four-year interval 1984-87, and several major pesticide runoff events have occurred since the station was discontinued. The effects of watershed size on annual herbicide chemographs are illustrated using 1989 data for atrazine in the Maumee River and Rock Creek (Figure 3). Peak atrazine concentrations were higher in Rock Creek than in the Maumee, with concentrations falling to near baseline between storms for Rock Creek. In contrast, the Maumee had one broad peak of atrazine concentrations. The distribution of annual maximum concentrations of atrazine and metolachlor for the Maumee River, the Sandusky River, Honey Creek and Rock Creek are shown in Figure 4. The graphs cover the 14 year period from 1983 to 1996. Annual maximum atrazine and metolachlor concentrations increase as watershed size decreases. The exception is that Rock Creek has lower maximum concentrations than the larger Honey Creek. This is because Rock Creek is less suitable for growing corn than Honey Creek, and consequently the amount of atrazine applied in the Rock Creek basin is disproportionately low. The distribution of atrazine and metolachlor concentrations within the lowest quartile of values during the May through August period of 1983-1996 is shown in Figure 5. As watershed size increases, the concentrations of samples within the lowest quartile of values for the time period increases. Concentration exceedency curves for atrazine and metolachlor for the Maumee River and Rock Creek are shown in Figure 6. These graphs include the 40% of the time with the highest concentrations for each river and cover the time period from water years 1985 through 1996. The highest concentrations are greater in Rock Creek. However, the concentrations exceedency curves cross, demonstrating that In Agrochemical Fate and Movement; Steinheimer, T., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

Downloaded by UNIV MASSACHUSETTS AMHERST on September 20, 2012 | http://pubs.acs.org Publication Date: April 3, 2000 | doi: 10.1021/bk-2000-0751.ch003

49

Maumee River

Rock Creek

, 1.1, 30

60

JLLI. 90

120 150 180 210 240 Days after October 1, 1989

270

300

330

Figure 1. Annual unit area hydrographs for the Maumee River and Rock Creek, 1989 water year.

In Agrochemical Fate and Movement; Steinheimer, T., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

360

50

500

à 250 f ο 1 2 5

•s

t

0

: 0* 1

0

Τi

i

Maumee Sandusky Honey Rock

40'

t •w 30G O 3

û

20+

ça

10+

X

B

00

O'

Maumee Sandusky Honey Rock

2015+ (D

Ar

Downloaded by UNIV MASSACHUSETTS AMHERST on September 20, 2012 | http://pubs.acs.org Publication Date: April 3, 2000 | doi: 10.1021/bk-2000-0751.ch003

Β

o. εMen .2 5 10+ 'So «s PQ

8 o O

5+ o-

Maumee Sandusky Honey Rock

Figure 2. Effects of watershed size on peakflow/stormvolume, storm runoff duration, and peak flow to basin area. A dashed lines separating groups of boxplots indicates a statistically significant difference between the populations on either side of the line (Mann-Whitney U, p=.05, ties omitted).

In Agrochemical Fate and Movement; Steinheimer, T., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

Downloaded by UNIV MASSACHUSETTS AMHERST on September 20, 2012 | http://pubs.acs.org Publication Date: April 3, 2000 | doi: 10.1021/bk-2000-0751.ch003

51

Days since October 1, 1989 Figure 3. Annual chemographs for atrazine for the Maumee River and Rock Creek, 1989 water year.

In Agrochemical Fate and Movement; Steinheimer, T., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

Downloaded by UNIV MASSACHUSETTS AMHERST on September 20, 2012 | http://pubs.acs.org Publication Date: April 3, 2000 | doi: 10.1021/bk-2000-0751.ch003

52

ο U

Maumee Sandusky Honey Rock

Maumee Sandusky Honey Rock Figure 4. Distribution of annual maximum concentrations of atrazine and metolachlor in relation to watershed size, 1983-1996. A dashed lines separating groups of boxplots indicates a statistically significant difference between the populations on either side of the line (Mann-Whitney U, p=.05, ties omitted). Within the atrazine group on the right, Sandusky is significantly different from Honey, but the other pairwise tests are not significant.

In Agrochemical Fate and Movement; Steinheimer, T., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

Downloaded by UNIV MASSACHUSETTS AMHERST on September 20, 2012 | http://pubs.acs.org Publication Date: April 3, 2000 | doi: 10.1021/bk-2000-0751.ch003

53

0.80+

Atrazine

*Ε 0.60+ 0.40-•a g 0.20-0.00+ Maumee Sandusky Honey Rock

Maumee Sandusky Honey Rock Figure 5. Distribution of atrazine and metolachlor concentrations within the lowest quartile of values for the May through August period, in relation to watershed size, 1983-1996. A dashed lines separating groups of boxplots indicates a statistically significant difference between the populations on either side of the line (Mann-Whitney U, p=.05, ties omitted).

In Agrochemical Fate and Movement; Steinheimer, T., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

Downloaded by UNIV MASSACHUSETTS AMHERST on September 20, 2012 | http://pubs.acs.org Publication Date: April 3, 2000 | doi: 10.1021/bk-2000-0751.ch003

54

Ο

5

10

15

20

25

30

35

40

% of time concentration is exceeded Figure 6. Concentration exceedency curves for atrazine and metolachlor for the Maumee River and Rock Creek, 1985-1996.

In Agrochemical Fate and Movement; Steinheimer, T., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

55

Downloaded by UNIV MASSACHUSETTS AMHERST on September 20, 2012 | http://pubs.acs.org Publication Date: April 3, 2000 | doi: 10.1021/bk-2000-0751.ch003

intermediate concentrations in the larger Maumee River are greater than those in the smaller rivers. The curves for the Sandusky River and Honey Creek fall between those for the Maumee River and Rock Creek. Nutrient and Sediment Concentrations Patterns. At all of the sampling stations, the shape of concentration exceedency curves varies among parameters. In Figure 7, concentration exceedency curves are shown for suspended solids, total phosphorus, nitrate + nitrite, chloride and conductivity for the Maumee River and Lost Creek. In order to plot parameters with differing concentrations on the same graph, each parameter was expressed as a percent of the concentration exceeded 0.5% of the time (99.5 percentile concentrations) for the period of record at that station. Suspended solids concentrations drop off much more quickly, relative to their 99.5 percentile concentrations than do conductivity values. The graphs "stack up" by parameter in the same way for both watersheds. The order from the bottom of the figure to the top is suspended solids, total phosphorus, nitrate + nitrite, chloride and conductivity. For a given parameter, the concentration exceedency curves vary with watershed size (Figure 8). For both suspended solids and nitrate + nitrite, the smallest watershed (Lost Creek) has the steepest drop from its 99.5 percentile concentration and the largest watershed (Maumee River) has the smallest drop. For both parameters the curves stack up in relation to watershed size, with the smallest watershed at the bottom of the stack. Nutrient and Sediment Loading Patterns. The temporal distribution of loading combines the effects of scale on the hydrologie response of the river to storm events the effects of scale on runoff-related concentrations patterns. Data for a particular station and parameter were ranked by instantaneous loading rate (concentration χ flow) from the highest to the lowest rate for the period of record. Loads associated with each sample were calculated (loading rate χ time). Cumulative loads and cumulative times were then calculated and expressed as a percentage of the total load and total time for the period of record. The percentage of the total load can then be plotted as a function of the percentage of the total time, as shown in Figure 9. For the Maumee River, the 20% of the time with the highest export rates of suspended solids accounted for about 90% of the total export of suspended solids while the 20% of the time with the highest chloride loads accounted for about 60% of the total chloride export. For both the Maumee River and Lost Creek the parameters stack up in a particular order, with chloride and conductivity at the bottom, followed by discharge, nitrate + nitrite, total phosphorus and suspended solids. Cumulative loading curves for the five watersheds are shown for suspended solids and nitrate + nitrite in Figure 10. For a particular parameter, the cumulative loading curves stack up in relation to watershed size. The smaller the watershed, the greater the proportion of the total load that is exported in a given percentage of time. Thus, the 5% of the time with the highest suspended solids export rates accounted for 67% of the total export rate from the Maumee River and more than 95% of the total export from Lost Creek. In Table II, the percentages of the total loads exported during the 1% of the time with the highest loading rates are shown for various parameters and streams. For suspended solids, the 1% of the time with the highest export rates accounted for 62% of the total export from Lost Creek and 32% of the total export from the Maumee River. For nitrate + nitrite, the 1% of the time with the highest export rates accounted for 20% of the total export from Lost Creek and 8% of the total export from the Maumee River.

In Agrochemical Fate and Movement; Steinheimer, T., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

Downloaded by UNIV MASSACHUSETTS AMHERST on September 20, 2012 | http://pubs.acs.org Publication Date: April 3, 2000 | doi: 10.1021/bk-2000-0751.ch003

56

Percent of time Figure 7. Concentrations exceedency curves for various nutrients, relative of 99.5 percentile concentrations, for the Maumee River and Lost Creek.

In Agrochemical Fate and Movement; Steinheimer, T., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

Downloaded by UNIV MASSACHUSETTS AMHERST on September 20, 2012 | http://pubs.acs.org Publication Date: April 3, 2000 | doi: 10.1021/bk-2000-0751.ch003

57

OH Ο

1

10

1 1 20 30 Percent of time

1

Τ­

40

50

Figure 8. Effects of watershed size on concentrations exceedency curves for suspended solids and nitrate.

In Agrochemical Fate and Movement; Steinheimer, T., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

Downloaded by UNIV MASSACHUSETTS AMHERST on September 20, 2012 | http://pubs.acs.org Publication Date: April 3, 2000 | doi: 10.1021/bk-2000-0751.ch003

58

Figure 9. Variations in percent of total load versus percent of time for various parameters for the Maumee River and Lost Creek.

In Agrochemical Fate and Movement; Steinheimer, T., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

Downloaded by UNIV MASSACHUSETTS AMHERST on September 20, 2012 | http://pubs.acs.org Publication Date: April 3, 2000 | doi: 10.1021/bk-2000-0751.ch003

59

Suspended Solids

• Maumee



Sandusky • Honey

A Rock m Lost 20 30 Percent of time 100-

50

Nitrate plus Nitrite

• Maumee • Sandusky

• Honey

m 1

1

20 30 Percent of time

Rock Lost

T"

40

10. Effects of watershed size on percent of loads versus percent of time for suspended solids and nitrate.

In Agrochemical Fate and Movement; Steinheimer, T., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

60

Downloaded by UNIV MASSACHUSETTS AMHERST on September 20, 2012 | http://pubs.acs.org Publication Date: April 3, 2000 | doi: 10.1021/bk-2000-0751.ch003

Table IL Percent of the total load accounted for by the 1 % of the time with the highest loading rates (fluxes). Suspended Chloride Watershed Discharge Sediment Nitrate Maumee 32% 11% 11% 8% Sandusky 34% 13% 14% 12% Honey 51% 16% 15% 11% Rock 58% 27% 25% 16% Lost 62% 20% 24% 18%

Hypothesized Causes of Scale Effects. We believe that many of the scale effects noted above are a direct consequence of the routing of water and chemicals through drainage networks during storm runoff events, coupled with differing pathways of chemical movement from land to streams . The pattern of concentrations and loading at a specific stream location depends on its position in the drainage network. Whenever two streams merge with differing concentrations of chemicals, the resulting concentrations of the mixed water will be intermediate to the two parent streams. The resulting mixed concentrations will depend not only on the concentration differences between the streams, but also on the flow of the streams. For mixing in flowing systems the resulting concentration is the sum of the instantaneous fluxes of the two streams divided by the sum of the instantaneous discharges. Several factors give rise to differences in concentrations and flows in two streams as they mix. Generally where streams join, their individual watersheds will have differing sizes. Consequently, even when a rainfall event occurs simultaneously over the entire watershed, the streams will be in different phases of their runoff response to the storm as they mix. Chemical concentrations vary in characteristic ways during runoff events, and peaks of chemographs may precede, coincide with, or trail the peak of the hydrograph, depending on the chemical (2,/2). In Figure 11, chemograph shapes for suspended sediments, atrazine and nitrate + nitrite are compared with the storm hydrograph for a June 1993 storm in Honey Creek. The asynchrony between peak concentration and peak flow contributes to the potential for dilution during mixing. One way to illustrate the asynchrony between chemical transport (time integrated instantaneous fluxes) and water discharge (time integrated instantaneous flow) during a storm runoff event is to plot double mass curves for individual storm events. For double mass curves, cumulative loads are plotted against cumulative discharges. To compare different parameters, the cumulative loads can be plotted as a percentage of the total storm load and cumulative discharge as a percentage of the total discharge for the storm. In Figure 12 double mass curves are shown for suspended solids, atrazine and nitrate + nitrite for the June 1993 storm event on Honey Creek (Figure 11). The first 50 percent of the storm discharge water accounted for 76% of the total storm load of suspended sediments, 58% of the total storm load of atrazine, and 36% of the total storm load of nitrate + nitrite. For suspended sediments the flow weighted average concentration of suspended solids was three times higher during thefirsthalf of the storm discharge than during the second half. The corresponding atrazine concentration was 1.38 times higher, while that for nitrate + nitrite was 1.63 times lower. Based on the distribution of these chemicals during this runoff event, the capacity for dilution during routing is greatest for suspended solids and least for atrazine. Considerable volumes of water with low sediment concentrations are present during the receding portion of the storm runoff event and considerable volumes of water with low nitrate + nitrite concentrations are present early in the runoff event. We believe that the asymmetry present in chemical transport is due to the different pathways of movement of pollutants from land surfaces to streams. Pollutants associated with surface runoff commonly exhibit a "first flush" In Agrochemical Fate and Movement; Steinheimer, T., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

Downloaded by UNIV MASSACHUSETTS AMHERST on September 20, 2012 | http://pubs.acs.org Publication Date: April 3, 2000 | doi: 10.1021/bk-2000-0751.ch003

61

11. Comparison of chemographs for suspended sediments, atrazine and nitrate with the hydrograph for a June 1996 storm in Honey Creek.

In Agrochemical Fate and Movement; Steinheimer, T., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

Downloaded by UNIV MASSACHUSETTS AMHERST on September 20, 2012 | http://pubs.acs.org Publication Date: April 3, 2000 | doi: 10.1021/bk-2000-0751.ch003

62

Ο

10

20

30

40

50

60

70

80

90

% of Total Storm Discharge 12. Percent of total load versus percent of total storm discharge for suspended sediments, atrazine, and nitrate for a June 1996 storm in Honey Creek.

In Agrochemical Fate and Movement; Steinheimer, T., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

100

Downloaded by UNIV MASSACHUSETTS AMHERST on September 20, 2012 | http://pubs.acs.org Publication Date: April 3, 2000 | doi: 10.1021/bk-2000-0751.ch003

63 phenomena. It is well known that in urban runoff studies, pollutant concentrations are particularly high early in the runoff event (first flush) (13). In field runoff studies from cropland, sediment and particulate phosphorus concentrations also have higher concentrations early in the runoff period than later. This generates the "asynchrony" in stream sediment concentrations that support scale effects through dilution. The pathways of most of the nitrate + nitrite export in this area involve tile flow and interflow. Since these flows are delayed relative to surface runoff, and persist for longer durations, most of the nitrate + nitrite is delivered to the stream late in the runoff period. The chemographs for soluble herbicides are generally much broader than the chemographs for sediments and particulate phosphorus, with peak concentrations occurring between peak sediment and peak nitrate + nitrite. The onfield kinetics of herbicide dissolution from the soil surface and upper soil layers may account for the extended delivery of pesticides from fields to streams. For some soils, pesticide contributions from tile drainage may also contribute to the broad shape of the pesticide chemograph. Another factor contributing to the observed scale effects involves increasing amounts of storage of pre-storm water in stream channels in a downstream direction. When a small order stream responds quickly to a rainfall event, and that stream enters a high order stream, the amount of pre-storm water present in the receiving stream channel generally increases in a downstream direction. Thus, there is more water present in the channel to dilute incoming local runoff, with its high peak concentrations, in a downstream direction. In addition, the kinematic wave movement of water at the flood front, as that front moves downstream, increases the volume of pre-storm water available to dilute local runoff. Conclusions and Observations 1. The patterns of chemical concentrations and transport in streams vary systematically with the position of the sampling station in the drainage network. These variations reflect a watershed scale effect and occur even though the landscape and land use may be nearly uniform throughout the drainage network. 2. Peak concentrations of chemicals derived from land runoff are generally higher in streams draining small watersheds than in streams draining large watersheds. 3. The durations of intermediate concentrations are longer for streams draining large watersheds than for streams draining small watersheds. 4. The export or loading of chemicals derived from land runoff occurs in shorter periods of time for small watersheds than for large watersheds. 5. The extent of these scale effects varies among parameters, being most evident for suspended solids. 6. We hypothesize that scale effects are largely a consequence of dilution accompanying the routing of runoff water through the drainage network. Changing channel storage with stream size and kinematic wave movement of the flood front likely also play a role in generating the observed scale effects. 7. We hypothesize that these differences in scale effects among parameters originate from differences in the pathways of movement of chemicals from land surfaces to streams. 8. Awareness of scale effects is important when comparing concentrations and loading data among streams of different sizes and when designing sampling programs aimed at characterizing either concentrations or loadings in stream systems. 9. Models aimed at predicting the concentrations of nonpoint derived pollutants in steam systems should reflect the same scale effects that are apparent in detailed long term data sets such as those we have accumulated. 10. Because of the complexity and variability of factors that interact to produce runoff within large watersheds, individual runoff events may diverge from the general patterns described above. In Agrochemical Fate and Movement; Steinheimer, T., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

64

Downloaded by UNIV MASSACHUSETTS AMHERST on September 20, 2012 | http://pubs.acs.org Publication Date: April 3, 2000 | doi: 10.1021/bk-2000-0751.ch003

Literature Cited 1. Saxton, K.E.; Shiau, S.Y. In Surface Water Hydrology; Wolman, M.G.; Riggs, H.C., Eds.; The Geological Society of America: Boulder, CO, 1990; pp 55-80. 2. Baker, D. B.; Krieger, Κ. Α.; Richards, R. P.; Kramer. J. K. In Perspectives on Nonpoint Source Pollution; EPA 440/5/85-001; U.S. EPA: Washington, DC, 1985; pp 201-207. 3. Baker, David B. Sediment, Nutrient and Pesticide Transport in Selected Lower Great Lakes Tributaries; EPA-905/4-88-001; U.S. EPA, Great Lakes National Program Office: Chicago, IL, 1988; pp 1-225. 4. Baker, D. B. Agriculture, Ecosystems and Environment 1993,46; 197-215. 5. Richards, R. P.; Baker, D. B. Environ. Toxicol. Chem. 1993, 12,13-26. 6. Fenelon, J. M. Water Quality of the White River Basin, Indiana, 1992-96; U. S. Geological Survey Circular 1150. U. S. Geological Survey, Information Services: Denver, CO, 1998; pp 1-7. 7. Resource Management Associates. Land Resource Information For the Lake Erie Drainage Basin, Co-occurrence of Land Resource Features. U. S. Army Corps of Engineers: Buffalo, NY, 1979; Vol. 2, pp 96-111, 125-138; Vol. 3, pp 180-189, 230-247. 8. US EPA. Methods for Analysis of Water and Wastes; US EPA: Cincinnati, OH, 1979. 9. Matthai, H.F. In Surface Water Hydrology; Wolman, M.G.; Riggs, H.C., Eds.; The Geological Society of America: Boulder, CO, 1990; pp 97-120. 10. Leopold, L.B. Water, Rivers and Creeks; University Science Books: Sausalito, CA, 1997; pp 39-57. 11. Huber, W.C. In Handbook of Hydrology; Maidment, D.R., Ed.; McGraw-Hill: New York, NY, 1993; pp 14.1-14.50. 12. Baker, D. B. J. Soil Water Conserv. 1985, 40, 125-132. 13. Livingston, Ε. H.; Cox, J.H. In Perspectives on Nonpoint Source Pollution, EPA 440/5/85-001; US EPA: Washington, DC, 1985; pp 289-291.

In Agrochemical Fate and Movement; Steinheimer, T., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2000.