| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Department of Natural Resources and Environmental Sciences, University of Illinois, W-503 Turner Hall, 1102 South Goodwin Avenue, Urbana, IL 61801
b Iowa Department of Natural Resources, Geological Survey Division, Iowa City, IA 52242
* Corresponding author (gmcisaac{at}uiuc.edu).
Received for publication July 31, 2002.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSCONCLUSIONSREFERENCES |
|---|
Abbreviations: DMRWQN, Des Moines River Water Quality Network • USGS, United States Geological Survey
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSCONCLUSIONSREFERENCES |
|---|
|
McIsaac et al. (unpublished data, 2002) demonstrated significant discrepancies among NO3 concentrations reported by different agencies in Illinois between 1967 and 1974. Values reported by one agency (Central Illinois Public Service) had a greater mean and variation than those reported by other agencies sampling from nearby locations, which probably reflected analytical error. Additionally, values reported by the Illinois Environmental Protection Agency between 1969 and 1971 appeared to be influenced by a conversion factor error.
The objective of this study was to reconsider the comparison of 1945 and contemporary NO3 concentrations in the Des Moines River near Des Moines in light of the following: (i) possible errors in the historical data used, (ii) variations in methods of sample collection, (iii) variations in location of sampling, and (iv) additional data collected since 1990.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSCONCLUSIONSREFERENCES |
|---|
Within the city of Des Moines, the Des Moines River is joined by the Raccoon River, which also drains the Des Moines Lobe. Below this confluence, the river flows to the east and southeast through the Southern Iowa Rolling Loess Prairie, which is of less recent glaciation than the Des Moines Lobe. Steeper slopes and more fully developed natural drainage systems reduce the need for artificial drainage. Additionally, compared with the Des Moines Lobe, a greater portion of the land in the Rolling Loess Prairie is used for hay and pasture and less is used for annual row crop production. Nitrate concentrations in the streams of this region tend to be less than those in the Des Moines Lobe (Iowa Department of Natural Resources, Geological Survey Division, 2001a), and consequently the NO3 concentrations in the Des Moines River tend to decline downstream from the Des Moines Lobe and downstream from the city of Des Moines.
Data
Hershey (1955) described, reproduced, and summarized previously published water quality data from several rivers in Iowa, including the 1945 Des Moines River NO3 concentration data. In a summary table, Hershey (1955)(p. 40) presented average concentrations of several constituents (NO3, chloride, calcium, magnesium, and sulfate) from several rivers and sampling periods. For each location, the averages were identified as either "mean," "weighted average," or "average." These designations are not clearly explained in the text. All data identified as "mean" were from samples collected in 1906–1907, for which discharge measurements were not reported, and therefore flow-weighted averages could not be calculated. All of the data identified as "weighted average" were from locations for which discharge was reported. Data identified as "average" were from two locations: Des Moines River in 1945 and the Raccoon River in 1946. Discharge data were reported for the Des Moines River but not for the Raccoon River. The "weighted average" designation may have referred to the method of compositing daily samples, such that the volume of each daily sample added to the composite was proportional to the quantity of discharge occurring on the day the sample was collected (Hershey, 1955, p. 9). Concentrations identified as "average" may have indicated that an equal volume of each daily sample was added to the composite sample regardless of the daily stream flow or it may have referred to an arithmetic average.
The 1945 average NO3 concentration for the Des Moines River at Des Moines was reported by Hershey to be 22 mg NO3 L-1, which corresponds to 5.0 mg NO3–N L-1, the value that Keeney and DeLuca (1993) used in their analysis. They had interpreted the designation of "average" to mean arithmetic average (T. DeLuca, personal communication, 2002) in contrast to the "weighted average," which they interpreted as flow-weighted averages, although these designations were not mentioned in Keeney and DeLuca (1993).
During the 1945 water year, samples for NO3 determination were collected from the 14th St. bridge, approximately 2.4 km downstream from the confluence with the Raccoon River and upstream of the Des Moines wastewater treatment plant. These data were originally published in Paulsen (1949), but methods of sample collection were not described. It is likely that samples were dipped from the surface of the river near the center of the channel, although it is possible that depth-integrated samples may have been taken with equipment developed for collecting samples for determination of sediment concentration. Samples were collected on most days and composited over a period of approximately 10 d (Paulsen, 1949). The composite samples were analyzed for NO3 using the phenoldisulphonic acid method (Collins, 1928). Additionally, on seven dates, a set of four to five samples was collected from different locations across the bridge and these samples were analyzed individually, without compositing (Hershey, 1955).
The United States Geological Survey (USGS) also reported NO3 concentration values of composited samples collected at Des Moines, upstream of the confluence with the Raccoon River from November to June during the 1955 water year (United States Geological Survey, 1958). Because the Raccoon River represents approximately 40% of the flow below the confluence with the Des Moines River, and because sampling was not conducted for a complete year, we did not include the 1955 data in our analysis.
Starting in 1975, the Des Moines River Water Quality Network (DMRWQN) collected Des Moines River water samples from the Highway 46 bridge, approximately 12 km downstream from the confluence with the Raccoon River and 4 km downstream of the Des Moines wastewater treatment plant (Lutz and Esser, 2002). In 1998, Highway 46 was closed and this sampling location was moved 1 km downstream to the Highway 65 bridge. These are considered equivalent sampling locations by the DMRWQN because there are no significant tributaries or N sources between the two locations. We also treat these as a single sampling location and refer to it as the Hwy. 46/65 bridges.
Starting in 1978, DMRWQN also collected samples near Runnells, Iowa, approximately 25 river km downstream from the Hwy. 46/65 bridges. The drainage area at Runnells is 30200 km2, which is 18% greater than the drainage area above USGS Stream Gauging Station 5485500 in Des Moines. The additional drainage area between Des Moines and Runnells is primarily from the Rolling Loess Prairie, rather than from the Des Moines Lobe.
At locations monitored by DMRWQN, dip samples were collected near the center of the channel, approximately once per week before 1982. Starting in 1982, sampling frequency was reduced to an average of approximately twice per month, with somewhat greater frequency during the growing season (April–September) and lower frequency during the winter (November–February). Samples were analyzed for NO3 using the cadmium reduction method. These data were obtained from the DMRWQN web site (Des Moines River Water Quality Network, 2002).
Daily river discharge data from 1941 to 2001 for the Des Moines River at Des Moines below the confluence with the Raccoon River (USGS Station 5485500) were obtained from the USGS National Water Information Service (NWIS) web site (United States Geological Survey, 2003). River stage and discharge were determined at the 6th St. bridge, approximately 1.2 km downstream from the confluence with the Raccoon River. The drainage area above this gauging station is 25600 km2. Concentrations of NO3 and other constituents measured by USGS during 1945 were also obtained from the NWIS web site.
Analysis and Interpretation of the Data
Using the 1945 concentration and discharge data obtained from the United States Geological Survey (2003), we attempted to reproduce the average concentrations of several constituents presented by Hershey (1955)(p. 40). We calculated arithmetic average and the flow-weighted average concentrations for several of the constituents reported by Hershey (NO3, fluoride, chloride, sulfate, calcium, and magnesium). Flow-weighted values were calculated by summing the products of the composite concentration and the average discharge during the period of respective sample collection and dividing the sum of the products by the sum of the discharges (Hershey, 1955, p. 14). Concentrations of the noncomposited samples taken from different locations across the cross-section were not included in the calculation of averages (Hershey, 1955, p. 14).
Because NO3 concentration can be a strongly related to stream flow, which may influence intertemporal comparisons, we attempted to quantify relationships between stream discharge and NO3 concentration by using nonlinear regression. First-, second-, and third-order polynomials were tested with discharge and the natural log of water yield (discharge per unit area) as the independent variable.
With time series data, there is frequently serial correlation in the observations and model residuals (predicted minus observed value), which can cause underestimation of error variance. Using the AUTOREG procedure of SAS Institute (2000), the Durbin–Watson method was used to test for the significance of residual autocorrelation. When autocorrelation was significant at the P < 0.05 level, the regression included the lagged model error term as an additional independent variable to correct for the effects of residual autocorrelation.
To evaluate the influence of compositing daily samples on the observed NO3 concentrations, we used the concentration–discharge relationship determined from the 10-d composites and the natural log of 10-d flow to estimate the expected concentration for the days during which individual samples were analyzed in 1945. Natural log of average daily flow on these dates was used as the independent variable. The estimated concentrations were then compared with the measured concentrations on those dates by linear regression and we tested whether the regression line was statistically different than the 1:1 line.
To characterize the temporal trends in discharge and NO3 concentrations during 1976–2001 at the Hwy. 46/65 bridges, we conducted multiple regression analyses with time as the independent variable, and considered first-, second-, and third-order polynomial models. Since NO3 concentration sometimes increases with discharge, we also conducted multiple regression analyses with time and natural log of discharge as independent variables to determine if any temporal trends in NO3 concentration were independent of trends in discharge. The above analysis was repeated with the 1982–2001 data to characterize trends under a consistent sampling frequency.
Because of the large gap in time between 1945 and the contemporary concentration data, we did not include 1945 in the trend analysis described above. Analysis of autocorrelation requires observations at consistent time intervals. Additionally, trend analysis would require an assumption about the shape of the trend between 1945 and the contemporary period. To compare 1945 NO3 concentrations to the contemporary value at comparable discharge, we developed best-fit relationships between annual NO3 concentration at the Hwy. 46/65 bridges and water yield during 1976–1981 and during 1982–2001. We used these best-fit relationships to estimate the expected concentration and the 95% confidence interval for individual observations at the 1945 discharge. We then compared the 1945 concentration to the expected average and the 95% confidence interval of individual observations. In developing best-fit relationships, we considered the following independent variables: annual water yield, April through September water yield, and average water yield during the previous one to four years. April through September water yield was considered for the following reasons: (i) this was thought to be the period of maximum NO3 availability in the soil, (ii) on average 68% of the annual water yield occurs during this period, and (iii) during 1982 to 2001 water sample collection was most frequent during this period. Annual water yields in previous years were considered as independent variables to test for any effects of droughts and floods that could persist for one or more years. During drought years, NO3 may accumulate in the soil or subsoil due to reduced crop production. This NO3 may contribute to riverine NO3 over several years following a drought. Conversely, one or several years of high flow may dilute or deplete the source of NO3 contamination, contributing to one or more subsequent years of relatively low NO3 concentration.
This analysis was conducted using arithmetic average annual concentrations and the flow-weighted average annual concentrations. The 1945 flow-weighted average was calculated as described above. For the contemporary samples, we used a period weighted average method (Coats et al., 2002) to estimate the annual load and divided the estimated annual load by average annual discharge. In this method, loads are calculated based on the assumption that the average of successive concentration measurements represents the concentration of the discharge that occurred between the two sample collection times.
The intertemporal comparison of concentrations may be influenced by differences in sampling location. In 1945, samples were collected upstream of the Des Moines wastewater treatment plant. The samples collected at the Hwy. 46/65 bridges in the contemporary period were collected 4 to 5 km downstream of the wastewater treatment plant. Keeney and DeLuca (1993) considered the contribution of the wastewater treatment plant, serving 350000 individuals, to be insignificant to the total N budget for the watershed for the 1980–1990 data. Based on an estimate of 3.3 kg N person-1 yr-1 in sewage effluent (Howarth et al., 1996) and 350000 residents, the wastewater treatment plant could contribute an average of 1160 Mg N yr-1 to the NO3 load at Des Moines. This represents only 2.5% of the estimated average 45800 Mg N yr-1 in NO3–N load calculated from 1976–2001 concentrations and discharge at Des Moines.
Additionally, the Saylorville Reservoir, upstream of Des Moines, was completed in 1977 and probably causes some loss of riverine NO3 due to denitrification or biotic uptake, depending on the depth and residence time (Howarth et al., 1996). Bierl (1982) concluded that the Saylorville Reservoir did not have a significant influence on the NO3 concentration in the Des Moines River. The lack of a statistical difference between riverine NO3 concentrations collected above and below the reservoir since 1982 by DMRWQN supports Bierl's conclusion (Donna Lutz, DMRWQN, personal communication, 2003). It is fortuitous for the intertemporal comparison that the Saylorville Reservoir represents a relatively small sink of NO3, while the Des Moines wastewater treatment plant represent a relatively small source of NO3.
Keeney and DeLuca (1993) compared the 1945 concentrations in Des Moines to concentrations that were collected approximately 35 km downstream at Runnells. The additional drainage area below Des Moines is of a different geological origin than the land upstream of Des Moines (the Des Moines Lobe). Streams draining the Des Moines Lobe tend to have high NO3 concentrations because of the intensity of tile drainage in that region (Iowa Department of Natural Resources, Geological Survey Division, 2001b). The contribution from tributaries below the city of Des Moines tends to cause dilution of NO3 in the main stem of the river. To determine whether the concentrations observed at Hwy. 46/65 bridges were statistically different than the concentrations observed at Runnells, we calculated the difference in concentrations between the locations for each day that both locations were sampled. We then tested whether the mean of these differences was statistically different than zero. Approximately 99% of the 684 observations at Runnells were collected on the same day as samples collected from the Hwy. 46/65 bridges. The change in average NO3 concentrations per river mile between Runnells and the Hwy. 46/65 bridges was used to estimate the difference in average concentration between the Hwy. 46/65 bridges and the 1945 sampling location.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSCONCLUSIONSREFERENCES |
|---|
|
|
|
|
During 1976–2001, there were no statistically significant (P > 0.05) temporal trends in the average annual water yield or April through September water yield, but there was a significant (P < 0.001) upward trend in discharge at sampling. This was probably due to the change in the sampling frequency after 1981, which reduced sampling of lower flows. For the 1982–2001 period, there were no statistically significant (P > 0.05) temporal trends in the average annual discharge, discharge at sampling, individual NO3 concentrations, annual arithmetic average, or flow-weighted average concentrations.
More than 80% of the variation in annual arithmetic average concentrations during 1982–2001 could be described by an equation that included the square and the cube of the natural logarithm of April through September water yield (Fig. 5) . Model residuals were normally distributed and autocorrelation was statistically significant (P < 0.05). When the lag of the residuals was incorporated into the model, the predicted values and the confidence limits were not a function of discharge alone. When this equation was used to estimate the arithmetic average concentration at the mean water yield for the 1945 water year, the expected mean value was 7.5 mg NO3–N L-1, and the 95% lower confidence limit of an individual observation at this discharge was 5.2 mg NO3–N L-1 (Table 2). Even if the effect of compositing the samples in 1945 was as large as 19% (the 95% upper confidence limit of our estimate), the equivalent arithmetic average concentration in 1945 would be 4.3 mg NO3–N L-1, which is also less than the 95% lower confidence limit of 5.2 mg NO3–N L-1. When the annual water yield was used as the independent variable, similar results were obtained but with a somewhat lower coefficient of determination (R2 = 0.75) and greater root mean square error (RSME). The observed concentration in 1945, including the upward adjustment for composite sample collection, was less than the 95% confidence interval for and individual observation at the 1945 water yield.
|
|
For the 1976–1981 period, linear models with the natural log of water yield provided the best description of the arithmetic annual average concentrations (autocorrelation was not statistically significant at P > 0.05). Using the annual water yield as the independent variable produced a higher R2 and a lower RSME than using the April through September water yield (Table 2). The 1945 arithmetic average NO3 concentration with the 19% upward adjustment was outside the 95% confidence interval for an individual observation in the 1976–1981 period when the annual water yield was used as the independent variable, and outside the 90% confidence interval when the April through September water yield was used as the independent variable. The confidence interval was larger when the April through September water yield was used as the independent variable, because the relationship has a greater RSME than when annual water yield was used as the independent variable. Including water yield during previous years as an additional independent variable did not significantly improve the explanatory power of the relationships in the 1976–1981 period.
Depending on the model used and depending on the assumed effects of composite sample analysis in 1945, the arithmetic annual average NO3 concentration in 1945 (3.6 mg NO3–N L-1) was between 44 and 57% of the expected mean value at an equivalent water yield during 1976–2001. After applying a 19% upward adjustment to the 1945 arithmetic average concentration to account for the effect of composite sample analysis, there appears to have been less than a 5% probability of observing an arithmetic annual average NO3 concentration of 4.3 mg NO3–N L-1 during 1976–2001 when water yields were equivalent to the 1942–1945 values.
A similar set of regression equations could be used to describe annual variations in flow-weighted NO3 concentrations between 1982 and 2001 (Table 3). April through September water yield provided greater explanatory power than annual water yield. Autocorrelation of the residuals was statistically significant (P < 0.05) if water yield in the previous three years was not included in the model. When this relationship was used to estimate the concentration at the 1945 average discharge (Fig. 6) , the expected mean concentration was 8.7 mg NO3–N L-1, and the 95% lower confidence limit of an estimated individual observation at this discharge was 6.1 mg NO3–N L-1, which is 22% larger than the observed value for 1945 (5.0 mg NO3–N L-1). Thus, after the 19% upward adjustment for compositing is applied to the 1945 flow-weighted concentration, the result is less than the 95% confidence limit of expected individual observations. When the annual water yield is used as the independent variable, however, the value of R2 is less, RSME is greater, and the 1945 NO3 concentration with the 19% upward adjustment for compositing was slightly (0.01 mg N L-1) greater than the 90% lower confidence limit of the expected observations (Table 3).
|
|
For the 1976–1981 period, the square and cube of the natural logarithm of annual water yield provided greater predictive power than the square and cube of the April through September water yield. Autocorrelation was not statistically significant at P > 0.05, and the 1945 concentration plus the 19% adjustment for compositing was less than the 90% confidence limit for individual observations at the 1945 water yield. The fact that 1976–1981 NO3 concentrations were more strongly correlated with annual water yield than April through September water yield suggests that the higher correlation between the 1982–2001 NO3 concentrations and April through September water yield was primarily a result of differences in the timing of sample collection rather than soil NO3 availability. The daily composite sample collection in 1945 and the weekly sampling during 1976–1981 both provide a more seasonally balanced assessment of annual NO3 concentrations than the 1981–2001 sampling approach that favored the April through September period. For this reason, the comparison between the 1945 and the 1976–1981 concentrations may be more reliable, even though it is based on fewer years of observation.
The 1945 flow-weighted average NO3 concentration during April through September was 5.3 mg NO3–N L-1. If the sampling strategy used during 1982–2001 had been used in 1945, it is likely that the resulting flow-weighted average NO3 concentration would have been between 5.3 and 5.0 mg NO3–N L-1. If we consider the maximum value (5.3 mg NO3–N L-1) and the 19% upward adjustment for compositing, the resulting value is 6.3 mg NO3–N L-1. This value lies outside the 95% confidence interval for the expected individual observations during 1982–2001 at the 1945 water yield, if the model that includes the average water yield during the three previous years is used. If the model that includes only the natural logarithm of April through September water yield is used, 6.3 mg NO3–N L-1 is inside the 95% confidence interval of expected individual concentrations for the 1982–2001 period at the 1945 discharge, but outside the 90% confidence interval of expected individual concentrations. This result is consistent with the result obtained from the comparison with the best fit model of the 1976–1981 concentrations, which is based on annual water yields.
The flow-weighted average NO3 concentration in 1945 (5.0 mg NO3–N L-1) was between 54 and 73% of the expected mean value at an equivalent water yield during 1976–2001, depending on the model used to describe NO3 concentrations during 1976–2001 and depending on the assumed effect of composite sample collection in 1945. After applying a 19% upward adjustment to the 1945 flow-weighted average concentration to account for the effect of composite sample analysis, there appears to have been an 11% probability or less of observing a flow-weighted average annual NO3 concentration of 5.95 mg NO3–N L-1 during 1976–2001 when the water yield was equivalent to the 1942–1945 values.
In general, the difference between 1945 and contemporary concentrations was greater for arithmetic averages than for the flow-weighted concentrations. This may reflect actual differences in the two measures and/or our ability to accurately estimate these quantities and their expected values in the contemporary period. The arithmetic average concentration represents the time-averaged concentration of water passing a point in the river while the flow-weighted average attempts to represent the concentration of the average parcel of water in the river during a water year. Our ability to estimate these quantities accurately depends on sampling frequency. Sampling in 1945 was frequent and regular, but sampling in 1982–2001 was less frequent, irregular, and was biased in favor of April through September flows. This may contribute to increased values of arithmetic average annual concentrations, which would inflate the difference between 1945 and the contemporary concentrations. However, when the 1976–1981 observations (weekly sampling) were used in the intertemporal comparison, the 1945 arithmetic average concentration was outside the 95% confidence interval for expected individual observations. Thus, the difference in sampling frequency during 1982–2001 does not seem to account for the differences in arithmetic average vs. flow-weighted concentrations.
The intertemporal comparison also depends on our ability to estimate the expected average concentration in the contemporary period at the 1945 discharge. The higher coefficients of determination for the arithmetic average concentration–discharge relationships, compared with those for the flow-weighted average concentrations, contributed to smaller confidence intervals. Flow-weighted concentration values depend on assumptions about the concentrations occurring between samples. Somewhat different results might be obtained if different methods of estimating flow-weighted concentration were used and if a more precise equation for estimating the expected concentrations in the contemporary period could be developed, perhaps by incorporating nitrogen budgets into the analysis.
Since we do not have a multiyear record of riverine NO3 concentrations for the 1940s, we do not know whether the values observed in 1945 were above or below the average for the 1940s. If the statistical relationships identified in the 1982–2001 period are applicable to the 1940s, they suggest that nitrate concentrations in 1945 were higher than average, after adjusting for flow, because water yield in the three previous years was below average. It is not known, however, if this relationship is applicable to the different land use and hydrologic conditions that prevailed in the 1940s. Consequently, we cannot infer that the similarities or differences between 1945 and the contemporary period are indicative of similarities or differences between the 1940s and the contemporary period.
Comparison of 1978–2001 Concentrations at the Highway 46 and 64 Bridges and at Runnells
Keeney and DeLuca (1993) compared the 1945 riverine NO3 concentrations in the city of Des Moines to concentrations measured approximately 35 km downstream at Runnells during 1980–1990. Between 1978 and 2001, the average difference in individual NO3–N concentrations (n = 679) measured at the Hwy. 46/65 bridges and at Runnells on the same date was 0.5 mg NO3–N L-1, with lower concentration at Runnells. The mean difference in concentrations is statistically different than zero (P < 0.001). Since the distance between the two locations is approximately 25 km, the average river NO3–N concentrations decline at a rate of 0.02 mg NO3–N L-1 km-1 in the downstream direction in the contemporary period. If this same rate of change is extrapolated 10 km upstream to the 1945 sampling location, the concentration measured at the Hwy. 46/65 bridges would be, on average, 0.20 mg NO3–N L-1 less than concentrations at the 1945 sampling location, and the concentrations at Runnells would be 0.70 mg NO3–N L-1 less than concentrations at the 1945 sampling location. Adjusting the concentrations to account for this would tend to increase the difference between the values in 1945 and those measured between 1976 and 2001. However, the actual influence of the sampling location used in 1945 in comparison with the Hwy. 46/65 bridges is unknown. This uncertainty, together with differences in sample collection and analytical methods used, confounds the comparison of the contemporary concentrations with the 1945 values.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSCONCLUSIONSREFERENCES |
|---|
The 1945 concentrations were based on analysis of samples that were composites of approximately 10 individual daily samples, while the contemporary concentrations are based on analysis of individual (noncomposited) samples. Comparison of concentration–discharge relationships of composited and noncomposited samples collected in 1945 suggested that concentrations of composited samples were between 1% less and 19% greater than (P = 0.05) the noncomposited samples collected at the same discharge.
There did not appear to be any statistically significant temporal trends in NO3 concentrations or discharge at the Hwy. 46/65 bridges during the 1976–2001 period. There was, however, a significant increase in the discharge at sampling over this period, probably due to a change in sampling frequency after 1981. To avoid any influence of the change in sampling frequency, the 1945 annual concentrations were compared separately to the 1982–2001 annual values and to the 1976–1981 annual values. Similar results were obtained from both sets of comparisons.
Depending on the relationship used to describe the arithmetic annual average NO3 concentrations during 1976–2001 and depending on the estimated effects composite sample collection in 1945, the arithmetic average riverine NO3 concentration in the Des Moines River at Des Moines in 1945 was between 44 and 57% of the expected average value at similar flows during 1976–2001. The probability of observing the 1945 arithmetic average concentration at an equivalent discharge during 1976–2001 appeared to be less than 5%.
The flow-weighted average NO3 concentration in 1945 was between 54 and 73% of the expected average value at similar flows during 1976–2001, depending on the model used to estimate the contemporary NO3 concentration and depending on the estimated effects of composite sample collection in 1945. The probability of observing the 1945 flow-weighted average concentration at an equivalent discharge during 1976–2001 appeared to be 11% or less.
In summary, NO3 concentration in the Des Moines River in Des Moines during 1945 appeared to be between 45 and 75% of the expected mean value at similar discharge during 1976–2001 near the west side of Des Moines. This apparent change in concentration from 1945 to the present is considerably greater than suggested by Keeney and DeLuca (1993). Drawing generalizations about water quality in the 1940s based on these data is not recommended, however, because of the limited number of observations in the 1940s, and because water yield in 1945 was not representative of the 1940s. Differences in sampling location, sampling methods, and analytical methods also confound the intertemporal comparison. Long-term changes in riverine N concentrations are best evaluated using consistent sampling locations and methods over multiple years.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSCONCLUSIONSREFERENCES |
|---|
Related articles in JEQ:
This article has been cited by other articles:
|
|
 |
|
  W. H. Renwick, M. J. Vanni, Q. Zhang, and J. Patton Water Quality Trends and Changing Agricultural Practices in a Midwest U.S. Watershed, 1994-2006 J. Environ. Qual., August 8, 2008; 37(5): 1862 - 1874. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Crop Science | |||
| Journal of Natural Resources and Life Sciences Education |
Vadose Zone Journal | ||||
| Soil Science Society of America Journal | Journal of Plant Registrations | The Plant Genome | |||
