Note: This fact sheet is also available in acrobat (.pdf) format (484 KB): Download it
One station is upstream from a single barnyard-runoff source and the other station is downstream from that same source. The barnyards investigated were identified by each watershed plan as critical nonpoint sources based on herd size, lot size, proximity to the stream, and downslope overland flow characteristics.
Otter Creek is within the Sheboygan River Priority Watershed, 15 miles west of Lake Michigan, in east-central Wisconsin (fig. 1). The drainage area of Otter Creek is 9.2 square miles at the downstream sampling station, and land use in the watershed is 67 percent agricultural (Bachhuber and Foye, 1993). Upstream and downstream sampling stations, each of which are equipped to continuously monitor streamwater levels and collect discrete water samples, were installed at Otter Creek in March 1994. Water samples are collected with a refrigerated water-quality sampler that is activated by the rise and fall of streamwater levels.
Halfway Prairie Creek is within the Black Earth Creek Priority Watershed, 20 miles northwest of Madison, in south-central Wisconsin (fig. 1). The drainage area of Halfway Prairie Creek is 16.1 square miles at the downstream sampling station, and land use in the watershed is 60 percent agricultural (Eagan and Morton, 1989). Upstream and downstream sampling stations were installed at Halfway Prairie Creek in April 1995. The upstream sampling station continuously monitors streamwater levels and precipitation and collects discrete water samples with a refrigerated water-quality sampler. The downstream station is equipped to collect water samples only.
Upstream-downstream sampling schemes have the inherent potential for upstream loading sources to mask the effects of the investigated source, because individual inputs are often small compared to the cumulative inputs from upstream (Spooner and others, 1985). To reduce the potential for this problem, project investigators added two enhancements to the sampling design used at Otter Creek in order to improve the isolation of barnyard runoff at Halfway Prairie Creek. First, the water-quality samplers are activated by precipitation and are programmed to collect time-integrated samples for an initial period. This enhancement will be added to the Otter Creek sampling design for the post-BMP monitoring period. Samples are then collected in response to the rise and fall of streamwater levels in a manner similar to that at the Otter Creek stations.
Two benefits of this approach are that (1) it allows for sampling of barnyard runoff in the receiving stream before streamwater level increases can be sensed, thereby effectively isolating the barnyard runoff from nonpoint-pollution sources upstream, and (2) it allows sampling during small storms in which local inputs from the barnyard are apparent, but little storm runoff from the upstream areas of the watershed is observed. A second enhancement to the Otter Creek sampling design is that the upstream and downstream stations at Halfway Prairie Creek are located close enough together to allow a direct electronic connection and, hence, the collection of concurrent samples from both water-quality samplers. This design allows for statistical comparisons between concurrent individual upstream and downstream concentrations in water samples.
Samples of streamflow were collected during 10 storms between April 1994 and June 1995 at Otter Creek, and 10 storms during April-June 1995 at Halfway Prairie Creek. Samples were analyzed for total phosphorus, ammonia nitrogen, biochemical oxygen demand (BOD), suspended solids, and fecal coliform bacteria. All are known to be potential indicators of nonpoint pollution from barnyards.
In general, event-mean concentrations of total phosphorus, ammonia nitrogen, and BOD were greater at the downstream station than at the upstream station for most periods of storm runoff at Otter Creek
Using a paired Student's t-test, it was determined that for each constituent, the average of the differences between upstream and downstream event-mean concentrations was significantly different from zero at the 95-percent confidence level. In other words, average downstream event-mean concentrations of total phosphorus, ammonia nitrogen, and BOD were significantly greater than average upstream event-mean concentrations. The statistically significant increases in event-mean concentrations downstream from the barnyard suggest that the investigated barnyard-runoff source is an important contributor to the concentrations of total phosphorus, ammonia nitrogen, and BOD in Otter Creek for the storms monitored.
Upstream and downstream event-mean concentrations of total phosphorus, ammonia nitrogen, and BOD during storms 1 and 2 did not follow the same general trends observed during the other eight storms. The most likely reason for this observation is that the threshold for initiation of sampling was set too high for the first two storms. Based on this experience, it was realized that the threshold could be set lower. If sampling is initiated sooner, the downstream site is more likely to capture the higher pollutant concentrations expected in the early part of the barnyard runoff.
Event-mean concentrations of total phosphorus, ammonia nitrogen, and BOD downstream from the barnyard on Halfway Prairie Creek were also generally higher than those upstream
Average downstream event-mean concentrations of total phosphorus, ammonia nitrogen, and BOD were significantly greater than average upstream event-mean concentrations at the 95-percent confidence level. Differences between upstream and downstream event-mean concentrations at Halfway Prairie Creek were less variable than those at Otter Creek. Time-integrated samples collected early in the storms at Halfway Prairie Creek may have helped to isolate barnyard runoff from other upstream sources of pollution loads. Similar runoff volumes among the storms and minimal seasonal changes between storms also may explain this lower variability.
The difficulty in finding differences between upstream and downstream event-mean concentrations during large periods of storm runoff is apparent for storm 8 at Halfway Prairie Creek. Because concurrent samples were collected, however, individual concentration comparisons may highlight changes between the upstream and downstream differences over time. Such comparisons would likely focus on the beginning of the storms, when most of the runoff is local and from the barnyard source. Although discrete comparisons have not yet been conducted, the statistically significant increases in event-mean concentrations downstream from the barnyard suggests that the investigated barnyard-runoff source is an important contributor to the concentrations of total phosphorus, ammonia nitrogen, and BOD in Halfway Prairie Creek for the storms monitored.
Using the Student's t-test to find differences between two independent data sets, one can develop an equation to estimate the minimum amount of decrease in post-BMP downstream event-mean concentrations necessary to be considered statistically significant. This 'minimum detectable change' is usually expressed as a percentage. If the pre- and post-BMP sample sizes are assumed to be equal (10 storms) and the variance in the differences between upstream and downstream concentrations is kept constant, minimum detectable changes for significance at the 95-percent confidence level are the following:
To clarify, the average downstream post-BMP event-mean concentrations of total phosphorus at Otter Creek would have decrease by at least 50 percent for the change to be considered statistically significant at the 95-percent confidence level. These values are conservatively high because variance in the downstream-upstream differences for average pre- and post-BMP event-mean concentrations was assumed to remain constant. One would expect the actual variance to decrease after BMP implementation, thus decreasing the minimum detectable changes. As discussed earlier, the differences in event-mean concentrations between upstream and downstream samples for Halfway Prairie Creek were less variable than those for Otter Creek. This is reflected in lower minimum detectable changes. According to the watershed plans (Bachhuber and Foye, 1993; Eagan and Morton, 1989), the minimum detectable changes are smaller than the pollutant reductions expected from BMP implementation at each site.
USGS Water Resources of Wisconsin Home PageUSGS Headquarters Home Page
http://wi.water.usgs.gov/pubs/FS-221-95/index.html>.
(fig. 2)
(fig. 3)
