Researchers design water quality sampling scheme
provided by University of Arkansas
wo University of Arkansas researchers have discovered that when it comes to accurate sampling of water quality, timing is everything, especially during a storm.
Nutrient and pollution loads in a stream or river are determined by sampling the water and applying some type of interpretation. Sampling schemes monitor flow rates continuously, take water samples intermittently and then calculate loads by either direct integration or a load calculation model. However, the unknown events between samples can lead to errors.
The problem arises because samples are commonly taken on a monthly basis, explained Marc Nelson, director of the Arkansas Water Resources Center Water Quality Lab. But we found that the pollution load increases dramatically during storms, which are rarely accounted for by monthly sampling.
Nelson and Soerens, assistant professor of civil engineering, based their findings on data acquired through five years of intensive sampling of the Illinois River and Moore's Creek, located in northwest Arkansas. Their database includes 1,200 discrete samples of six different water quality parameters and covers base flow conditions as well as 19 storm events. They also accumulated data from two years of sampling on the Kings River in the White River basin. These data were used to compare results from automatic sampling and manual cross-sectional sampling. They found that the two sampling techniques are correlated, but they can produce different results.
Monthly sampling assumes that the sample concentration represents the concentration in the stream between samples. This leads to errors when concentrations are not steady, which is the case during storms.
Much or most of the load in a stream is transported during storms, explained Soerens. This load is missed unless storms are intensively sampled. For example, during one storm in January, 1998, the measured total phosphorus on the Illinois River surpassed the phosphorus estimated for the entire previous year, which was based on monthly collection data.
By sampling storm runoff events at 30-minute intervals on both waterways, researchers were able to develop sub-sampling and data analysis techniques. They used these results to determine the minimum sample interval required to accurately determine pollutant loads during storms.
We found that the optimum sample interval depends on the type of pollutant being measured and the size of the drainage basin, said Soerens. As a whole, we found that as the time between samples increased, the error in the load estimate increased.
On the Illinois River, which is not prone to flash flooding, the optimum sampling interval for total suspended solids (TSS) during a storm event is 2.5 hours. However, at Moore's Creek, a stream subject to flash flooding, the optimum sampling interval for TSS is 45 minutes.
The results indicate that pollutant load determinations are very sensitive to the frequency of sampling during storm events, said Nelson. All of the parameters except nitrate have most of their concentrations peak during the first part of a storm. This load is missed unless the storms are intensively sampled.
In the future, the researchers will use data from these sites to investigate the role of non-point sources in calculating the total daily maximum load (TMDL) for a body of water. Non-point source pollution is the accumulation of sediments, nutrients, or other contaminants that originate from the surface of the land, often as a result of agriculture or forestry operations. These sources usually enter streams as a result of storms. Storm water moving rapidly over land or concrete and asphalt can pick up pollutants that diminish water quality.
Soerens and Nelson used the data from their samples to evaluate the methods of load calculation. They found that direct integration gives better short-term results, but models can give good results in the long term if storm samples are included.