Why it's so hard to study non point source pollution.
Neil Gillies, Cacapon Institute
Many of the potential sources of nutrients and sediment in West Virginia’s Potomac Highlands are non point in nature and, in most cases, it is precipitation in the form of rainfall or snowmelt that moves non point source pollutants into our streams. That relationship can be as simple as the direct washing of nutrients and sediment off the land into a stream via surface runoff, in which case increased concentrations are tied directly to precipitation. Some constituents, such as nitrate - the fully oxygenated form of nitrogen- once mobilized by precipitation can also move through the ground towards our streams; in this case there is a much more complex relationship between precipitation and concentration. Of course, not all precipitation causes runoff or significant infiltration. Those conditions depend on the intensity and/or duration of precipitation, type of ground cover, and the conditions preceding the precipitation (soil moisture, whether ground is frozen, etc.).
Watershed size plays an important role in water quality. It is essential to understand that the effects of stream-order/drainage area are as likely to have an impact on water quality concentrations as do land-use processes. In small watersheds, pollutants in the stream are relatively close to their source and their concentrations are relatively unaffected by instream processes that can change their concentration in the water column as they travel downstream. These instream processes include chemical and biological removal from the water column, chemical changes (such as microbial conversion of ammonia to nitrate), and simple dilution as new sources of water enter the stream. Smaller watersheds are therefore better indicators of the source of pollution. The larger the watershed, the more each parameter accumulates the effects of chemical and biological change, changes due to the influx of water from an increasing number of locations, and an increasingly large number of potential sources.
Cacapon Institute (CI) has been studying water quality in the Potomac Highland’s region since 1985. This report will use a tiny bit of CI’s data to illustrate what water quality data in non point source impacted watersheds looks like. Specifically, it will look at two nutrients: total phosphorus and nitrate-nitrogen.
Looking for Ghosts- Phosphorus
Total phosphorus is the sum of all forms of phosphorus: organic and inorganic, suspended and dissolved. While phosphorus does not move easily though ground water, erosion can transport large amounts of sediment-bound phosphorus to surface waters (Mueller et al, 1995). The literature indicates that over 75 percent of annual watershed runoff can occur during a small number of severe precipitation events (Edwards and Owens, 1991). Because P primarily moves in runoff, Sharpley (1995) estimates that over 90% of the annual P load can be delivered by these few events.
Phosphorus concentrations below point sources such as sewage treatment plants are often very high. However, phosphorus from non point sources, even where sources are extremely abundant, can be elusive. Phosphorus that enters the stream attached to sediment quickly drops from the water column to the stream bottom. Phosphorus that enters the stream in dissolved forms is rapidly taken up by microbes and plants, or chemically adsorbed to sediment – and mostly disappears from the water column as well.
The bar graph below shows three years of total phosphorus concentration data at two different stream sites in Hardy County. The light blue line shows the amount of rainfall that fell in the 30 days prior to sampling. The “blue” site is an intensely agricultural watershed that has large amounts of phosphorus applied to its farmlands every year. The “yellow” site is from a heavily forested watershed, but the site happened to have a phosphorus bearing spring that delivers water at the sampling site from within the stream channel and along the edges. Even though the phosphorus levels were usually low, these two sites typically have among the highest phosphorus levels seen in CI’s studies. Over the three years, only three high phosphorus “spikes” were captured. These spikes were somewhat related to the cumulative precipitation line, but a period of relatively heavy precipitation did not necessarily result in really elevated phosphorus concentrations. Each of the phosphorus spikes were associated with heavy rain that fell in the 24 hours prior to sampling.
The phosphorus pattern is consistent with what is expected of a mostly non point source derived parameter that is mostly present in the water column only fleetingly in association with rainfall and snowmelt events. Studying phosphorus is a bit like looking for ghosts. You would see a similar pattern for sediment in the water – actually even more extreme. Knowing what is happening with phosphorus is extremely important for the health of the Chesapeake Bay and our headwater streams, and it's extremely frustrating that it is so difficult to collect meaningful data.
Indicator Species - Nitrate-Nitrogen
Nitrogen makes up 78 percent of the earth's atmosphere, and cycles naturally through our ecosystems in many ways, including microbial action, and decomposition. Various forms of nitrogen are essential for plant and animal life, and nitrate-nitrogen is one of these. Much of the earth's natural supply of nitrate is produced by bacteria and algae that convert nitrogen gas to nitrate through various chemical reactions. In addition, nitrogen applied to crops as ammonium and organic N can be converted to nitrate by various chemical and biological pathways (Keeney, 1989). Nitrate readily dissolves in water, is chemically stable over a broad range of environmental conditions and moves easily through ground and surface waters (Mueller et al, 1995). Once in the stream, nitrate concentrations are much more persistent in the water column than phosphorus. However, microbial processes that occur at the stream bed interface process nitrate in a number of ways including removal from the stream via conversion to elemental (gaseous) nitrogen. These processes have the greatest effect on in-stream nitrate during low flow conditions.
Correll et al (1994) compared the concentrations of aqueous species of nitrogen (N) in different environmental settings in the Chesapeake Bay watershed. They found nitrate was the major dissolved form of nitrogen detected, with concentrations 10 to 20 fold higher than dissolved ammonium and organic nitrogen. For this reason, nitrate-nitrogen (NO3-N) is often the best quantitative indicator of nitrogen in WV’s rivers.
The bar graph above shows three years of nitrate concentration data at the same two stream sites in Hardy County discussed for phosphorus and the light blue line shows the amount of rainfall that fell in the 30 days prior to sampling. Again, the “blue” site is from an intensely agricultural watershed and the “yellow” site is from a heavily forested watershed. Unlike phosphorus, the nitrate concentration differences between the two sites were very large. Concentrations at the agricultural site varied enormously, and these variations were clearly related to precipitation, but in a rather complex way. The type of precipitation that leads to the highest nitrate concentrations for the longest periods of time is saturating rainfall that leaves the ground sodden and mobilizes nitrate to move through the ground toward our streams. The periods during 1998 and 1999 when the nitrate concentrations at the agricultural site were very low corresponded to a period of very low precipitation. In fact, the region was in a severe drought from October 1998 through much of 1999.
CI's scientists consider nitrate to be our most useful indicator of land use influences on water quality but, like all non point source pollutants, it's most informative when it's been raining.
So far, we have looked only at concentrations of nitrate and phosphorus. In many ways that is the most relevant way to look at water quality when you are concerned about local waters. The plants and animals in the stream are affected by the amount of a substance in the water, not by how much is passing by on its way downstream. However, when you are interested in assessing downstream impacts of any pollutant, you want to know how much of a substance is being "delivered" – you want to know the load. The load is the total amount of a substance that passes by some point in a certain amount of time - as in pounds per hour or tons per day. It is equal to the concentration of the substance times the total volume of water (often measured in cubic feet per second -cfs), a calculation that requires that an accurate flow measurement.
The following series of graphs show concentrations and loads related to stream flow at a single site that drains 109 square miles in Hardy County's Lost River watershed. The first graph shows total phosphorus concentrations between July 1998 and September 2000 and flows on the days that water samples were collected. The pattern should be familiar, with most phosphorus concentrations being fairly low and not varying much. One high concentration spike is noted in February 2000 that seems to be related to the high stream flows on that day. However, the rest of the data did not show a notable relationship between flow and phosphorus concentration. This is a pretty good indication that high flows alone might not indicate that you should expect to find high concentrations of phosphorus.
The load graph (above) tells a very different story. Elevated phosphorus loads were solely related to elevated stream flows. Days with low flow had loads that weren't large enough to register on this graph. You can see a couple of "blips" when hurricanes Dennis and Floyd hit the region with glancing blows. But the one particularly high load spike occurred during a region-wide storm in February 2000, and the next two data points on the graph were actually the following two days (February 19-21, 2000). What if we had not been able to sample on those days?
The graph below shows nitrate-nitrogen concentrations between July 1998 and September 2000 and flows on the days that water samples were collected. The pattern should be familiar, with nitrate concentrations varying greatly over time. One particularly high concentration spike was noted in early 1999, on a low flow day. The data did not show a notable relationship between flow and nitrate concentration. This is a pretty good indication that high flows alone might not indicate that you should expect to find high concentrations of nitrate.
The load graph (above) tells a very different story. As with phosphorus, elevated nitrate loads were always related to elevated stream flows. Unlike phosphorus, the elevated levels remained higher longer because of the ability of nitrate to continue flowing into streams both overland and through the ground. Periods with low flow had loads that weren't large enough to register on this graph. You can see a couple of "blips" when hurricanes Dennis and Floyd hit the region with glancing blows. But the one particularly high load spike occurred during a region-wide storm in February 2000, and the next two data points on the graph were actually the following two days (February 19-21, 2000).
So, if you can just get out and collect samples during storms at sites that have flow information, you can get a good estimate of the loads that affect downstream waters. Right? Well, not really. The following graphs take a much closer look at what happens during a storm.
Anatomy of a Storm
Storm sampling is very difficult. It requires a willingness to sample whenever a storm is happening, and the luck to be both available and in the right location to capture it. The following graphs follow concentration and load changes of total phosphorus and nitrate-nitrogen before, during and after a storm in the Lost River watershed in June 1998. The day started in the morning with a regularly scheduled sampling run. While processing the samples in the laboratory later that day, a big storm hit the watershed. Lab staff was back out in the field sampling by 7:00 pm, collected sequential samples until shortly after 10:00 pm, went back to the lab to process bacteria samples, collected another sample on the way home at nearly 3:00 am the following morning, then collected yet another sample on the way to work at 9:30 am the same day. Yes, it helps to be a bit nuts when you do this kind of work.
Total phosphorus (TP) concentrations rose rapidly to a peak early in the storm, then fell rapidly - reduced by 50 percent six hours after the peak was reached, 65% after 12 hours. Thirty eight hours after the storm, TP concentrations had dropped to pre-storm levels. Loads followed a similar, if even more precipitous pattern.
Nitrate-nitrogen (NO3-N) concentrations and loads reached a plateau at the same time as TP and retained that level throughout the storm interval. Thirty eight hours after the storm, NO3-N levels remained somewhat elevated.
River stage remained steady during the storm sampling interval from 6/15/98 at 7:15 p.m. until 6/16/98 at 9:30 a.m., and the river was very muddy throughout this time period. Nitrate loads and concentrations generally corresponded with river flow. However, turbidity (an indirect measure of sediment) peaked with TP early in the storm and dropped more than 80% by 0930 the following morning even as the river level remained high. Nitrate remained high because nitrate continues to flow into the steam via subsurface and subsurface pathways even after the storm has passed. Phosphorus and sediment concentrations and loads rose and fell quickly because they only move into the stream during periods of rapid surface runoff.
So, the river was high and muddy for many hours, but collecting a sample for phosphorus or sediment at any specific time during the storm would not have yielded the same result as another time.
And that's why it's so hard to study non point source pollution.