Research Topics » Long-Term Stream Monitoring

By Will Jobs, Collins Research Fellow, Vassar College Environmental Research Institute
Published: April 25th, 2011

Long-term stream monitoring is an important way to track the health of a watershed. While establishing a baseline of data to follow changes in stream quality, it also provides a way to see how land use changes, restorative actions, and application of chemicals throughout the watershed impact our waterways.

Introduction

As we come to learn more about the connection we all share with our watershed, the importance of the health of our streams becomes increasingly obvious. Because everything that happens to the land within a watershed eventually ends up in the streams, monitoring the health of streams is a way to assess the health of the entire watershed. This data provides information about the quality of the groundwater and drinking water in an area as well as the ability of the watershed to support wildlife.

The site on the Casperkill where the sonde and in-water HOBO pressure sensor are stationed.

Unfortunately, analyzing a single water sample is not a good way to understand or characterize the health of a stream or its watershed. Instead, stream monitoring is usually done over long periods of time (at least one year) in order to capture seasonal changes as well as changes in human intervention over time (such as nutrient inputs from fertilizer and altered stream banks). Even better than collecting multiple samples over a long period of time is real-time stream monitoring, in which data is collected at regular frequent intervals. Real-time monitoring is not subject to bias from researchers sampling only during good weather or only during the day so the data more accurately reflects stream quality. In addition, real-time monitoring data captures all states of a stream – whether in flood or drought, during natural disturbances and human intervention.

One method of long-term real-time stream monitoring is with a sonde, a device with multiple probes that collects data at programmed intervals and is deployed directly into the stream. This allows researchers to assess multiple things including: how land use changes have affected stream quantity and quality; the success of restorative actions taken throughout a watershed; and how long chemicals (like road salt) persist in groundwater. As part of a series of ongoing studies of the Casperkill Creek (see research articles “flashy streams” and “sewage in the stream”, as well as the “Health of the Casperkill” document), the Vassar Environmental Research Institute installed a sonde into the Casperkill to monitor numerous parameters over time.

The Casperkill Creek was chosen for this study for a number of reasons, including the fact that it flows through Vassar, it flows through a highly urbanized area, and it encounters a number of different land use types along its banks. The stream is rated class C (as of August 2008) by the NYSDEC which indicates it is not suitable for swimming or other contact activities but is able to support fish populations. Previous work has shown that the stream health varies quite a bit in parallel to the land use changes. The creek also suffers from much human influence, including some leaky sewer systems and road salt inflows in the winter.

Back to top

Research

The goal of this study is to establish a baseline of data and a system of long-term monitoring for an urban stream to identify land-use effects, separating this from water quality changes throughout the year. Currently collecting data every 20 minutes, the sonde in the Casperkill measures many parameters that include those in the following table. Note that some of these parameters are no longer collected, but past data remains in the database. Parameter names below reflect their names on the data access page.

The sonde database is accessed from the Data Exchange page in the Research
section and allows viewing, downloading, and graphing of all the collected data.

Recently, this data from almost four years’ monitoring has been reviewed, reorganized, and stored in a database which is now publicly accessible on the Data Exchange page. The user-friendly interface enables the user to view or download the entirety of the dataset or any subset of it and even allows the user to graph the data directly on the page. This new system allows the researchers working on this study to collaborate more easily by maintaining a single location with the latest, high quality data as well as providing a way to more easily analyze the data for trends. It also provides a means for the public to access data being generated on the Casperkill and will hopefully serve as a model for future datasets on other streams.

The data set acquired so far is extensive (over 120,000 rows of data) and much analysis of the data is still to be done. Trends involving chloride have been the subject of much attention recently and with good reason: each winter road salt is applied gratuitously to roads regardless of how traveled the roads are, and this road salt is eventually washed into groundwater and streams. As chloride starts to have effects on biota and microbial processes when concentrations rise above 100 mg/L, it is important that streams be monitored to assure levels do not rise significantly past that mark. As the graph below shows, lower temperatures correspond to higher chloride levels in the Casperkill because they are associated with winter snowstorms and road salt applications. Furthermore, while chloride levels usually drop when the discharge of the stream rises (as the amount of water increases which dilutes the chloride), in the winter months this trend is reversed because these discharge events are associated with snow and so the influx of road salt outweighs the dilution from the extra water in the stream, leading to a net rise in chloride.

A plot of chloride and discharge data from the sonde database shows the normal pattern of falling chloride with discharge events in autumn but a rise in chloride with discharge events in winter, associated with road salt applications.

The research group encourages the public to look through the data and observe many of the trends known from environmental science. For example, make a hydrograph by plotting discharge against precipitation to see how a precipitation event leads to a slightly later discharge event with a peak discharge and return to base flow. Or plot pH against precipitation to see how the slight natural acidity of rain (due to dissolved carbon dioxide) causes the pH to fall (become acidic). Any interesting trends or noticed problems with the data can be reported to the Collins Research Fellow at the Vassar Environmental Research Institute by emailing This e-mail address is being protected from spambots. You need JavaScript enabled to view it .

Back to top

Management

The Casperkill Watershed Alliance held a public information session on February 5, 2010 about the impacts of and alternatives to road salt with speakers from the Cary Institute.

Long-term datasets like these are essential to providing officials with the information they need in order to make sound decisions regarding the environment. Not only does long-term monitoring provide a baseline of data to which all future data can be compared to observe trends over time, it also gives researchers a way to see how specific actions (e.g. urban development, the creation of a landfill, chemical spills, etc.) have affected the stream in the short- and long-term.

One such example of how this data can be used to impact future municipal decisions is by monitoring how changes in the application of road salt have affected the levels of chloride in the stream. As described above, chloride levels above a certain threshold can be detrimental to wildlife and, eventually, humans. The data shows high chloride levels in the Casperkill year-round, suggesting levels of road salt currently used have infiltrated into the groundwater which recharges the stream year-round. This issue motivated the Casperkill Watershed Alliance to have a Public Road Salt Information Session this past February to discuss the ecological impacts of road salt as well as road salt alternatives and ways to reduce its use. Presentations from this workshop are available in the Resources section.

Back to top

Methods

Data has been collected since July 12th of 2007 on the Casperkill Creek at a suburban site between Route 376 and Spackenkill Road using a YSI 6920 sonde with YSI 6136 turbidity, YSI 6562 DO, YSI 6560 conductivity, and YSI 6561 pH attachments. Furthermore, until September 30th of 2008 ammonium levels were measured using a YSI 6883 Ammonium Ion-Selective Electrode, while nitrate was measured until July 14th, 2010 and chloride until July 27th, 2010 with YSI 6884 and YSI 6882 Ion-Selective Electrodes, respectively. Past measurements with the ion-selective electrodes are included in the online dataset but are no longer taken due to instability of their measurements and their range of error being larger than the values measured in the Casperkill. Measurements were taken every five minutes until December 4th of 2007, when measurement frequency was reduced to once every twenty minutes to conserve battery power. Up until December 23rd, 2007 all measurements were made with a single sonde (“Wendy”). On that date a second sonde (“Casper”) was deployed in place of the first, and since then every 10-14 days the sonde in the field has been switched with the other sonde, which had been sitting in lab and whose probes were all calibrated the day of switching. The sonde that had been in the field is then brought back into the lab to gather its data, check probe calibrations, and charge its batteries until its next deployment.

One day in April the sonde probes were covered with frog eggs!

Chloride levels are measured by comparing the specific conductance to a regression curve generated using ion chromatograph (IC) measurements of chloride levels of water at the time the sondes are swapped. The relationship between chloride and specific conductance is direct and linear. A different regression curve is made for each sonde as they give somewhat different measurements of specific conductance in the stream. This enables us to get chloride measurements independent of the sonde.

Discharge of the stream is measured indirectly using two HOBO pressure sensors, one of which is submerged in the creek to get the absolute water pressure, the other which is maintained in an office at Vassar College, approximately 1.2 miles away, to give atmospheric pressure. The difference in pressures gives the water pressure, which can be related to a polynomial rating curve for the Casperkill (see “Health of the Casperkill”). Precipitation data is measured every 20 minutes with a rain gauge at the Vassar Farm and Ecological Preserve, less than half a mile from the sonde. In cases where the weather station at the farm did not collect data (due to lack of sunlight charging the batteries via the solar panels), gaps have been filled with hourly precipitation data from the NCDC climatic weather service for the Dutchess County airport (approximately three miles away).

The quality of the data is ensured by carefully reviewing the data. Data files from the sonde, as well as the HOBO pressure sensors and the weather station data are combined into quarterly files, which are periodically reviewed in conjunction with calibration logs for the sonde to identify any clearly erroneous data due to bad calibrations or dying batteries. The sondes and probes are carefully and thoroughly cleaned every time they are removed from the Casperkill and are maintained as directed by their manual. Data is uploaded to the database (on the Data Exchange page) after each quarter, and whenever the dataset has been modified (for example, to improve our IC-chloride regression or after reviewing the data).

Back to top

References

• Vassar College Environmental Research Institute. "Health of the Casperkill." February 2009.
• Findlay, S. "Salt in Dutchess County Waters -- Where, when, so what?" February 2011.

Back to Research Topics