Lisa A. Senior

Estimated fecal coliform bacteria concentrations using near real-time continuous water-quality and streamflow data from five stream sites in Chester County, Pennsylvania, 2007–16

Several streams used for recreational activities, such as fishing, swimming, and boating, in Chester County, Pennsylvania, are known to have periodic elevated concentrations of fecal coliform bacteria, a type of bacteria used to indicate the potential presence of fecally related pathogens that may pose health risks to humans exposed through water contact. The availability of near real-time continuous stream discharge, turbidity, and other water-quality data for some streams in the county presents an opportunity to use surrogates to estimate near real-time concentrations of fecal coliform (FC) bacteria and thus provide some information about associated potential health risks during recreational use of streams. The U.S. Geological Survey (USGS), in cooperation with the Chester County Health Department (CCHD) and the Chester County Water Resources Authority (CCWRA), has collected discrete stream samples for analysis of FC concentrations during March–October annually at or near five gaging stations where near real-time continuous data on stream discharge, turbidity, and water temperature have been collected since 2007 (or since 2012 at 2 of the 5 stations). In 2014, the USGS, in cooperation with the CCWRA and CCHD, began to develop regression equations to estimate FC concentrations using available near real-time continuous data. Regression equations included possible explanatory variables of stream discharge, turbidity, water temperature, and seasonal factors calculated using Julian Day with base-10 logarithmic (log) transformations of selected variables. The regression equations were developed using the data from 2007 to 2015 (101–106 discrete bacteria samples per site) for three gaging stations on Brandywine Creek (West Branch Brandywine Creek at Modena, East Branch Brandywine Creek below Downingtown, and Brandywine Creek at Chadds Ford) and from 2012 to 2015 (37–38 discrete bacteria samples per site) for one station each on French Creek near Phoenixville and White Clay Creek near Strickersville. Fecal coliform bacteria data collected by USGS in 2016 (about nine samples per site) were used to validate the equations. The best-fit regression equations included log turbidity and seasonality factors computed using Julian Day as explanatory variables to estimate log FC concentrations at all five stream sites. The adjusted coefficient of determination for the equations ranged from 0.61 to 0.76, with the strength of the regression equations likely affected in part by the limited amount and variability of FC bacteria data. During summer months, the estimated and measured FC concentrations commonly were greater than the Pennsylvania Department of Environmental Protection established standards of 200 and 400 colonies per 100 milliliters for water contact from May through September at the 5 stream sites, with concentrations typically higher at 2 sites (White Clay Creek and West Branch Brandywine Creek at Modena) than at the other 3 sites. The estimated concentrations of FC bacteria during the summer months commonly were higher than measured concentrations and therefore could be considered cautious estimates of potential human-health risk. Additional water-quality data are needed to maintain and (or) improve the ability of regression equations to estimate FC concentrations by use of surrogate data. Introduction In Chester County in southeastern Pennsylvania, many streams provide opportunities for recreational activities, such as swimming, fishing, and boating. However, the periodic presence of elevated levels of fecal coliform (FC) bacteria in some streams indicates possible elevated human-health risks associated with exposure to potential pathogens in these waters. Fecal coliform bacteria are a group of bacteria found in the digestive systems of warm-blooded animals and are one of several types of fecal-indicator bacteria (FIB) that are indicative of the potential presence of feces and associated pathogens (disease-causing organisms). In order to protect human 2 Estimated Fecal Coliform Bacteria Concentrations Using Near Real-Time Continuous Data, Chester County, Pa., 2007–16 health during recreational use of streams and other waterways, standards for the presence of FC and other indicator bacteria have been established. In Pennsylvania, the FIB water-contact standards set by the Pennsylvania Department of Environmental Protection (PADEP) apply to FC concentrations in the water column, with standards specifying lower concentrations for the swimming season (May 1–September 30) than for the remainder of the year. One of the most common methods to determine FC concentrations involves membrane filtration of a grab sample and subsequent incubation for 24 hours (National Environmental Methods Index, 2016). However, this method introduces at least a 1-day delay in assessing bacteria concentrations. In addition, because of the type of sample (grab), the method is limited to determining bacteria concentrations at a single specific point in time at the sampling location. Interest in developing near real-time estimates of FIB concentrations has resulted in studies that evaluated bacteria surrogates, such as turbidity, that can be measured near real time in situ. Near real-time estimates of FIB concentrations can be used to improve protection of human health from exposure to potential pathogens during recreational use of surface waters. Regression equations that incorporate turbidity as a variable have been used to estimate bacteria concentrations for beaches along Lake Erie (Francy and others, 2013; Zimmerman, 2008) and in streams in Kansas (Rasmussen and Ziegler, 2003) and Georgia (Lawrence, 2012). In Chester County, recent data are available on FC bacteria concentrations in discrete grab samples and near real-time continuous turbidity at 5 sites with gaging stations on streams used for recreational activities, including 3 sites on Brandywine Creek, 1 site on French Creek, and 1 site on White Clay Creek (fig. 1). Discrete grab samples for FC bacteria analysis at these five stream sites have been collected by the U.S. Geological Survey (USGS), in cooperation with the Chester County Water Resource Authority (CCWRA) and Chester County Health Department (CCHD). Collection of the earliest discrete bacteria data began in the 1970s at the three Brandywine Creek sites and in 2012 for the other two stream sites (table 1). Near real-time continuous turbidity and other water-quality data currently (2016) are collected by USGS, in cooperation with the CCWRA, CCHD, and City of Wilmington, Delaware, using sensors installed at these 5 and 4 other gaging stations. Thus, as of 2016, sufficient data are available to allow development of regression equations to estimate near real-time bacteria concentrations at 5 sites but with additional FC data collection, regression equations could be developed to estimate near real-time FC concentrations at the other 4 gaging stations. The earliest continuous turbidity record for the Brandywine Creek sites began in 2005, and the record for the White Creek and French Creek sites began in 2011 and 2006, respectively (table 1). The availability of data and interest in improving healthrelated guidance for recreational use of the streams in Chester County made possible the development of regression equations for the estimation of near real-time FC concentrations so that the public might be better informed about possible actions to take to reduce chances of illness or infection. A preliminary unpublished evaluation of the use of surrogates, such as turbidity, to estimate FC concentrations for the three Brandywine Creek sites for the period 2007–10 was conducted by USGS for CCHD and CCWRA in 2012. Results of the preliminary evaluation indicate that turbidity and water temperature are better estimators of FC bacteria than turbidity alone. On the basis of these preliminary results and availability of data, it appeared that turbidity along with other water-quality properties might be used to estimate near real-time concentrations of FC bacteria and, thus, provide the public and health officials with an immediate indicator of potential health risk associated with recreational use of the selected streams in Chester County. In 2014, the USGS, in cooperation with the CCHD and CCWRA, initiated a more formal study to develop regression equations for the estimation of near real-time FC concentrations from available near real-time continuous data (collected at 15to 30-minute intervals and transmitted hourly). On further analysis, inclusion of variables other than turbidity and water temperature appeared likely to improve the strength of the regression-based approach to estimating bacteria concentrations. However, additional data beyond that collected through 2014, especially for the two stream sites with the discrete bacteria record beginning in 2012, are thought to be needed in order to develop statistically significant regression equations. Recreational Water Standards and Criteria for Fecal Indicator Bacteria In Pennsylvania, the water-contact standards for FC concentrations in the water column established by the PADEP are 200 coliforms per 100 milliliters (coliforms/100 mL) during the swimming season (May 1–September 30) and 2,000 coliforms/100 mL during the remainder of the year (Pennsylvania Department of Environmental Protection, 025 Pa. Code § 93.7, accessed May 31, 2012, at http://www.pacode.com/ secure/data/025/chapter93/s93.7.html). The PADEP standards of 200 or 2,000 coliforms/100 mL are defined as the geometric mean of FC concentrations in a minimum of five consecutive samples collected on different days within a 30-day period with no more than 10 percent of the FC concentrations exceeding 400 coliforms/100 mL in a 30-day period during the swimming season (May 1–September 30) (Pennsylvania Department of Environmental Protection, 025 Pa. Code § 93.7). The units of coliforms/100 mL used by PADEP are equivalent to units