1 Introduction

Urban detention basins provide an impoundment to store runoff from the surrounding catchment and then meter it out slowly into a receiving body of water [1, 2]. Water quality from stormwater systems can often fail to meet clean water standards [3]. This failure to meet standards has led to stormwater research focused on water quality [4,5,6,7]. The constituents of water that flows into detention basins can vary from the first flush, which contains more sediment, chemicals, solid waste, and metals [8], to later flows into the basin [9,10,11].

The varying water quality of stormwater that flows into detention basins requires methods to assess those waters at different pool surface elevations. A vertical water quality assessment method is needed because a portion of first flush flows will remain in the basin and mix with subsequent flows that have less constituents, giving an inaccurate representation of what happens throughout a storm event. Additionally, most research to date has failed to address the changing water quality throughout a storm event, focusing only on the first flush [7, 9], or has used expensive autosamplers to attempt to take samples at different times [12]. The project sought to fill the need for a simple low-cost passive vertical water quality water sampling system for detention basins, and included development and implementation of the system within three detention basins to compare water quality.

2 Methods and materials

2.1 Passive vertical water quality assessment method

A passive vertical water quality assessment system for detention basins needs to sample stormwater as it enters the basin, fills the basin to different elevations and eventually leaves the basin at the outflow point. Ideally, a passive vertical water quality assessment method should be designed so no personnel would need to be present while the basin fills from large storm events. Storm events often happen when personnel are not available and certain events have hazardous conditions such as lightning and flooding. An automated water sampling system would work for sampling, but for many the expense and the technical knowledge needed is a limiting factor [13].

Challenges for automated water samples are finding electricity or setting up solar power, the sampler can typically be only set to one elevation and needs to be placed or checked prior to the storm event. There are passive stormwater samplers for first flush that are commercially available, for approximately $200 USD and are normally limited to a 1 L capacity. These samplers are designed to let in water at the level it is placed and once filled no other water can enter the sampler. As these samplers have a 1 L capacity there would be situations where multiple samplers at each elevation would be needed to provide the volume of water needed for complete analyte analysis. Given the cost and challenges of these samplers, authors sought to design a simpler sampling method that would use readily available materials to reduce costs and may better fill the need for assessing vertical water quality.

The simple passive vertical water quality sampling method designed during this study used materials that are widely available. The design used standard five gallon plastic buckets that are readily available and are 8.927 L in volume with a 30.25 cm diameter opening at the top, 26.23 cm diameter on the bottom, and 36.83 cm height. Each bucket had eight 1.27 cm holes drilled in a level line just under the lip of the bucket and a water tight lid placed on top. The plastic buckets were made from High-density Polyethylene (HDPE) and conforms to the stormwater sample collection protocols of the Environmental Protection Agency [14]. The holes allowed basin stormwater to flow into the bucket at a specified vertical elevation. Once filled, the internal pressure in the bucket would no longer allow water to enter the bucket, preventing any water from a higher vertical elevation from mixing with water in the bucket. This design ensures that water samples come from a particular vertical elevation within the basin. Additionally, the large size of the buckets allows for standardized sampling methods, such as a churn splitter to mix the samples which is used to reduce bias among samples.

Before the buckets were placed out they were acid washed (10% Hydrochloric acid solution) to reduce contamination of the water sample. The buckets can be strapped to any sturdy pole, or in this case T-posts, at the elevation chosen. Straps used in this case were ropes and elastic cords so the bucket would not move prior, during, or after it filled with stormwater. Additionally, they were easy to remove when retrieving the bucket. Bucket placement within a basin can be determined based on study design, but placing buckets outside turbulent flows within the deepest parts of the basin will ensure the buckets function as intended. The buckets can be retrieved once conditions allow for safe retrieval. In this study, that was between one to 24-h after storms when water levels were low enough so a person using waders could safely retrieve the buckets. For difficult to reach or deep basins, floats could be placed on the ropes and a quick release mechanism designed to retrieve buckets from a sturdy boat.

The simple sampling design allows for any elevation to be chosen for sampling that fits with the design of the study (Fig. 1). For this study, buckets were arranged to sample stormwater at following elevations as measured from the basin bottom: 0.3 m, 0.6 m, 0.9 m, 1.2 m, and 1.5 m. An equal spacing of 0.3 m was chosen to make sure the peak pool elevation of an event was collected knowing peak elevations of events would vary over the period of the study. The bucket for 0.3 m elevation was buried partially into the ground to ensure holes were set at exactly 0.3 m. As the water receded from the basin, grab samples can be taken at each of the same elevations as the buckets. Because water in detention basins recedes at a slower rate, the ability to schedule personnel for grab samples is much easier.

Fig. 1
figure 1

Arrangement of the buckets for sampling detention basin pool surface levels along with details of the buckets

Full size image

2.2 Study area

This study was conducted at three stormwater detention basins in Fargo, North Dakota, USA, being: (1) Fisheye Basin (Fisheye) located at 46° 52′ 02.16″ N 96° 51′ 24.40″ W; (2) Scheels Basin (Scheels) located at 46° 51′ 29.69″ N 96° 51′ 56.90″ W, and (3) The Fargo Project (TFP) located at 46° 51′ 10.10″ N 96° 51′ 10.10″ W. The basins were chosen because of their similarities in regard to size, geographic proximity, topography, and surrounding land use. One key difference between the detention basins was the greening of the TFP site, including in the removal of the concrete channel, while the other basins did not have the same greening. Within these three basins, stormwater runoff was detained and discharged out over a period of 12–36 h depending on the severity and duration of the storm and the stormwater volume.

Sampling was conducted during storm events with 2.54 cm or more of precipitation within a one-hour period. This level of precipitation creates enough stormwater runoff to fill the three basins to at least the 0.6 m pool elevation. There were two events during 2016 and one event in 2017 that met this criterion. Because the three basins had different characteristics, events did not fill the basin to similar levels. Thus, comparison of elevations among the three basins was not sensible. To facilitate comparisons the following three stages were identified: (1) “Initial” filling of the basin with first flush; (2) “Peak” corresponding to the highest elevation that stormwater filled the basin; and (3) “Outfall” being the last grab sample before the basin completely drained.

2.3 Sample analysis

Once removed from the basin, sample buckets were immediately mixed using a SP Bell-Art churn splitter (Bel-Art-SP Scienceware Wayne, NJ, USA) 4 L to resuspend particulates and take a standard sample of the bucket water. All samples were processed and preserved according to the North Dakota Department of Environmental Quality (NDDEQ) and sent to the NDDEQ Lab in Bismarck, North Dakota, USA to analyze for Total Suspended Solids (TSS), total dissolved solids, nutrients complete (Total Kjeldahl Nitrogen (TKN), NO2, NO3, NH3-N, and P), Escherichia coli (E. coli), 15 general chemistry parameters, eight nutrient analytes, 15 trace elements, oil and grease, and diesel range organics. E coli samples were transported the day the sample was taken and were kept cold while in transport to the NDDEQ’s lab. The NDDEQ E. coli analysis used the Colilert-18/Quanti-Tray/2000 to quantify E. coli (Idexx Laboratories, Inc., Westbrook, Maine). The quantification method is based on the “most probable number” methodology. The results are reported in colony-forming units (CFU)/100 mL. Samples were diluted if the upper detection limit was reached and rerun to determine CFU/100 mL, after that if CFU reached the upper limit of 16,000 CFU then 16,000 CFU/100 mL was reported, but the actual CFU could have been much higher.

Selected analytes were tested for these three factors: (1) events (July 2016, September 2016, and June 2017); (2) stages (Initial, Peak, and Outfall); and (3) site (Fisheye, Scheels, TFP). The analysis design was a block design, with the site as the block and events and stage as fixed factors [7]. The analysis used PROC GLM in the SAS® software, Version 9.4 of the SAS System for Windows (Copyright © 2015 SAS Institute Inc. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc., Cary, NC, USA) and the Tukey adjustment was used for multiple comparisons among the three factors.

3 Results and discussion

3.1 Passive vertical water quality assessment method

The passive vertical water quality sampling system performed as designed. Having a series of buckets at different heights meant that the peak elevation of stormwater was collected even if the storm events varied over time and among basins. In addition, if stormwater samples were needed at other elevations, the arrangement of buckets would have generated the needed samples. Because of the simplicity of the method and use of easily acquired materials, there was no need to have expensive automated systems or costly commercially purchased equipment siting in a basin waiting for an event to occur. The authors suspect that more expensive equipment may encourage curiosity and mischief with the equipment; whereas buckets on poles tied together with rope appeared to be uninteresting to local residents, both human and animal. This method would benefit agencies that have limited man power and limited funds.

One of the drawbacks of the method is the samples sit until retrieval of buckets; therefore, samples are not immediately cooled or chemically stabilized. The amount of time that samples sat depended on how fast water drained from the basin as to when personnel could retrieve the buckets. Only automated samplers solve this problem, while the commercially available stormwater sampler systems have the same issue of the sample sitting until retrieved. This method would not be appropriate for volatile organic analysis because volatile substances could dissipate before the bucket could be retrieved and a water sample taken. Advantages of the method include being low-cost, low technical knowledge needed, and a method that can be done by almost any group. Due to its simplicity reduced training and maintenance is needed and there is little that can go wrong when compared to automated systems. The passive vertical water sampling system reduced vertical water sampling to a simple process that allows for precise vertical water sampling in an uncomplicated and safe manner.

3.2 Water quality comparisons

Water quality data were compared among the three storm events and the three stages (initial, peak, and outfall) and among the three sites. Results from a select group of analytes are reported to demonstrate the capabilities of the method. The other analytes had few differences among the events, stages, or basins, and were below levels that would trigger water quality concerns.

The TSS analysis found no statistically significant differences among sites and events (p ≥ 0.05). However, there were significant differences among stages (p ≤ 0.05). The TSS readings were highest at the initial stage and then decreased with dilution and residency time (Fig. 2). There appears to be settling of particles over time as the water remained in the detention basin. This is in line with other research regarding suspended solids, which also shows suspended solids settling with residency time [15].

Fig. 2
figure 2

Total suspended solids levels at the initial, peak, and outfall stage showing storm events averaged over the three sites. Different letters following the x axis labels represent significant differences at the p ≤ 0.05

Full size image

Because the simple passive vertical water sampling method found the same response of TSS in detention basins as reported in the literature, this supports that the method is capable of being an effective assessment method for detention basins. Assessing TSS is important for stormwater because TSS has been shown to transport harmful pollutants and can negatively impact receiving water bodies [16]. Using this simple sampling method allows researchers and managers to assess their TSS related water quality issues and determine if actions, such as retrofits [17], are able to improve TSS.

Total phosphorus (TP) is a critically important factor in freshwater and is reported here to give more context as to how the simple passive vertical water quality sampling method will function as an assessment method. There were significant differences for TP among the storm events and stages (p ≤ 0.05), but not sites (p ≥ 0.05) (Fig. 3). Measurements of TP were greatest during the initial flush of water. Once basins reached their peak, these numbers decreased and remained relatively constant through the outfall. The TP levels for the July 2016 event were significantly higher than the other two events. These results establish that the method is able to discern differences in levels of TP that would be informative in the management of stormwater.

Fig. 3
figure 3

Total phosphorus levels at initial, peak, and outfall stages showing storm events averaged over the three sites. Different letters following the x axis labels represent significant differences at the p ≤ 0.05

Full size image

The analysis of E. coli is included as another example of the capability of the sampling method. The recreational standard for E. coli in the state of North Dakota (NDCC §33-16-02) states that surface water for recreational (boating, fishing, and swimming) purposes should not exceed 126 coliform forming units (CFU)/100 mL. Even though detention basins are not often thought of as recreational spaces, this is an area that has been retrofitted for human recreational use. Throughout the study, all E. coli samples taken failed to fall within the recreational standard of 126 CFU/100 mL (Fig. 4). The September 2016 event had E. coli levels that were significantly higher than the other two events (p ≤ 0.05). Neither the sites nor stages were significantly different (p ≥ 0.05). During the September 2016 event, all but one sample was at 16,000 CFU/100 mL, the highest value that could be measured. These results show that this simple method will be able to differentiate among events for this critically important water quality issue. The results from the E. coli analysis was an important factor in determining the design for a subsequent study by Olson et al. [7] of E. coli in detention and retention basins in Fargo.

Fig. 4
figure 4

E. coli levels at initial, peak, and outfall stages showing storm events averaged over the three sites. Different letters following the x axis labels represent significant differences at the p ≤ 0.05

Full size image

4 Conclusions

The simple passive vertical water quality sampling method used readily available and inexpensive materials, did not require any exposure of personnel to hazardous conditions, and replicated water quality results found in other studies. The method had a similar response to TSS found in other detention basin studies, showing its reliability as a sampling method. Additionally, the capability of the method was demonstrated in its ability to detect differences among elevation stages and storm events for both TP and E. coli. This method has ability to be used by researchers, managers, and citizen science as a way to allow for precise vertical water sampling providing ample amounts of water for analysis in a low cost, uncomplicated, and safe manner.