Correlation

When analysed, the correlation between particle size (d50) and metal concentration was shown to be ‘very weak’ based on the definition by (Evans, 2006). It has been widely reported that, in highway runoff with smaller particle sizes generally show higher concentrations of metals (Herngren et al., 2006; Dong and Lee, 2009), due to the increased surface area compared those with larger PSD for a given volume.

It is possible that these results are due to agglomeration of particles, meaning that ‘larger’ particles are actually a number of smaller particles. If this were the case, the larger particles would have the higher concentration of pollutant, as they would inherit the sum of the constituent smaller particles. Agglomeration of particles is likely to occur during the build-up of sediment on the car park surface, where they are resident for the time between rainfall events.

Another possibility is that the range of particle sizes acquired by rainfall runoff process is too small to demonstrate this trend. It is possible that given a larger range of intensities, different particle sizes may be mobilised, i.e. the energy from the intensities in this study have kinetic energy too similar to each other to provide a range of particle sizes.

4.4.8 First Flush

One of the objectives of this study was to determine whether a ‘first flush’ of pollutants occurred in runoff entering a channel drain. This is, to say, an initial higher load of pollutants, caused by a prior build up in the period since prior wash-off from a rainfall event.

If a first flush had occurred during the study, a peak in mass and/or pollutant concentration would have been expected over one of the time periods. The results show that neither parameter was shown to vary with time, so therefore no first flush effect was demonstrated.

It is unlikely that the reason for a lack of first flush was insufficient time for it to take effect. This is because by definition, first flush comes from the initial stages of runoff. The results show that sediment and associated pollutants have been washed off. Therefore, further wash-off should not be considered as initial, or ‘first’.

The only way in which a first flush could have occurred, is in the event of a greater influx of sediment/pollutant in the first 5 minutes, with a lower than average amount in the remainder of the 10 minute interval. This is deemed to be unlikely, given no variation was shown throughout the whole duration of the sampling.

This could be examined further by shorter time periods, but as detailed in the Results section, for lower intensities, no runoff was generated in the first 5 minutes, hence the first measurement period was 0 – 10 minutes.

Although a first flush has been demonstrated multiple times (Deletic and Maksimovic, 1998; Wicke, 2009) it is not universally accepted. Bach (2010) argue that the number of variables affecting concentrations makes it difficult to see a general trend. Therefore, although a first flush effect was not demonstrated in this study, the number of variables may account for this, and a repeat study may differ.

4.4.9 Relevance to Legislation

The Draft National Standards for SUDS (2015) suggest that the first 5 mm of runoff should be stored on site, and not discharged, due to the potential pollutants. From the results of this study, 5mm would have fallen at 12.5, 7.7 and 5.6 minutes for the 24, 39, and 54 mm/h intensities respectively.

If this was to be substantiated based on the results of this study, mass and/or pollutant concentration would have peaked at the 20 minute measurement for 24 mm/h, and the 10 minute measurement for 39 and 54 mm/h, the points at which 5mm of runoff would be generated. The results show that neither

parameter was shown to vary with time, so therefore there does not appear to be a higher pollutant load in the first 5mm of runoff based on this study.

4.4.10 Comparison to Highways

Although using a rainfall simulator, this study was intended to be representative of actual rainfall events affecting a car park. Crabtree et al. (2006) covered a range of natural rainfall events, and thus presumably covered a wide, representative range of intensities. Therefore, comparing the results obtained in this study with Crabtree et al. (2006) will indicate whether they are representative of natural rainfall. Table 4.3 displays the results of this study to those obtained by Crabtree et al. (2006).

Table 4.3 Comparison of Results With Highways Study (Crabtree, 2006)

There was a greater range of values obtained in this study, and for all metals but Cd, the maximum concentration was higher. The average values also followed a similar trend. This is likely due to the intensities used in this study, being relatively high, compared to Crabtree et al. (2006) which encompassed all rainfall events. Despite this, the average concentrations are all in the same order of magnitude.

It can also be observed that the average concentrations reported by Crabtree et al. (2006) all fall within the range of values from this study. This suggests that

the runoff from a car park is similar to that of a highway. It also suggests that the methodology of using a rainfall simulator is robust, and comparable to rainfall events.

4.5 Conclusion

Rainfall intensity was shown to generally have an effect on the chemical characteristics (heavy metal concentrations), but not the physical characteristics (particle size) at the intensities studied.

The effect on metals concentrations was shown to be non-linear, with variation generally shown between the low intensity and the two higher intensities, but not between the higher intensities themselves. This is likely due to Capacity Factor (Cf), the capacity of a given rainfall intensity to mobilise sediments pollutants.

Time was shown not to have an effect on the physical or chemical characteristics of the runoff. Therefore, no first flush effect was demonstrated in the runoff generated from the intensities studied.

4.5.1 Further Work

The key limitation of the study was the inability to simulate lower rainfall intensities, more representative of the common events in the UK. Therefore, further work would be recommended using a purpose built rainfall simulator, as with the other studies mentioned throughout, in order to get the desired intensity. The unique element from this study of capturing the runoff entering the channel drain could then be applied.

Additionally, more samples could be taken, on more sites. This would require greater resources, and committed partners on the sample sites in order to aid with logistics, such as providing thousands of litres of water to refill the water butts.

5 Assessment of Wet Vac Sediment Collection Method

as a Representation of Rainfall

This paper is prepared as a short communication to a journal such as Science of the Total Environment.

The following communication bridges the two preceding chapters, which cover the accumulation and transport phases of the Conceptual Model. The validity of the sampling method used to collect accumulated sediment is compared to rainfall events, demonstrating which process affects characteristics.

Source  Accumulation 

_Transport

The objective of this communication is to: ‘Validate the wet-vac method as representative of a rainfall event.’

5.1 Introduction

The ‘wet-vac’ method of collecting sediment from urban areas has been utilised extensively. Studies including those by Bris et al. (1999), Deletic and Orr (2005) and Kayhanian et al. (2012) used the method, and as such, a version of this was used in Chapter 4 as the main sampling methodology. Analysis of the effectiveness of the method has been reported as 95 % by Kayahnian et al. (2012), which is interpreted as collection of 95 % of particles on the highway/carpark surface.

Additionally, the use of simulated rainfall has been used successfully in studies such as Herngren (2005), Egodawatta (2007), Egodawatta (2008) and Egodawatta (2009). It was also used in Chapter 5 of this study to determine the characteristics of runoff entering a channel drain.

Whilst the high level of collection efficiency of the wet vac method reported by Kayhanian et al., (2012) is of great use in determining what builds up and remains on a paved surface, it is unknown if this is representative of what washes off in a rainstorm. As documented in Chapter 5, analysing the actual

runoff is logistically rather difficult, requiring either labour intensive or highly weather dependent manual sampling, or expensive autosamplers. The aim of investigating runoff into channel drains posed extra complications to those of, for example, a drainage ditch, due to the smaller size.

Therefore, this objective of this study is compare the characteristics of samples taken from the ‘wet-vac’ method, to those collected in runoff in order to determine whether the method is representative of sediment transported by runoff generated from a rainfall event.