|
Discussion |
1 Groundwater
Tracing Unit, Department of Earth Sciences, University College London,
Gower Street, London WC1E 6BT,
UK
2 School of Environmental
Sciences, University of East Anglia, Norwich NR4 7TJ,
UK
3 White Cottage, Lower
End, Great Milton, Oxford, OX44 7NL,
UK
4 3 Grangefield Way,
Aldwick, Bognor Regis, West Sussex, PO21 4EG,
UK
5 both formerly with WRc
plc, Medmenham, Marlow, Bucks, SL7 2HD,
UK
T.C. Atkinson writes: Highway drainage has long been recognized as a hazard to groundwater abstractions (e.g. Atkinson & Smith 1974; Price et al. 1992; Hiscock et al. 1995). Lacey & Cole (2003) have performed a valuable service by setting out clearly the methods and data sources from which the risk of a spillage from road or rail tankers can be estimated. Their road example indicates a frequency of about one spill in 740 years per km of trunk road outside built-up areas. Expressed in this way, as an average recurrence interval between spills on a short section of road, spillages may seem comfortingly rare (although Lacey & Cole do not say this). A casual reader might draw the conclusion that traffic on highways in general presents only moderate risks of contaminating water supplies. This is misleading because Lacey & Cole present only the annual risk, i.e. the probability that a spillage might occur in any period of one year. For water supplies it is better to consider the total risk over the whole period in which the hazard might exist. Because roads and railways can be expected to continue in use for decades or even for over a century, the overall risk of a spillage occurring during their lifetime of use may be quite large.
Lifetime risk may be calculated from annual risk as follows, using data from Lacey & Cole (2003). The annual risk on a 2 km stretch of road was 0.0027 or one spill per 370 years. It follows that the probabilitythat there will not be a spillage during any one year is(1 – 0.0027) = 0.9973. Now estimate the lifetime of the hazard, i.e. the length of time in years for which the road and the water abstraction will both be in use. For the sake of argument, suppose this is 30 years. Assume also that the annual risk of spillage on the relevant sectionof road will remain unchanged over this time. The probability that there will be no spillage at all duringN years is given by the Nth power of the annual probability. Thus for N = 30 years, the probability ofno spillage is (0.9937)30 = 0.9221. Finally, there must either be no spillage at all during N years, or at leastone spillage will occur. Together these possibilitieshave a combined probability of 100%. Therefore the probability that at least one spillage will occur during30 years is (1 – 0.9221) = 0.07791, or 7.8%.
Thus, in Lacey's & Cole's (2003) example there is an almost 8% chance that a spillage will occur on the road in question during a 30 year period in which tanker flows remain constant. Few people would regard this as a comfortingly low risk for contamination of drinking water.
A more extreme illustration is provided by the M1/M25 motorway interchange in Hertfordshire, England, which presents a hazard to a groundwater abstraction several km away as demonstrated in a tracer study by Price et al. (1992). No analysis of spillage risk for this busy intersection has been published, but one was made at the time of the tracer studies using broadly similar methodology to Lacey & Cole (2003). Motorways are expected to remain in use for over a century and an estimate of 120 years was used, leading to a lifetime risk of 26% that at least one spillage would occur. This risk is high, but it must be considered in relation to the other findings of the tracer study. The tracer concentration was attenuated by more than 107 times during its passage through the Chalk aquifer from a drainage soakaway at the motorway to the water abstraction affected. Details are given by Price et al. (1989, 1992). Engineering and procedural measures were subsequently put in place to manage future spills and reduce the likelihood that liquid contaminants might enter the soakaway following an accident.
Table 1 shows lifetime risk as a function of annual risk and the duration of the hazard. A hazard exists for as long as a target for contamination and a potential source both exist simultaneously. Lifetime risks are c.10% or more wherever the duration of a hazard exceeds one tenth of its recurrence interval, which will often be the case for roads or railways. Reducing annual risk is intrinsically difficult and not a very effective response, as lifetime risks will remain above 1% even if annual risk is as low as 1 in 1000. These high lifetime risks strongly reinforce Lacey's & Cole's conclusion that hazardassessment, pollution prevention measures and development of contingency plans are elements of soundmanagement as far as protection of public water supplies are concerned. As roads, motorways, railways and water abstractions are all designed and managed by different bodies, active cooperation and coordination between them is needed. The references given below emphasise the value of GIS methods and tracer studies for hazard identification and assessment.
|
Whether the concept of lifetime risk is necessarily ‘better’, or more appropriate, than annual risk seems to the authors to depend on the context and purpose for which the estimate of risk is to be used. If, as suggested in the paper, the risk of spillage is quantified in terms of the expected rate r of occurrence of spills per unit of time (and if this rate is assumed to remain constant), the expected number of spills in an interval of duration t would be rt. The probability of experiencing at least one spill in such an interval would then be given by 1 – exp(–rt). The conversion between annual and lifetime risks is therefore very straightforward.
Regarding the wording of the third paragraph of the discussion, one should beware equating the chance of a spillage, as estimated in the paper, with the risk of contamination of drinking water. This is because not all spills give rise to contamination of a water-source. Also, as we have pointed out, the possibility of a spill turning into a contamination event can and should be reduced by structures and emergency procedures designed to prevent the spilt material from entering the aquatic environment.
 |
References |
|---|
 TOP References |
|---|
Atkinson, T.C., Smith, D.I., Rapid groundwater flow in fissures in the Chalk: an example from south Hampshire. Quarterly Journal of Engineering Geology, 7 1974. 197–205.
Hiscock, K.M., Lovett, A.A., Brainard, J.S., Parfitt, J.P., Groundwater vulnerability assessment: two case studies using GIS methodology. Quarterly Journal of Engineering Geology, 28 1995. 179–194.
Lacey, R.F., Cole, J.A., Estimating water pollution risks arising from road and railway accidents. Quarterly Journal of Engineering Geology and Hydrogeology, 36 2003. 185–192.
Price, M., Atkinson, T.C., Wheeler, D., Barker, J.A., Monkhouse, R.A., Highway drainage to the Chalk aquifer: the movement of groundwater in the Chalk near Bricket Wood, Hertfordshire, and its possible pollutionby drainage from the M25. Technical Report WD/89/3 1989. British Geological Survey.
Price, M., Atkinson, T.C., Barker, J.A., Wheeler, D., Monkhouse, R.A., A tracer study of the danger posed to a chalk aquifer by contaminated highway runoff. Proceedings of the Institution of Civil Engineers, Water Maritime and Energy, 96 1992. 9–18.[ISI]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
