MAXIMUM ALLOWABLE CONCENTRATIONS THE PERCENTILE APPROACH John Batty Environment Agency EU CMA Plenary Meeting Berlin 2 May 2007
Introduction: How to use MACs In Council Working Party discussions a range of views on the approach for short term standards have been expressed: Reference values only (not used in compliance) Mandatory values (used as component of classification) Absolute standard Percentile standards Whichever of these is selected one thing is clear: We should investigate immediately if the standard is exceeded…. Trigger further monitoring
“Ideal” Environmental Quality Standards To achieve consistency an EQS should include: A limit - e.g. a concentration of 0.07ug/litre A “Summary statistic” - how often the limit may be exceeded- e.g. 5% of the time The period over which compliance assessed The degree of statistical confidence required to demonstrate non-compliance The proportion of the time over which assessment takes place for which failure is acceptable e.g. 1 year in 10 calendar years
Assessing compliance with Short Term Standards There are 31,000,000 seconds in a year Yet we will only sample, at best, a few hundred of these The problem is that we do not know the quality for the rest of the time If a standard is to be considered an absolute value - it must NEVER be failed! Do we really mean that? Continuous monitoring provides the only reliable mechanism to demonstrate absolute compliance. UNAFFORDABLE ! In reality, we tolerate failure some of the time - when we are not monitoring. A percentile approach
Breaches of Short Term Standards
Effect of sampling rate on compliance
Assessing compliance with Short Term Standards We have seen that an absolute limit is a poor compliance measure unable to capture all events - risk of false negatives increased sampling frequency will show more failures while low rates of sampling favour the polluter under very frequent sampling, the standard will almost always be exceeded
Effect of sampling rate on the severity of an absolute limit -1
Effect of sampling rate on the severity of an absolute limit -2
Proposal ... If we are to implement a mandatory short-term standard it should be expressed as a 95% percentile (i.e. we allow the concentration to be exceeded for 5% of the time)with a 50% level of confidence. A look-up table may be provided that sets out the number of failures permitted for a given sampling rate