Daily Air Quality Index - reporting performance assessment
Having fully adopted the Daily Air Quality Index (DAQI) when it came into force on 1 January 2012, King’s have examined the index’s reporting performance during two recent particulate (PM) episodes.
We examined the impact of reporting PM using the new fixed midnight to midnight day required by the new DAQI, as opposed to the running 24 hour mean used by the previous index and still widely reported by other networks.
The new DAQI also contains the concept of "triggers". Triggers are based on hourly values and predict the pollution level at a monitoring site in advance. This short-term predictive element is a key component of the additional public information provided by the new DAQI.
The first episode examined was a PM2.5 dominated event where pollution peaked on 31 January 2012. Trigger performance was examined at all 76 PM10 and PM2.5 analysers in our database from across the southeast and London.
The daily banding was correctly predicted in advance at 71 sites with the remaining 5 being PM2.5 analysers where index level 6 (moderate) was predicted but index level 7 (high) was attained by the end of the day. This under estimation for PM2.5 in some locations was attributed to the atypical temporal distribution of PM2.5 during this episode.
The original work by King’s for the Committee on the Medical Effects of Air Pollutants (COMEAP) to calculate the triggers included AURN data from across the country. In all, 270,000 site days were used in the analysis to calculate the PM10 triggers. We were therefore also keen to examine the performance of triggers outside the southeast and London.
We examined trigger performance using data from the AURN PM10 instrument at Headingley in Leeds. This site was chosen because pollution peaked at ‘high’ and hence enabled analysis of transition from ‘low’ through ‘moderate’ to ‘high’ and back again.
As in the southeast, the episode started building on 30 January, peaked on the 31st and then cleared late on 1 February. On all days examined (30 January to 2 February), DAQI defined triggers correctly predicted the forthcoming day’s pollution levels. Further, as COMEAP intended, the triggers responded rapidly to rises in ambient levels and enabled prediction of daily levels as early as 0800h.
The redundant system of reporting running 24 hour running means by contrast would have introduced a lag in response and would not have reported the ‘high’ peak until the last hour of the 31st whereas the ambient levels peaked and the trigger activated 9 hours earlier.
Further, the running 24 hour mean reporting would have continued to report ‘high’ right through into the afternoon of the following day when DAQI levels were ‘moderate’. This lag was repeated into 2 February where running 24 would have reported ‘moderate’ for 8 hours despite the episode clearing in the afternoon on the previous day.
In their
review COMEAP specifically considered this lag issue of running 24 hour reporting (section 3.3) and recommended the change to fixed 24 hour day reporting combined with triggers. This recommendation was
formally adopted by Defra for the DAQI.
The second episode examined was primarily driven by local emissions during 6 February 2012. During this PM10 dominated episode, alerts triggered on LondonAir, and on our other websites and apps, correctly predicted the day’s banding at all 21 analysers which measured ‘moderate’ pollution.
The two specific aspects of the DAQ examined here, fixed midnight to midnight reporting and the use of triggers to predict pollution levels earlier in the day have clearly resulted in better and more accurate reporting of pollution to the public.
