Air quality expert group - Fine particulate matter (PM2.5) in ... - Defra

PM2.5 in the UK
164
A2.6: Photochemical Trajectory Model (PTM)
43. The PTM moving air parcel trajectory model has been used to estimate the
mid-afternoon mass concentrations of a number of suspended particulate
matter (PM) components for each day of 2006 at a rural location in Harwell,
Oxfordshire, southern UK. A large number of equally probable and randomlyselected
96-hour three-dimensional air mass trajectories were used to describe
the variability of the atmospheric transport paths during each day. A chemical
kinetic description was given for the major PM formation processes. Further
details are given in Derwent et al. (2009b).
PM2.5 Component, µg/m 3
25
20
15
10
5
0
January
DEVELOPMENT OF A MASS CLOSURE SCHEME FOR THE PTM
February
March
April
May
Biomass fires in Russian
Federation
June
July
August
September
October
Number
December
Harwell, Oxfordshire 2006
PM OC
PM EC
PM NH4
PM NO3
PM SO4
Partisol PM2.5
Figure A2.6.1: Mass closure for PM2.5 as predicted by PTM for Harwell,
Oxfordshire, 2006.
44. The PTM output comprised the histories of the chemical development of the 51
model species along the 30 randomly-selected 96-hour trajectories arriving at
Harwell, Oxfordshire, at 15:00 (GMT) on each day of 2006. Examination of the
30 sets of results for each day and each model species revealed the presence
of a significant amount of scatter about the average result for the arrival point.
There was evidence that this scatter increased dramatically on more polluted
days compared with cleaner days. There was apparently much more variability
in the model description of air mass origins on polluted days compared with
clean days. This reflected the increasing stagnation and decreasing wind speeds
associated with pollution episodes. The arithmetic mean of the 30 sets of results
was taken as the best estimate of the model result for that day and these
estimates have also been evaluated against observations for 15:00 (GMT) (see
Figure A2.6.1).
45. Model performance requirements have focused on achieving mean fractional
biases in the range ±0.2 and 50% of the model results within a factor of two
of the observations. On this basis, model performance against observations was
found to be satisfactory for NOx, NH3, HNO3, PM ammonium, PM sulphate, PM
nitrate and PM2.5. The lack of adequate observations precluded a satisfactory
evaluation of model performance for EC and OM.