All measured fluxes in this thesis were calculated and filtered to meet stringent
quality controls using a programme written in LabVIEW. In brief, this programme,
which developed further between each measurement campaign, read in flux data files,
applying the standard rotations (Baldocchi, 1993) to the coordinate frame using the
following procedure: Uhcr = < ( V 2 + U 2) (2.5) £/,„= ^ ( W 2+ U hJ ) (2.6) Cos n = U I U ,„ (2.7) Sin r\ = V I U hor ( 2 . 8 ) Coj O = [/* „ ,/t/,„, (2.9) Sin 6 = W / Ul0f (2.10)
Ucorr = -u x Sinq + v x Cos q (2.11)
VCorr= u x Cos 0 x Cos q + v x Cos 0 x Sin q + w x Sin 0 (2.12)
W Corr = - u x Sin 0 x Cosq - v x Sin 0 x Sin q + w x Cos 0 (2.13 )
where, U, V and W represent the average 3D wind components averaged over a given
flux averaging period (typically 30 minutes) and u, v, w are their instantaneous values.
This sets the averages o f the vertical and cross-wind components (v and w) to equal
Chapter II 67
corrected. Once rotated, data from the sonic anemometer was used to calculate fluxes
o f sensible heat:
H = WT ' p Cp (2.14)
momentum:
t = w ' u ' (2.15)
and frictional velocity:
( T \
U.
\ p J
(2.16)
as well as the M onin-Obukov length which is a measure o f the height at which
mechanical turbulence (wind shear) gives way to buoyant production o f turbulence
(Stull, 1988). Here K is the von Karman’s constant (0.41) and g is the acceleration due
to gravity (9.8 m s '1)
L = - o C v u*3 T (2.17)
K g H
The average wind speed (WS) was calculated using Eq. (2.18) and averaging periods
where the mean wind speed dropped below 1 m s '1 were rejected and not included in
the final data analysis.
t2 , t / 2 , rjr 2
WS = 'l( U 2 + r + W ) (2.18)
Concentration measurements o f VOC in ion counts per second (ICPS) were converted
v o c iR H , x 1 x 1 ° 9 x Pstu x 224Q0x { L , j + L ) x T r a n s ) ( 219)
PPB ( k x t x M 2 1 X M3 1 x p d x 6.022 x 10 23 x Tsld x TransRH,)
where ft/7, and TransRH,, were the ion count and transmission o f the target VOC, Ts,d
and Pstd standard temperature (273.15 K) and pressure (1013 mbar), Td and Pd, the
drift tube temperature and pressure, M21 and Trans, the primary ion count and
transm ission number, M37, the ion count o f the reagent water cluster and k and t, the
9 3 1
reaction rate constant and reaction time which had values o f 2 x 10' cm s' and 1.05 x
1 O'4 s respectively.
PPB values were then converted into fig VOC m '3 using the relationship:
XvoclMg m Xvoc [ppb]x l E ^ x IE* x M r
f 1013x22.4x(273 +
r„,n,t.)
( 273 x p
(2.20)
where M r was the molecular weight o f the target VOC, P was the ambient pressure in
mbar and Tsonic was the temperature o f sampled air as measured by the sonic
anemometer. These values were then paired with the associated vertical wind velocity
and the flux was calculated using Eq. (1.6). The resultant flux was then multiplied by
3600 s h '1 to give a measurement o f the VOC flux in units o f fig m 2 h ’.
Finally calculated fluxes were subjected to a data quality test whereby each
averaging period was tested for non-stationarities. The stationarity test followed the
theory outlined by Foken & Whichura (1996), which states that a time series % is
stationary, when the flux (Fx) is equal to the mean average flux o f its components {Fxj,
Fx2, Fx3...). Here we took Fx to be the flux over the averaging periods, and the
components Fx, to Fx6 to be the flux calculated from individual 5 minute blocks o f the
Chapter II 69
o f Fxi - Fyfi differed by more than 60% o f the value o f Fx the time series was
considered non stationary and the data were discarded. Time series where the fluxes
differed between 30% and 60% were considered stationary, but to be o f a lower
quality. High quality stationary data was taken to be any time series where the fluxes
Chapter III
3.
Mixing ratios and eddy covariance flux
measurements of volatile organic compounds from an
urban canopy (Manchester, U.K.)
This chapter details the deployment o f the disjunct eddy covariance sampling
system, developed in chapter 2, for the measurement o f VOC above the city o f
Manchester. The sampler was located on the roof o f Portland Tower, an 80 m office
block located in the city centre, and sampled for a period o f three weeks between the
5th and 21st o f June 2006. In addition to the disjunct eddy covariance flux
measurements, a secondary sampling technique, virtual disjunct eddy covariance was
also deployed, tested and compared to results from the DEC system. As each system
relied on the PTR-MS for VOC concentration measurements, the two systems
operated in alternate half hours allowing an indirect comparison o f the two techniques
to be made.
The following work was submitted to the journal o f Atmospheric Chemistry
and Physics on the 4th o f December 2007 and accepted for publication in ACPD on
January the 8th, 2008. The manuscript is currently under review. The authors and their
contributions are listed below.
Ben Langford (Lancaster University & CEH): Developed the disjunct eddy covariance system and software, developed the virtual disjunct eddy covariance software, operated the instruments during the field campaign, post processed the raw data and wrote the manuscript.
Brian Davison (Lancaster University): Helped compile the hardware for the disjunct eddy sampler, helped with the installation o f the systems and with the compilation o f the manuscript.
Eiko Nemitz: (CEH) Co-wrote the software for the virtual disjunct eddy covariance system, helped with the installation o f the systems and the compilation o f the manuscript.
Chapter III 71
Nick H ewitt (Lancaster University): Helped with interpretation o f results and the com piling o f the manuscript.
Mixing ratios and ed d y c o v a r ia n ce flux m e a su r e m e n ts
o f v o la tile o rg a n ic c o m p o u n d s from an urban ca n o p y
(M an ch ester, U.K.)
B. Langford1, B. Davison1, E. Nemitz2 and C. N. Hewitt1.
[1] Lancaster Environment Centre, Lancaster University, Lancaster, LA I 4YQ
[2] Centre for Ecology & Hydrology (CEH) Edinburgh, Bush Estate, Penicuik, EH26
OQB, U.K.
Correspondence to: C.N. Hewitt ([email protected])
Abstract
Concentrations and fluxes o f six volatile organic compounds (VOC) were
measured above the city o f Manchester (U.K.) during the summer o f 2006. A proton
transfer reaction-m ass spectrometer was used for the measurement o f concentrations,
and fluxes were calculated using both the disjunct and the virtual disjunct eddy
covariance techniques. The two flux systems, which operated in alternate half hours,
showed reasonable agreement, with R2 values ranging between 0.2 and 0.8 for the
individual analytes. On average, fluxes measured in the disjunct mode were lower
than those measured in the virtual mode by approximately 19%, o f which at least 8%
can be attributed to the differing measurement frequencies o f the two systems and the
subsequent attenuation o f high frequency flux contributions. Observed fluxes are
thought to be largely controlled by anthropogenic sources, with vehicle emissions the
Chapter III 73
a fraction o f the isoprene present. Fluxes o f the oxygenated compounds were highest
on average, ranging between 60 - 89 pg m'2 h’1, whereas the fluxes o f aromatic
compounds were lower, between 19 - 42 pg m'2 h '1. The observed fluxes o f benzene
were up-scaled to give a city wide emission estimate which was found to be
significantly lower than that o f the National Atmospheric Emissions Inventory
(NAEI).