Upload
others
View
7
Download
0
Embed Size (px)
Citation preview
Improvements to the nitrogen dioxide observations by
means of the MKIV Brewer spectrophotometer
H. Diémoz 1,2 A. M. Siani 2 V. Savastiouk 3 A. Redondas 4 K.Eleftheratos 5 C. S. Zerefos 5 M. Vrekoussis 6 G. R. Casale 2
M. Stanek 7 J. P. Burrows 8 A. Richter 8 A. Hilboll 8
1ARPA Valle d'Aosta, 2Sapienza Università di Roma, 3IOS, 4AEMET, 5Academy of Athens,6EEWRC, 7�HMÚ and 8University of Bremen
14th WMO-GAW Brewer users group meetingTenerife, March 2014
1 Introduction
2 Instrumental characterisation
3 Updates to the algorithm
4 Field measurement campaigns
5 Reprocessing of long-term data sets
6 Future research
IntroductionMeasurements with Brewer spectrophotometers
MKIV
426 � 453 nm (visible)
original algorithm: Kerr (1989)
updates (not implemented):Barton (2007)
MKIII
349 � 363 (UV-A)
Cede et al. (2006)
IntroductionMeasurements with Brewer spectrophotometers
What's wrong with NO2 measurements using the Brewer?
overestimations (>200%) with thecurrent algorithm
I other ground-based instrumentsI satellite radiometers
random deviationsI due to interferences (O4, H2O, Ring)I sensitivity to instrumental settings
(e.g. wavelength misalignments)
noiseI depends on both the algorithm and
used wavelengths
no calibration service for NO2
IntroductionMeasurements with Brewer spectrophotometers
Open questions
can we obtain better performances bychanging the operational wavelengths?
how large is the measurementuncertainty?
do direct sun and zenith sky estimatesagree within their uncertainties?
how to perform a Langley plot with avariable absorber?
is it possible to reprocess long-termseries?
IntroductionMeasurements with Brewer spectrophotometers
What was done
1 mathematical framework and numericalsimulations
I accurate characterisation of Brewer#066
I parameterisation of all in�uencingfactors
2 updated spectroscopic dataset
3 new (polarised) AMFs for ZS geometry
4 variable-Langley calibration (Izaña)
5 Montecarlo uncertainty budget
6 reprocessing of 4 long-term series
1 Introduction
2 Instrumental characterisation
3 Updates to the algorithm
4 Field measurement campaigns
5 Reprocessing of long-term data sets
6 Future research
Instrumental characterisation
Fundamental to reveal the weak signal of NO2
dispersion
�neutral density� �lters
temperature dependence
etc.
[Diémoz et al., 2011, Brewer Meeting, Beijing]
1 Introduction
2 Instrumental characterisation
3 Updates to the algorithm
4 Field measurement campaigns
5 Reprocessing of long-term data sets
6 Future research
Updates to the algorithmNoise reduction
~γ ·−→βR ≡ 0
~γ · −−→αO3 ≡ 0
~γ ·−→δA ≡ 0
~γ ·~1 ≡ 0
~γ ‖ αNO2
6 slits =more degrees of freedom =
noise reduction
Updates to the algorithmInterfering factors
Considered factors
noise
oxygen dimer (O4)
water vapor (H2O)
Ring e�ect
sensitivity to NO2
e�ective temperature
wavelengthmisalignments 0
0.01
0.02
0.03
0.04
0.05
400 420 440 460 480 500
opticaldepth
wavelength (nm)
totalO3
NO2H2OO4
CHOCHO
Updates to the algorithmIncreasing the DOF
0
1
2
3
4
5
0 500 1000 1500 2000 2500 3000
noise
(DU)
micrometer steps
For every in�uencing factor a diagram was assessedand the optimal grating positions (i.e. wavelengths) were found
Updates to the algorithmJump scans
420 430 440 450 460 470 480
wavelength (nm)
a)
b)
c)
11 slitsO4 and H2O included in the retrieval
Updates to the algorithm
Further updates
all cross sections were updated to recent laboratory measurements(NDACC 2012 guidelines)
I NO2 - Vandaele 2002 @ 220 K (instead of Graham 1976)I O3 - Bogumil 2003 @ 223 K (instead of Vigroux 1952)I Rayleigh - Bodhaine et al. 1999 (instead of UV Brewer coe�cients)I O4 - Hermans 2003 (previously neglected)I H2O - Hitran + Py4CATS (previously neglected)
I0-e�ect taken into account
new weighting factors
polarised AMFs for ZS geometry (SCIATRAN, full-spherical)
Updates to the algorithmRadiative transfer calculations
0.9
1
1.1
1.2
1.3
1.4
1.5
0 10 20 30 40 50 60 70 80
DU
solar zenith angle (deg)
[Diémoz et al., 2013, SPIE, Dresden]
1 Introduction
2 Instrumental characterisation
3 Updates to the algorithm
4 Field measurement campaigns
5 Reprocessing of long-term data sets
6 Future research
Field measurement campaignsThe Izaña campaign
38 days of measurements
5 days removed because offog/rain
no relevant contamination bySaharan dust on ratios
2 days for initial checks
16 days for direct-sunmeasurements (3 steps)
15 days for zenith-skymeasurements (3 steps and 2polarizations)
5 days for O3 and O4 direct-sunmeasurements
Field measurement campaignsLangley in varying conditions
Improving the Langley: linear changes with time
y(t)
µ(t)=a
1
µ(t)+ b(t)
=a1
µ(t)− δ − ξt
(1)
Field measurement campaignsLangley in varying conditions
(a δ ξ
)T= A
†(
y1µ1
... ynµn
)T(2)
Field measurement campaignsLangley in varying conditions
-0.1
0
0.1
0.2
0.3
0.4
8 10 12 14 16 18
NO2columndensity
(DU)
time(hours)
Izana clim. (AM)Izana clim. (PM)
linear �t
site reference daytime increasing rate (1013 molec
cm2 h)
Zugspitze/Garmisch Sussmann et al. (2005) 5�15Izaña Gil et al. (2008) 6
Paci�c Ocean Peters et al. (2012) 8.7±0.5Izaña present study 7.0�12.5
Field measurement campaignsThe Izaña campaign
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
-40 -30 -20 -10 0 10 20 30 40
DU
airmass
directzenith (parallel)
zenith (perpendicular)
Measurements using the three di�erent geometries are equivalentwithin their respective uncertainties
(airmass < 0 morning; > 0 afternoon)
Field measurement campaignsThe Izaña campaign
0.06
0.07
0.08
0.09
0.1
0.11
0.12
0.13
0.14
0.15
0.06 0.07 0.08 0.09 0.1 0.11 0.12 0.13 0.14 0.15
Brewer
(DU)
NDACC (DU)
�ty=x
The Brewer does not overestimate anymore
Field measurement campaignsComparison among di�erent algorithms
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
260 262 264 266 268 270 272 274 276
DU
day
Izaña (2012) � Pink: Kerr's algorithm;green: �lters and Rayleigh updates; blue: new algorithm.
Field measurement campaignsUncertainty budget
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
10 20 30 40 50 60 70 80
NO2uncertainty
(DU)
solar zenith angle
�lters correctionPoisson
dark currentcross sections
NO2 e�ective heightNO2 e�ective temperature
α Angstromwavelengthdead time
O4 correctionresolution
total
Monte Carlo uncertainty budget> 10 factors taken into account
Field measurement campaignsComparison among di�erent algorithms
-3
-2
-1
0
1
2
3
260 262 264 266 268 270 272 274 276
DU
day
Much less noise than Barton's algorithm (2007)...
Field measurement campaignsComparison among di�erent algorithms
-3
-2
-1
0
1
2
3
264.2 264.3 264.4 264.5 264.6 264.7 264.8 264.9
DU
day
... and much smaller dependence on wavelengths misalignments(crosses: hg test misalignments on same scale)
Field measurement campaignsOther retrievable quantities
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
-100 -50 0 50 100
Lp
solar zenith angle
-0.4
-0.2
0
0.2
0.4
0.6
0.8
-100 -50 0 50 100
Lp
solar zenith angle
New product: degree of linear polarisation of the sky(left: clear sky; right: fog)
Field measurement campaignsOther retrievable quantities
-20
-15
-10
-5
0
5
10
15
20
291 292 293 294 295 296
O4opticaldepth
(ratio)
day
New product: oxygen dimer (O2�O2) optical depth(too weak absorption by H2O in the blue band)
Field measurement campaignsThe Saint-Christophe campaign
The Brewer #066 calibration can be successfully transferred
1 Introduction
2 Instrumental characterisation
3 Updates to the algorithm
4 Field measurement campaigns
5 Reprocessing of long-term data sets
6 Future research
Reprocessing of long-term data setsThe stations
�Minimum-Amount Langley Extrapolation�and �Bootstrap Estimation� techniques
Herman (2009)
Reprocessing of long-term data setsWavelength misalignments
-1
-0.5
0
0.5
1
1.5
2007 2008 2009 2010 2011 2012 2013
NO2(D
U)
date
Athens, Brewer #001
Reprocessing of long-term data setsWavelength misalignments
-1
-0.5
0
0.5
1
1.5
2007 2008 2009 2010 2011 2012 2013
NO2(D
U)
date
Athens, Brewer #001
Reprocessing of long-term data setsWavelength misalignments
-0.05
0
0.05
0.1
0.15
0.2
0.25
0.3
1 1.5 2 2.5 3 3.5 4 4.5 5
retrievedNO2(D
U)
µNO2
step 1002step 998
reference step (1000)
Reprocessing of long-term data setsWavelength misalignments
-0.5
0
0.5
1
1.5
2
2.5
2007 2008 2009 2010 2011 2012 2013
NO2(D
U)
date
�Piecewise� calibrationAthens, Brewer #001
Reprocessing of long-term data setsExploratory data analysis
stationIQR (DU)
new algorithmIQR (DU)
old algorithm
overestimationby old
algorithm
Saint-Christophe 0.19�0.3 1.0�1.3 340%Hradec Králové 0.3�0.5 1.3�1.9 275%
Rome 0.28�0.6 1.2�1.8 275%Athens 0.3�0.7 0.8�1.5 120%
Years: 2007�2013
Reprocessing of long-term data setsCorrelation with in situ concentrations
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
DU
2008 2009 2010 2011 2012 201310
20
30
40
50
60
70
80
µg
m3
date
in situ concentrationBrewer VCD
Aosta, Brewer #066 (monthly averages)Spearman's rs = 0.7 when compared to in situ concentrations
Reprocessing of long-term data setsSeasonal cycle
0
0.5
1
1.5
2
J F M A M J J A S O N D
NO2(D
U)
month
Aosta, Brewer #066
temperature inversionsand stagnation
increased photolysisduring summer
increased NO2 lifetimeduring winter (lowertemperatures)
increased emissions inwinter
Reprocessing of long-term data setsWeekly cycle
0
0.5
1
1.5
2
2.5
S M T W T F S
NO2(D
U)
day of week
Rome, Brewer #067Signi�cant di�erences between weekdays and weekends
in all analysed stations
Reprocessing of long-term data setsDaily cycle
0
0.2
0.4
0.6
0.8
1
6 8 10 12 14 16 18
NO2(D
U)
hour (UT)
Aosta, Brewer #066Morning rush hours � No (inverse-)U shape
Reprocessing of long-term data setsSpaceborne estimates
much morecomparable, nowoverall bias -2.4%
low correlation
lack of seasonalitycompared to satellites 0
0.2
0.4
0.6
0.8
1
1.2
2007 2008 2009 2010 2011 2012 2013
DU
date
BrewerOMI
SCIAMACHYGOME2a
Rome, Brewer #067Monthly averages
1 Introduction
2 Instrumental characterisation
3 Updates to the algorithm
4 Field measurement campaigns
5 Reprocessing of long-term data sets
6 Future research
Future research
AOD retrieval at 440 nm (STSM)
more accurate Brewer-satellites comparison (STSM)
Ring e�ectI more accurate zenith sky measurements
stratosphere-troposphere partitioning (ds + zs)
Moon measurements (already started)
View publication statsView publication stats