Upload
others
View
7
Download
0
Embed Size (px)
Citation preview
T i t l e
Atmosphere Monitoring
CAMS-71 policy user workshop
22. January 2018, Copenhagen, EEA
Source-receptor calculations for main European City Areas: status of 2017 products and their validation
Michael Schulz (MetNo)Martijn Schaap (TNO)Alvaro Valdebenito, Augustin Mortier, MatthieuPommier, Hilde Fagerli, Richard Kranenburg
AtmosphereMonitoring
Service Overview: Regional contributions to pollution in EU cities
A B
AtmosphereMonitoring
How to get there on the CAMS website ?
Click here
http://policy.atmosphere.copernicus.eu
Click here
Click here
Click here
Or bookmark:
AtmosphereMonitoring
Daily Forecasts of Source Contribution Split
• Available for 34 European Cities
• 4 Day forecasts of PM10 and Ozone
• Source contribution split into
Local – National – EU – Other countries Natural
• Country source contribution split (experimental)
• Access to previous periods
• Consistent with CAMS 50 forecasts
• Results available in json file format
A
AtmosphereMonitoring
Source Contribution Split // Example PM10 Copenhagen Today
(Others=Natural+Rest of World)
(Others=Rest of Europe+World)
A
AtmosphereMonitoring
Episode analysis
• Features as for daily forecasts
• BUT
• With two models (EMEP + TNO) for PM10
EMEP model for Ozone
• Report available one month after episode
• Activation with more output
on demand (up to 5 times per year)
• Service evolution test
Daily simulation from April 2018 onwards
B
AtmosphereMonitoring
We provide:• 4 Day forecasts of PM10 and Ozone• Available for 34 European Cities• Source contribution split into
– Local– National – EU– Other countries – Natural
We operate two models:• EMEP• LOTOS-EUROS
We use two source apportionment techniques:• Ensemble based (EMEP)• Labelling (LOTOS-EUROS)
Why?• Increases service reliability• Allows service development
S o u r c e a p p o r t i o n m e n t s e r v i c eB
AtmosphereMonitoring
Can we trust in these results?
Are results model dependent? (Are the models simulating the same?)
How much do results depend on city area (=local) choice?
Are there non-linearities affecting the result?
Can we validate with measurements?
In-Depth Investigation undertaken with several model experiments using harmonised EMEP and TNOfor December 2016 Episode
AtmosphereMonitoring
Are results model dependent?
Brussels4-7 Dec 2016
PM10ChemicalCompositionevolution hourly
for comparableComponents
TNO
EMEP
Composition split
AtmosphereMonitoring
Are results model dependent?
BrusselsPM104-7 Dec 2016
TNO
EMEP
Source Country split
AtmosphereMonitoring
Are results model dependent?
Brussels4-7 Dec 2016
ElementalCarbonSootCountrySourceContributionsTNO
EMEP
Source Country to Soot/Elemental Carbon split
AtmosphereMonitoring
How much do results depend on city area (=local) choice?
Gridded administrative areaof the city
1 model grid cellin the center
9 model grid cellon the center
Whats the best choice for the city area? => three experiments
AtmosphereMonitoring
Two different city examples and city definitions used in experiments
Berlinbig city
Zagrebsmall cityRather large Admin
AtmosphereMonitoring
Local/External Pollution split for PM10 for 3 city area definition experiments
4-7 Dec 16Berlin
Zagreb
ADMIN area 1 grid box area 9 grid box area
AtmosphereMonitoring
W h a t s h a l l w e u s e f o r c i t y a r e a s ?
Gridded GADM administrative
area of the city
1 model grid cellin the center
9 model grid cellon the center
Gridded OECDFunctional
Urban Areas
Admin and 9 grid results are mostly similar, small cities most sensitive to definition=> we are open for suggestions what to use!
?
AtmosphereMonitoring
Are non-linearities affecting the results?
PM10 Hourly Contributions Dec 2017
Others gets negativeOthers = Total-(Local+EU)
Now and then, yes….
2 EMEP experimentsWith different perturbations5% and 50% emission reduction(Instead of 15% standard)
AtmosphereMonitoring
Are there non-linearities affecting the results?
4-7 Dec16Rome
EMEP model two experiments
50% emission perturbation experiment
5 % emission perturbation experimentOthers gets negative
Others = Total-(Local+EU)-
++-
AtmosphereMonitoring
Are non-linearities affecting the results?
PM10 Hourly Contributions Dec 2017Now and then, yes….
But is it a problem?
Better statistics will be Produced…
AtmosphereMonitoring
Can we validate with measurements?
Berlin
Example PM10, Berlin, 2.12.2016Size of dot = measured PM10 concentration
City sites
Background sites
The idea:Derive City concentrationand Background concentrationfrom NRT observations
Derive External concentrationFrom difference City-Background
Plot latest NRT observationsalong with Source contribution time series
AtmosphereMonitoring
PM 10 measurements in episode Dec 2016 near Berlin
Hourly PM10 at stations in outer circle around Berlin (20-70 km distance to center)
Hourly PM10 at stations in center of Berlin (0-20 km distance to center)
There is potential butalso further work needed
AtmosphereMonitoring
Example of two days of PM10 measurements
Berlin
Size of dot = measured PM10 concentration
Dec 2Dec 4
Dec 4 large external contributionDec 2 relatively large local contribution
AtmosphereMonitoring Reanalysis source allocation PM10
Model results for UBA station Neuglobsow
Evaluation using source tracers
Wood combustion through Levoglucosan
E v a l u a t i o n s o u r c e p r o f i l e s - U B A p r o j e c t
Levoglucosan measurements
Mo
del
led
wo
od
co
mb
ust
ion
OC
AtmosphereMonitoring
Summary
• Hourly local/external SR relationships are available for PM10 and Ozonefor 34 EU cities and in more depth for selected episodes in 2017
• Results are available as 4 day forecasts since winter 2016
• EMEP results are very similar to TNO model, despite different SR method
• SIMILAR if meteorology, city area, emissions are harmonised
• Major differences can be attributed to model differences in aerosolcomponent treatments (seasalt, road dust, water, SOA …)and boundary layer parameterizations
• Non-linearities seem to influence Local/External split by up to ca 20%
• With current model resolution (0.25x0.125 degrees)the city definition is more relevant for small cities for SR results
AtmosphereMonitoring
Outlook / Plans
• More cities will be included to fill gaps on map ( Milano, Frankfurt, Lyon, Lille )
• Publication of December 2016 episode study
• Inclusion of January 2017 episode in interim assessment report
• More episodes in 2018, flexible hindsight activation by all partners and users
• New choice of city area definition (“local”), ideally with user input
• Experimental daily forecast of country source attribution from April 2018 onward
(harmonized city area definition, new CAMS emissions, composition output, more cities, 2 models)
Visualization of uncertainty in graphs
• Inclusion of NRT observations in daily forecast charts (based on background and city stations)Cooperation with TFMM twin site study
THANKS for the attention