Dispersion modeling
& EnFuser
Research manager/FMI
Ari Karppinen+ the team
11/2016
AQMg : the aims
1. Development and evaluation
of air quality models : from
microscale to global scale
2. Integration of meteorological
models (including climate) and
dispersion models
3. Efficient use of all available
measurement information
4. Application of models, and
dissemination of information:
!
Integrated use of models and data
Monitoring Models Satellite
Goal: operational system taking into account all
sources of information
ECMWFHIRLAM
RCR
HIRLAM
MBE
AROME,
LAPS
Weather prediction
models
Dispersion models -
long-range, regional
Dispersion and effects
models – urban, local
PALM
FINFLO,FLUENT(CFD),
DNS-code development
SILAM LRT,
meso-scale,
radioactivity , pollen
Modelling system - FMI
OSPM (NERI), street
canyon
UDM-FMI, urban
CAR-FMI, roadside
HILATAR
LRT, meso-scale
Aerosol process models:
UHMA (U Helsinki, FMI)
MONO32 (U Helsinki, Stadia)
SALSA(UH,UKU,FMI)
SILAM-APMs
MPP-FMI, Meteorological
pre-processing model
ESCAPE, chemical
accidents
BUOYANT, fires
EXPAND (FMI, YTV)
population exposure
Jalkanen, J.-P., L. Johansson, and J. Kukkonen, 2016. A comprehensive inventory
of ship traffic exhaust emissions in the European sea areas in 2011. Atmos. Chem.
Phys., 16, 71–84, 2016. http://www.atmos-chem-phys.net/16/71/2016/acp-16-71-
2016.pdf.
Modelled geographical distribution of CO2 emissions
from global shipping in 2015.
A new detailed dataset of the emissions of shipping globally
AROME
NWP modelHIRLAM-MBE
NWP model
Physiography,
forest mapping
Aerobiological
observations
SILAM now
Satellite
observations
Phenological
observations SILAM
CTM model
EVALUATION:
NRT model-measurement
comparison
Aerobiological
observations
Meteorological
data: ECMWF
Online AQ
monitoring
Phenological
models
Fire Assimilation
System
HIRLAM-RCR
NWP model
AQ
products
CLRTAP/EMEP
emission data
http://silam.fmi.fi
Satellite
observations
Globalboundary cond.
+own simulations
WRF
TVM
SILAM v.5: modules and capabilities
• Modules
• 8 chemical and physical transformation modules (6 open for operational use),
• 6 source terms (all open),
• 2 aerosol dynamics (one open)
• 3D- and 4D- Var
• Domains: from global to beta-meso scale (~1km resolution)
• Meteo input:
• ECMWF
• HIRLAM, AROME, HIRHAM, ECHAM, and any other who can write GRIB-1 or GRIB-2
• WRF
SOx
Acid-basic
CB4
Pollen
General PM
Radioactive
Passive
self-decayLong-lived
multi-media
Transformations
Area
Point
Nuclear bomb
Source types
Map of
species
masses
Em
iss
ion
Tra
ns
form
ati
on
Dynamics
Advection
diffusion
Aerosol
dynamics
Bio-VOC
Pollen
Sea salt
Simple
Basic
Transformation
Dry
Wet
Deposition
Initialization,
3D-Var adjust.
Control
Type: forward, adjoint, 4D-Var
European AQ forecast ( SO2, NO, NO2, CO, O3, PM10, PM2.5)
see also : http://silam.fmi.fi/
CAMS: Copernicus Atmosphere Monitoring service
( https://atmosphere.copernicus.eu/ )
+ Clearly largest
forecasting ensemble up
to date, for main gaseous
and PM pollutants
+ A concerted effort with
a better overall reliability
and versalitility
- Can still be improved:
mass closure of PM, non-
antropogenic PM
- Structure and
treatments of models are
variable (e.g. data
assimilation, evaluation)
Example: Global forecasts of aerosol concentrations - sulphates
Global AQ forecast ( SO2, NO, NO2, CO, O3, PM10, PM2.5)
see also : http://silam.fmi.fi/
Forest fires,
volcanoes,
etc…
29/11/2016 13
The other “extreme”:
Nested Multi-Scale System in the PALM
Large-Eddy Simulation Model
Practical example:
Helsinki City
Urban planning
29/11/2016 14
The Domain: 4 * 2 & 2*1 km
Resolution: 1m ; 0.1 s
29/11/2016 15
29/11/2016 16
29/11/2016 17
The options:
29/11/2016 18
Nested LES-modelling proved to be
a useful tool for City planning
Fusion of meteorological and air
quality information
Idea: to combine ALL available
information (models, measurements,
land use, traffic, population data..)
to achieve the ”optimal” view of the
state of environment
Background(we already have everything we need..)
• Availability of open access (environmental) data has improved in the past few years rapidly (see e.g . https://www.europeandataportal.eu/en)
• real time air quality measurements
• air quality modelling results (nowcasting/forcasting)
• meteorological measurements & model data
• land use data
• Detailed 3D data on urban areas
• OpenStreet maps
• BUT: e.g. in India & China, the situation could bemuch better, especially with the openess of data
29.11.2016 21
Background(..to make something useful out of it )
• we want to give a detailed & accurate view of current AirQuality esepcially at urban (=populated) areas
• and..
• we also want to forecast airquality (1-2 days ahead)
• Taking into account the huge increase in availabilityof crucial/useful data , we should be able to thismuch better than we used to do with ”traditional” 1st/2nd or even 3rd generation modelling methods..
• but.. HOW to do it ???
29.11.2016 22
2323
Regional AQ Forecasts
Emission
Inventories
(if available)
Meteorological DataAQ measurements
GIS-
Datasets
FMI-ENFUSER
• Information fusion using open access
and globally available datasets
• Hourly average NO2, PM2.5, PM10 and
O3
• Adapts to recent measured AQ data
• Can be calibrated and used without local
emission inventories
Operational modelling system
24
FMI-ENFUSER(runs every 1-6h)
Information
extraction(every 10 minutes)
Calibration
update(1-4 times per year)
Model output(figures, animations,
gridded data, etc)
Web Portal
For selected cities in Finland FMI-ENFUSER produces high resolution
(15x15 m) AQ information
- Includes AQ forecasts for the next 12-48 hours
- Model data accessible through a Web Portal
NO2, PM2.5, PM10, O3, AQI
Amazon Web Service, HerokuApp, Dropbox
25
The EnFuser (= our solution)
FMI-ENFUSER = (The Finnish Meteorological Institute’s ENvironmental
information FUsion SERvice)
Developed in projects : EU/PESCaDO, CLEEN MMEA, TEKES-INKA*, TEKES-CITYZER*, CLIMOB*
• Combination of statistical
modeling(LUR) and dispersion
modeling
• Dispersion modeling without
emission data (~inverse modeling)
• On the other hand, statistical
modeling using physically
realistic weighting factors based
on dispersion modelling
• model calibrated based on
measurements
• E.g. 100 000+ observation
(Helsinki area)Real meteorology taken into account in the statistical
modeling
26
Land use data
• OpenStreetMap (OSM)
• Land use, street network and buildings
• 5 x 5m, open global data
• (<= similar work done in SP!)
• Building are important for many reasons:
• E.g. proxy for PM2.5 source
• Combining emission surveys and OSM data
we can create a detailed small scale wood
Combustion emission mapping for
Helsinki area
• Turbulence /winds / street canyons
• The OSM-makes it possible to identify
the street canyon environments
-> realistic meteorological scenarios
29.11.2016 27
Most recent
measurments
Regional scale
concentrations
(SILAM)
1800 x 1200 gridpoints
– all the points have
gone through the
adjustment cycle
• Meteorological data (HIRLAM/HARMONIE) fromFMI’s Open Data portal
• SILAM regional AQ forecasts
• Long range transport of pollutants and boundaryconditions for ENFUSER
• Globally available
• Important for forecasting
• STEAM shipping emissions
• Global emission inventory
• Important for coastal locations
Synergy with other FMI models
29.11.2016 28
29
Approach Benefits
Due to calibration, information on emission sources not
absolutely necessary
- Known emission inventories can still be included
Latest sensor measurements & modelled data can be included
in the pool of information- Individual sensor quality (and measurement height) taken into account
- Weather forecast + regional background forecast => AQ forecasting possible
Example emission inventory: shipping
emissions (Rotterdam, FMI-STEAM)
can be included as ’puff emissions’
minute by minute.
Industrial emissions can also be
included with the same methodology.
30
Built to estimate
seasonal averages
based on Land-
Use-Regression
Built to estimate
hourly concentration
in HMA based on
Dispersion
modelling
Built to estimate
hourly concentration
in Finland, based on
Dispersion and LUR
modelling with
fusion
Performance evaluation
Root mean squared error for estimated seasonal NO2 concentrations
Online /real-time evaluaton
29.11.2016 31
Integrated solution offered together with a sensor network
• High resolution assessment of air quality in an area ratherthan a collection of point observations
Tool for administrative users
• Warning system and pre-emptive decision making
Indoor air quality management
• Optimization of air intake control using ENFUSER’s forecastsand 3D air quality data
Mobile applications with GPS tracking
• Assessment of personal pollutant exposure on a minute-by-minute basis
High resolution annual pollutant concentration maps for cities
• Assessment of annual average air quality right outside of yourhome
Planned practical applicatios
29.11.2016 32
A demonstration in Langfang, China, utilizing 5 installed PM2.5 sensors.
Continues with more sensors (3x) and regional AQ model data being included.
Also, EnFuser will be demonstrated(2016 ) in Delhi, India as a part of CLIMOB project.
33
Demonstration in Asia
Figure: A closer look at the hourly PM2.5 concentration in Langfang,
given by FMI-ENFUSER visualization toolbox.
An example of estimated PM2.5 concentration time
series based on the sensor data in Langfang. The
selected example location is just outside of Silver City
Hotel, Langfang
• Johansson, L., Epitropou, V., Karatzas, K., Karppinen, K., Wanner, L., Vrochidis, S., Bassoukos, A., Kukkonen, J. and Kompatsiaris I. Fusion of meteorological and air quality data extracted from the web for personalized environmental information services. Environmental Modelling & Software, Elsevier, Volume 64, February 2015, Pages 143–155, 2014
Conclusions
29.11.2016 34
We are ready to…
utilize the data form
dense urban
measurement São
Paulo!
for Real time , high-
resolution AQ-mapping
+ forecasts
Huge potential
• The methodology would bridge the gap between modeling and the measurements especially in difficult environments like megacities (= like São Paulo)
• Basic requirements :
• dense measurement
network : good coverage
of all relevant environments
• Supporting information
• available:
• land use, traffic,
• population density
Thank You for Your
attention !