Off-line model integration:
EU practices, interfaces and possible strategies for harmonisation
Barbara Fay, Deutscher Wetterdienst, Offenbach, Germany Sandro Finardi, ARIANET, Milano, Italy
Copenhagen, COST728/NetFAM workshop 1 21-23 May 2007 'Integrated Transport Models' Outline Part I
1. overview: European model systems main systems, applications
2. coupling issues input data, downscaling/nesting, modularity
3. model harmonisation – integrated model systems
Copenhagen, COST728/NetFAM workshop 2 21-23 May 2007 'Integrated Transport Models' COST728 Working Group 2: Integrated systems of MetM and CTM/ADM: strategy, interfaces and module unification
2.1 Coupling of off-line models (Barbara Fay, DWD, Germany) D2.1 Overview of existing integrated (off-line and on-line) mesoscale systems.
Copenhagen, COST728/NetFAM workshop 3 21-23 May 2007 'Integrated Transport Models' 1. Overview of existing integrated (off-line and on-line) mesoscale systems • a tentative state-of-the-art inventory on existing methodologies, approaches, models and practice in different countries for building the integrated (off-line and on-line) MetM and CTM mesoscale systems. •COST728/732 Model Inventory (http://www.mi.uni- hamburg.de/index.php?id=539) • in Europe + WRF + US EPA + York University, Canada • sampled from COST728 members through questionnaire Æ as provided, patchy, only active COST countries • edited, but no rankings, represents knowledge and opinions of individual authors Meteorology Chemistry & transport Met & chem. & transport
Mi MITRAS AERMOD AERMOD, AERMOD_Urban Cro- MICTM MICTM Chensi Scale M-SYS MERCURE MetMs COST Meso-NH MIMO 732 RCG VADIS and Meso- ALADIN, ALADIN/A, ADMS-Urban AERMOD BOLCHEM Scale ALADIN/PL ALADIN-CAMx AURORA CALMET/CALPUFF CTMs ARPS CALGRID CAMx CALMET/CAMx COST CALGRID, CLM CHIMERE, CHIMERE (ARPA) ENVIRO-HIRLAM from 728 FVM, GEM/LAM CMAQ, CMAQ(GKSS) GEM/LAM-AQ GESIMA, GME EMEP M-SYS COST GRAMM, Hirlam, HRM ENVIRO-HIRLAM EPISODE MC2-AQ COSMO Lokalmodell aLMo, FARM COSMO-LM-ART 728 LAMI, LME, LMK FLEXPART, FLEXPART/A COSMO-LM-MUSCAT COSMO Lokalmodell LME_MH FVM MCCM model LAPS, MC2 GEM/LAM-AQ GEOS-CHEM MEMO (UoT-GR) MEMO (UoA-GR), ( UoA-PT) GRAMM HYSPLIT MERCURE inven- MERCURE, MESO_NH LOTOS-EUROS LPDM Meso-NH METRAS MARS (UoT-GR), (UoA-PT) RCG tory MINERVE MM5 (UoA-GR) MATCH MECTM TAPM MM5 (UoA-PT), (UoH- MOCAGE MUSE UK),(GKSS) NAME III OFIS (WG4) NHHIRLAM, RAMS RCG SILAM SAIMM, TAMOS SPRAY 3 TAMOS TRAMPER , UM TCAM TREX WRF_ARW UAM-V Macro- GEM GMElam CAM-CHEM ENVIRO-HIRLAM scale HIRLAM TAMOS CHIMERE, CHIMERE (ARPA-IT) GEM_AQ EMEP FLEXPART, FLEXPART/A GEOS-Chem IMPACT LPDM MATCH MOCAGE NAME III 5 SILAM STOCHEM TAMOS TCAM Model systems covered in - 16 countries / 38 institutions / > 25 systems
Main model systems MetMs:
MM5 COSMO HIR- ECM ALA GME UM RAMS CAL- ECHAM NCEP WRF LM LAM WF -DIN DWD MET
11 9 6 6 3 3 3 3 3 2 2 1
CAMx Chimere CAL- CMAQ Match MEMO CTMs: GRID 5 4 3 3 2 2
Copenhagen, COST728/NetFAM workshop 6 21-23 May 2007 'Integrated Transport Models' Model applications - Diagnostic / climatologic Transport, AQ assessment + impact scenarios, episodes, source apportionment - Forecasting: transport + chemistry, AQ (UAQ, coastal and industrial AQ), management: gases incl. ozone, PM, pollen grains - radioactivity (and environment) emergency forecasting
Copenhagen, COST728/NetFAM workshop 7 21-23 May 2007 'Integrated Transport Models' On-line coupled MetM - CTMs BOLCHEM (CNR/ISAC, Bologna) ENVIRO-HIRLAM (DMI) COSMO LM_ART (FZ Karlsruhe) COSMO LM-MUSCAT (IfT Leipzig) MCCM (Inst.Environm. Atmos.Research FZ Karlsruhe, Garmisch- Partenkirchen, Germany) MESSy: ECHAM5(and planned: LM) (MPI-C Mainz , Uni Bonn) MC2-AQ (York Univ, Toronto, Warsaw Univ.) GEM/LAM-AQ (York Univ, Toronto, Warsaw Univ.) OPANA = MEMO +CBM IV + SMVGear (Univ. Madrid) ECMWF (passive prognostic stratos. ozone tracer, GEMS chenistry) GME (DWD, passive prognostic stratos. ozone tracer)
Copenhagen, COST728/NetFAM workshop 8 21-23 May 2007 'Integrated Transport Models' On-line coupled models
Model name On-line coupled chemistry Time step for coupling Feedback BOLCHEM Ozone as prognostic chemically None active tracer ENVIRO-HIRLAM Gas phase, aerosol and Each HIRLAM time Yes heterogeneous chemistry step WRF RADM+Carbon Bond, Each model time step Yes Madronich+Fast-J photolysis, modal+sectional aerosol COSMO LM-ART Gas phase chem (58 variables), each LM time step Yes aerosol physics (102 variables), pollen grains COSMO LM-MUSCAT Several gas phase mechanisms, Each time step or time Yes aerosol physics step multiple MCCM RADM and RACM,photolysis Each model time step (Yes) (Madronich), modal aerosol MESSy: ECHAM5 Gases and aerosols Yes MESSy: ECHAM5-COSMO Gases and aerosols Yes LM (planned) MC2-AQ Gas phase: 47 species, 98 chemical each model time step None reactions and 16 photolysis reactions GEM/LAM-AQ Gas phase, aerosol and Set up by user – in most None heterogeneous chemistry cases every time step GME Progn. stratos ozone passive tracer ECMWF model (IFS) Prog. stratos O3, GEMS chemistry Each model time step Yes OPANA=MEMO+CBMIV 2. Coupling issues and problems
2.1 Input data • measurements -> pre-processors: point measurements to 3D Only diagnostic,possible advantage: no divergence from obs • models input format GRIB (majority?) netCDF coupling time step 15 min: COSMO LMK 2.8km, UM (4km) 1 h (majority) 3h (6h) Copenhagen, COST728/NetFAM workshop 10 21-23 May 2007 'Integrated Transport Models' 2.2 Downscaling / nesting
• requirements: higher resolution - improved vertical turbulence - suitable heterogeneous surface characteristics, fluxes - subgrid-scale orography, - urban structures, - local emission sources + transformations - local circulations (sea breeze, mountain-valley winds) - improved chemical boundary conditions • achieved through (self-) nesting of MetMs and CTMs 2-way interactive nesting for MetMs MM5, RAMS, (COSMO-LM)
Copenhagen, COST728/NetFAM workshop 11 21-23 May 2007 'Integrated Transport Models' 2.3 Modularity
Requirements: • high modularity • high compatibility, e.g. no COMMON blocks but direct parameter passing • flexible IO strategies
2.4 Interfaces -> part II, Sandro
Copenhagen, COST728/NetFAM workshop 12 21-23 May 2007 'Integrated Transport Models' 3. Model harmonisation - integrated model systems
• COST710 (1994-1998): Harmonisation in the pre-processing of met data for dispersion modelling - including harmonisation questionnaire • COST715 (1998-2004): Met applied to urban air pollution problems
• regular conferences on Harmonisation for regulatory modelling • harmonised local-scale model evaluation: - Model Evaluation Kit - American Standard Evaluation Tool COST728, WG4: harmonised evaluation
Copenhagen, COST728/NetFAM workshop 13 21-23 May 2007 'Integrated Transport Models' 3. Integrated model systems • PRISM (FP5): Earth system modelling, software infrastructure • PRISM support: COSMOS, OASIS coupler • ENSEMBLES (FP6): climate change ensemble prediction system for Earth system models • ENES European system for Earth system modelling including PRISM, ENSEMBLES … • ESMF Earth system modeling framework (US) • FLUME flexible Unified Model Eenvironment (UK MetOffice) • CURATOR info on earth system/climate models, intercomparison projects and IPCC assessments (US) • GEMS global+reginal Earth system monitoring using satellites and in-situ data, ECMWF, data assimil. and forecasting • GENIE Grid ENabled integrated Earth system model (UK) • GO-ESSP global orga of Earth system science portal (UK,NOAA,NASA...):access to obs and simulated data Off-line model integration: EU practices, interfaces and possible strategies for harmonisation
Part II
Interfaces between meteorological and air quality models
Sandro Finardi
Copenhagen, 21-23 May 2007 Why should we care about interfaces ?
In principle, interfaces should be simple pieces of software connecting Met. Models outputs and AQ Models inputs with minimum influence on results, but:
• Often MetM and AQM have been built separately and are applied on different grids at different space scales
• Often MetM cannot provide all the variables required by AQM or some fields can be estimated by parameterisations and algorithms not compatible with modelling methods implemented in AQM
• Some AQ models need to rely on “standard” meteorological products, e.g. average met. variables while turbulence, atmospheric stability, mixing height, and dispersion coefficients are diagnostically estimated.
• Sometimes re-computation or “filtering” of dispersion parameters can guarantee AQ modelling robustness for practical applications.
The communication between off-line coupled meteorological and AQ models is a problem of often underestimated importance: • Interfaces can limit the possibility of AQ models to access and exploit all the information provided by new generation mesoscale meteorological models, and can make difficult model intercomparison, • Interfaces can be used to improve boundary layer low description AQ Modelling System Conceptual Scheme
Geographic & Physiographic Data
Larger Scale Forecast Meteorological Model & Meteo Data Emission Inventories Interface Module
IC/BC Larger Scale Emission Air Quality Model Fields Pre-Processor & AQ Data Concentration Fields
Population Exposure Common interface tasks :
• Interpolation of Met. data to match grid differences (geographic projections, vertical grid system, resolution)
• downscaling of meteorological data
• Estimation of boundary layer scaling parameters, mixing height and eddy diffusivities (or σi and TL for Lagrangian models)
• Emissions related computations (e.g. biogenic VOC emission, wind blown dust).
• Enhancement of physiographic data and parameterisations (e.g. urbanisation)
• Air quality fields and data treatment for AQ models IC/BC
COST728/NetFAM workshop 'Integrated Transport Models' COST728/732 Inventory - Coupled Models involved:
MetMs:
EU Met. Services: LM, ALADIN, HIRLAM, UM
Other EU: METRAS, MEMO
US Community: MM5, RAMS, ARPS, WRF
CTM/AQMs:
EU Met. Services: LPDM, TRAJEK, NAME III, MATCH, SILAM, DERMA, ARGOS
Other EU: TCAM, FARM, SPRAY3, CHIMERE, AirQUIS, FLEXPART, AURORA, EURAD, MECTM, MARS, MATCH, AUSTAL, MISKAM, HYSPLIT, SCIPUFF
US Community: CAMx, UAM-V, CMAQ, REM-CALGRID, CALPUFF
COST728/NetFAM workshop 'Integrated Transport Models' Three main practises :
• Development of integrated systems (mainly Met services) – Interfaces built on specific model features and needs
• Use/Customisation of US Community modelling systems – MM5+CAMx, MM5+UAM-V, MM5/WRF+CMAQ – Customisation of available interfaces (e.g. MCIP)
• Interfacing of self developed AQ models with EU Met Services and US Community Met. Models – Development of model specific or general purpose interfaces
COST728/NetFAM workshop 'Integrated Transport Models' Different choices possible effects :
Some interface choices can have relevant effects on the AQ simulation results and mask actual meteorological and air quality model results, examples:
• Minimum KZ value effect
• Reconstruction or direct use of modelled surface fluxes
• Air quality initial conditions effects
COST728/NetFAM workshop 'Integrated Transport Models' Examples from AQ simulations in Torino and Roma
Meteorological driver: • Roma: RAMS • Torino: RAMS, LAMI
Air Quality Model: • FARM (Chemical Transport Model)
Interface Module: • GAP/SurfPRO: fields interpolation and meteo processing
Finardi et al., 2007, Env. Mod. and Sof. Gariazzo et al., 2007, Atm. Env.1h Minimum KZ effect: O3
VillaAda Ada: (RM/U) urban
120
100
80
60 [ppb] 40
20
0 20/06/2005 21/06/2005 22/06/2005 23/06/2005 24/06/2005 Cavaliere:Cavaliere (RM/R) rural
120
100
80
60 [ppb] 40
20
0 20/06/2005 21/06/2005 22/06/2005 23/06/2005 24/06/2005
Computed (Roma, dust, Kzurb=0.1) Computed (Roma, dust) Measured Minimum KZ effect: NO2
VillaAda Ada: (RM/U) urban
120
100
80
60 [ppb] 40
20
0 20/06/05 21/06/05 22/06/05 23/06/05 24/06/05 Cavaliere:Cavaliere (RM/R)rural
120
100
80
60 [ppb] 40
20
0 20/06/05 21/06/05 22/06/05 23/06/05 24/06/05
Computed (Roma, dust, Kzurb=0.1) Computed (Roma, dust) Measured Minimum KZ effect: other experiences
Time SeriesPlot of Plot Simulated of Simulate and dObse andrved Observed Surface Surface Ozone Ozone Concentrations Concentrations in Columbus, in Columbus, GA, GA, the MinimumUsing Vertical a minimum Eddy vertical Difusivity eddy of diffusivity10-4m2/s was of 1.0m2/s.used in CMAQ.
Ted Russell, 2003, “Air Pollutant Transport, Control and Modeling Issues in the Eastern United States”, Georgia Inst. of Technology Direct use or reconstruction of surface fluxes
Torino – Lingotto: surf. fluxes and disperion parameters
350 1.0 300 0.9 H 0.8 u 250 0 * 0.7 200 0.6 SurfPRO SurfPRO 150 0.5 MINT MINT 100 (m/s) u* 0.4 0.3 50 0.2 sens. heat flux (W/m2) flux heat sens. 0 0.1 -50 0.0 0 6 12 18 0 6 12 18 0 6 12 18 0 6 12 18 0 6 12 18 0 6 12 18 19-21/07/1999 19-21/07/1999
2000 60 1800 50 1600 zi Kz 1400 40 1200 SURF SurfPRO 1000 30 P2F MINT 800 Kz (m2/s) 20 600 mixing (m) height 400 10 200 0 0 048121620048121620048121620 0 6 12 18 0 6 12 18 0 6 12 18 19-21/07/1999 19-21/07/1999 Direct use or reconstruction of surface fluxes
Torino – Lingotto: effects on concentrations
Mint SurfPRO OBS
350 300 NO 250 2 200 150 MG/MQ 100 50 0 0.00 12.00 0.00 12.00 0.00 12.00 19-20-21/07/1999 Mint SurfPRO OBS 300
250 O3 200
150 MG/MQ 100
50
0 0.00 12.00 0.00 12.00 0.00 12.00 19-20-21/07/1999 Direct use or reconstruction of surface fluxes
Cause: mountain foot thunderstorms
08:00 17:00
COST728/NetFAM workshop 'Integrated Transport Models' Nesting effects through IC/BC:
TO - Consolata: PM10, daily average, +48+24 225225 225 200200 200 175175 175 150150 150 125125 125 g/m3 g/m3
μμ 100100 100 7575 75 5050 50 2525 25 00 0 1/61/6 1/7 1/7 31/7 31/7 30/8 30/8 29/9 29/10 28/11 28/12 27/1 computed observed TOTO -- Gaidano:Gaidano: PM10,PM10, dailydaily average,average, +48+24 225 225 200 200 175 175 150 150 125 125 g/m3 g/m3 μ
μ 100 100 75 75 50 50 25 25 0 0 1/61/6 21/61/7 11/731/ 31/77 20/830/8 9/929/ 29/99 2 19/109/10 8/1128/1 28/111 2 18/128/12 7/127/ 27/11 computed observed Nesting effects through IC/BC:
Torino is out of the Po valley for CHIMERE topography, causing initial concentration underestimation
CHIMERE - IC FARM +48
COST728/NetFAM workshop 'Integrated Transport Models' Discussion and Conclusions (1)
• A large variety of air quality modelling systems are developed and applied for research, operational forecast and air quality assessment over Europe. • The use of different meteorological drivers, air quality models and interface modules can be considered a scientific richness but creates problems of model result inter-comparison and make difficult stakeholders choices for practical applications. • Difficulties in model development collaboration in Europe are evident and can limit an effective exploitation of scientific advance. • Model harmonisation remains an important issue despite earlier efforts, e.g. COST710 (1994-1998) which are continued in the regular Harmonisation conferences and recent COST Actions (728, ES0602). • Development of community models can foster scientific cooperation, state-of-the-art knowledge dissemination and tools harmonisation (US experience), but it seems still hardly feasible in Europe.
COST728/NetFAM workshop 'Integrated Transport Models' Discussion and Conclusions (2)
Some basic steps to help harmonisation are desirable and feasible:
• Definition of agreed guidelines for off-line and on-line integrated modelling
• Modular modelling, flexible IO strategies to permit different model use and testing
• Definition of standards for the distribution of meteorological fields for air quality applications (weather forecast standards are not suitable)
• Continental scale air quality fields distribution harmonisation
• Definition of guidelines for interfaces development and application.
• Volunteer sharing of software implementing parameterisations for interfaces ?