Overview of COSMO Activities at Empa Stephan Henne, Dominik Brunner, Gerrit Kuhlmann, Gianluca Mussetti

Size: px
Start display at page:

Download "Overview of COSMO Activities at Empa Stephan Henne, Dominik Brunner, Gerrit Kuhlmann, Gianluca Mussetti"

Transcription

1 Overview of COSMO Activities at Empa Stephan Henne, Dominik Brunner, Gerrit Kuhlmann, Gianluca Mussetti Atmospheric Modelling and Remote Sensing Group Laboratory for Air Pollution/Environmental Technology, Empa, Dübendorf

2 Research Questions and Tools Quantification of greenhouse gas emissions (national to continental scale) Simulating effect of specific sources on air pollution levels Investigating air quality and greenhouse gases using remote sensing technologies Urban climate and air quality COSMO GHG (Tracer) Passive tracer transport Applied to CO 2, NO 2 FLEXPART-COSMO Passive tracer transport Source sensitivities Inverse modelling of surface fluxes (CH 4, N 2 O, halocarbons) COSMO urban Urban parameterisations Urban heat island Benefit of urban vegetation COSMO ART (ICON ART?) Air quality Switzerland BAFU project starting spring 2018

3 COSMO-GHG ESA Project SMARTCARB Modelling of CO 2, CO and NO 2 plumes from cities and power plants for designing a future CO 2 satellite Presentation Dominik SDSC Project Carbosense4D Starting 2018 Combining COSMO-GHG simulations and data from dense CO 2 sensor network in Switzerland Provide accurate, highres 4D CO 2 field for satellite validation Column mean dry air mole fraction of CO 2 (XCO2) and cloud cover on 3 July UTC

4 FLEXPART-COSMO Source Sensitivities and Flux Inversion Bermomünster tall tower Inlet at 212 m Source sensitivities for Swiss tall tower site Beromünster Driven by COSMO-7 analysis fields Annual average footprint Bayesian flux inversion: minimisation of cost function J = 1 2 x x b T B 1 x x b Mx χ o T R 1 Mx χ o

5 BAFU Project: Swiss CH 4 Emissions (Gg/yr) (Gg/yr) (Gg/yr) NIR 2017 Inversion Mean a-posteriori difference from 8 sensitivity runs each Henne et al., ACP, (2016)

6 FLEXART-COSMO at High Resolution Aim Optimisation of FLEXPART-COSMO for COSMO output with high resolution (COSMO-1, COSMO-2) Status FLEXPART too diffusive for convection resolving scale (<2 km) Suspected reason: use of instantaneous wind input Developments Test suite for FLEXPART simulations to test transport consistency Forward and backward simulations with the same source have to yield same receptor concentrations COSMO-2 runs with output of hourly averaged wind fields Re-evaluation of transport implementation in FLEXPART-COSMO

7 Receptor concentration (backwards vs forwards): Default Configuration FLEXPART-ECMWF FLEXPART-COSMO COSMO-7 FLEXPART-COSMO COSMO-2 Almost perfect agreement for FLEXPART-ECMWF (comparable with Seibert und Frank, 2004) FLEXPART-COSMO Backward runs show slightly reduced concentrations COSMO-2: generally reduced concentrations

8 Receptor concentration (backwards vs forwards): Different COSMO-2 Inputs FLEXPART-COSMO COSMO-2 FLEXPART-COSMO COSMO-2-free FLEXPART-COSMO COSMO-2-avg Free run shows very similar concentrations Similar negative deviation for backward runs Use of average wind fields Similarly small concentrations and negative deviation for backward runs

9 Receptor concentration (backwards vs forwards): Updated Transport Description FLEXPART-COSMO COSMO-7 FLEXPART-COSMO COSMO-2 FLEXPART-COSMO COSMO-2-avg Updates Calculation of relative vertical wind speed consistent with COSMO Inaccuracy in particle transport (update of local topography) Almost perfect agreement for all configurations COSMO-2 concentrations remain factor of 2 lower!

10 SNF DACH: N 2 O Sources from Grasslands ScaleX 2016 Field Campaign June July 2016 Southern Bavaria Intensively used grassland Treatment: fertilization using slurry Micro meteorological measurements Sonics, lidar, sodar KIT: Flux chamber measurements of N 2 O Empa: Online measurements of N 2 O isotopes in air PhD of Erkan Ibraim

11 FLEXPART-COSMO Site Scale Simulations FLEXPART-COSMO optimised for micrometeorological applications Direct use of measured u, v, u *, L M in near-field simulation Simulated N 2 O concentration Using N 2 O flux from chamber measurements for whole grassland

12 Empa Project: Urban Climate Simulations PhD thesis Gianluca Mussetti shared with Laboratory for Multiscale Studies in Building Physics Correctly represent effect of urban heat island in mesoscale (climate) model What resolution of the mesoscale model is required to resolve the intra urban variability? What is the effect of urban vegetation? Mesoscale Model Mesoscale modelling + Urban Canopy Model Large scale resolved Low computational cost Urban Canopy Model Able to model an entire city Double Canyon Effect Parameterization (COSMO-DCEP, Schubert et al., 2012) Multi-layer model, multiple canyon directions No street vegetation

13 Heatwave July 2015, Zurich: Impact on Air Temperature (2 m) Average early morning temperature 1000 m 500 m 250 m

14 Urban Vegetation COSMO-DCEP/BEP-TREE BEP-TREE (column model) Krayenhoff et al. 2014, 2015 Considering canyon and above building vegetation Radiation budget solved explicitly by raytracing method Drag and transpiration of vegetation parameterised Implementation into COSMO-DCEP ongoing Krayenhoff et al., 2014

15 Summary: Model Empa FLEXPART-COSMO Improved vertical transport description Modification for site scale application (nudging to mircomet observations) Implementation of NetCDF input of COSMO fields COSMO-GHG GHG module fully implemented in COSMO POMPA (GPU) Updates to COSMO and int2lm for more flexible definition of tracers Online emission module (ongoing) COSMO-Urban Incorporation of BEP-TREE urban canyon parameterisation with vegetation into COSMO-CLM- DCEP (ongoing)

16 Backup Slides

17 From Air Parcels to Mole Fractions S x Sensitivities (FLEXPART) Emissions (Inventory) Contributions by: Emission uptake Mole fraction at particle "origin" χ = + 1 K i,j k χ b m i,j χ t χ k E i,j + Base mole fraction

18 Atmosphärische CH 4 Beobachtungen Schauinsland (1200 m) Gäbris (1250 m) Mai 2016 Nov 2016 Jungfraujoch bis

19 Seasonality of CH 4 Emissions DJF MAM JJA SON Mean a-posteriori difference (all years, all sensitivity runs)

20 Concept for validation of Lagrangian Particle Models Forward and backward simulations with the same source have to yield similar receptor concentrations receptors source

21 Tested Windfields und Model Versions Windfelder ECMWF: IFS analysis (0.2 x 0.2 ) together with standard ECMWF version of FLEXPART. COSMO-7: analysis MeteoSchweiz containing instantaneous wind fields (previously our default setup for inverse modelling studies on Swiss scale). COSMO-2: analysis MeteoSchweiz containing instantaneous wind fields. COSMO-2-free: free COSMO run without data assimilation but otherwise same settings as COSMO-2 analysis. Output of instantaneous wind fields. COSMO-2-avg: Like COSMO2_free but with output of hourly averaged wind fields. Modelle FLEXPART-ECMWF FLEXPART-COSMO FLEXPART-COSMO (modified)

22 Windfelder + Grenzschichthöhen: Beispiel :00 UTC