NMMB Tutorial, NCEP, Maryland, April 1st, 2015

Carlos Pérez García-PandoColumbia University

NASA Goddard Institute for Space Studies

(On behalf of Oriol Jorba from Barcelona Supercomputing Center, Spain)

Contributors: Z. Janjic, T. Black, R. Vasic (NCEP), S. Basart, A. Badia, M. Spada, J.M. Baldasano, E. DiTomaso, G.S. Markomanolis, K. Serradell,

A. Folch, A. Martí, M. Gonçalves (BSC), D. Dabdub (UCI), R. Miller, K. Tsigaridis (NASA GISS), J. Soares (FMI)

The NMMB/BSC-CTM model: description and results

• Multiscale: regional to global scales

• On-line coupled aerosols and chemistry allowing consistency and feedbacks

NMMB/BSC-CTM

Nonhydrostatic Multiscale Model on the B-grid (NMMB)meteo variables/parameters

→ Janjic and Gall (NCAR/TN 2012)→ Janjic and Vasic (EGU2012)→ Janjic et al. (MWR 2011)→ (...)

BSC Chemical Transport Model(gas/aerosol variables: mass mixing ratios)

NMMB/BSC-Chemical Transport Model (Overview)

GAS-PHASECHEMISTRY

→ Jorba et al. (JGR 2012)→ Badia and Jorba (AE 2014)→ Badia et al. (GMD 2015 in prep)

→ Pérez et al. (ACP 2011)→ Haustein et al. (ACP 2012)

→ Spada et al. (ACP 2013)→ Spada et al. (AE 2014)

→ Spada et al. (GMD 2015 in prep)

AEROSOLS

Aerosols I• Mass-based approach. Dry sizes remain constant throughout the model simulation

• The wet size of hygroscopic particles changes with relative humidity for radiation and sedimentation calculations

• Dust: 8 size bins emitted as a function of friction velocity through a physically based dust emission scheme. Tested at global and regional scales (Northern Africa, Middle East and Europe)

• Sea-salt: 8 size bins emitted as a function of 10-m wind speed. 5 emission schemes available and thoroughly evaluated

• Black Carbon (BC): 2 tracers, one hydrophobic and one hydrophilic. Hydrophobic BC converted to hydrophilic with an e-folding time of 1.2 days (as in GOCART)

• Organic Matter (OM):

Primary Organic Matter (POA): 2 tracers, one hydrophobic and one hydrophilic. Hydrophobic OM converted to hydrophilic with an e-folding time of 1.2 days (as in GOCART)

Secondary organic aerosols (SOA): 4 gaseous and 4 aerosol-phase tracers. SOA produced by the reversible partitioning of the semi-volatile gaseous O3 oxidation products of isoprene and terpenes (Tsigaridis and Kanakidou, 2003; 2007). Anthropogenic SOA produced from toluene and xylene is under development

4

Aerosols II• Sulfate (SO4): 1 hydrophilic tracer, 4 additional prognostic tracers (SO2, DMS, H2O2, H2SO4)

and 3 online or climatological oxidants (OH, O3, HO2). Includes gas-phase oxidation of SO2, DMS and H202 by OH, and aqueous-phase oxidation by H202 and O3 (Sander et al., 2006, 2011)

• Nitrate (NO3) and Ammonium (NH4): as calculated by EQSAM thermodynamic equilibrium model but not tested yet (Metztger et al. 2002)

• Dry deposition from the bottom layer includes aerodynamic and surface resistance (Zhang et al., 2001)

• Gravitational settling follows the Stokes approximation including the Cunningham correction factor accounting for reduced viscosity for small aerosols

• In-cloud and below cloud scavenging from grid-scale (Ferrier) and sub-grid scale (BMJ) clouds

• Below cloud scavenging follows Slinn (1984) (directional interception, inertial impaction and Brownian diffusion)

• Vertical convective mixing follows the BMJ adjustment scheme (instead of a mass flux scheme)

• Radiation: RRTM SW/LW aerosol radiative feedback

• Optical properties: GADS refractive indexes (special for dust), Mie-scattering of spheres (Mischenko, 2000)

Gas-phase chemistry• OH, O3, HO2: for aerosol calculations we can use online gas-phase simulations or off-line

climatologies

• Carbon-bond CBM-IV and CB05 mechanisms implemented (Gery et al., 1989; Yarwood, 2005)

• Coupled with Fast-J photolysis scheme (Wild et al., 2000)

• Mechanism implemented through KPP kinetic pre-procesor (Damian et al., 2002)

• KPP coupling allows a straightforward modification of chemistry kinetics and reactions. Suitable for sensitivity studies

• Implemented an EBI solver for CB05 as in CMAQ. Includes 51 chemical species and 156 reactions. Working version and thoroughly tested

• Stratospheric ozone: linear model Cariolle and Teyssèdre (2007) or Monge-Sanz et al. (2011)

• Dry deposition velocities depend on aerodynamic resistance, quasilaminar sublayer resistance and canopy or surface resistance (Wesely et al., 1986, 1989)

• Cloud chemistry: wet scavenging, mixing, and aqueous chemistry. Deposition follows Byun and Ching (1999) for grid-scale and subgrid-scale clouds

6

Module_Solver_Grid_Comp

Solver_initialize

USE module_bsc_initialize

call bsc_initialize

Solver_run

USE module_bsc_chemistry module bsc_dynamics_routine_chem module_bsc_deposition module_bsc_sedimentation

call hdiff adv2_chem mono_chem radiation sedimentation turblence

cucnv gsmdrive

chem_driver

NEMS/NMMB • Chemistry implemented on-line within NMMB routines• Inline aerosol dynamic emissions and wet depositions• On-line with modular implementation of chemistry and

gas depositions• Same advection, diffusion routines as NMMB

!*** AEROSOL - GAS TRACERS: LOCATIONS IN TRACER ARRAY!*** [-----------------------------TRACERS-------------------------------------------------]!*** [------MET-----][--WATER---][-------------------CHEM------------------------------]!!*** [---------------------------CHEM---------------------------------------------------------]!*** /ash bins/dust bins/salt bins/om bins/bc bins/so4 bins/no3 bins/nh4 bins/gas species/!***

Module_BSC_Chemistry

call emissions photolysis gas_dry_deposition gas_wet_deposition chemistry_mechanism aerosol_mechanism

USE module_bsc_...

Integration within NMMB

7

Configure file (not exhaustive)num_dust: 8 # Number of dust transport bins (0 or 8 works - 4 also planned)num_salt: 8 # Number of sea salt tranport bins (0 or 8 works - 4 also planned)num_om: 0 # Number of organic matter transport bins (0 or 6 works)num_bc: 0 # Number of black carbon transport bins (0 or 2 works) num_so4: 1 # Number of sulfate transport bins (0 or 1 works)num_no3: 0 # Number of nitrate transport bins (0 or 3 works)num_nh4: 0 # Number of ammonium transport bins (0 or 1 works)

bc_aero: false # Do we have aerosol BC for parent domain on regional runs (true or false)?

chem_mech: 0 # 0 no gas-phase chemistry # 1 CB-IV chemical mechanism from KPP # 2 CBM05 chemical mechanism from KPP # 22 CBM05 chemical mech. with EBI-solver from CMAQ # 6 SO4-SOA-GLOBAL MECHANISM # 7 SO4-SOA-NI-GLOBAL MECHANISM

diff_chem: true # Lateral diffusion for chem/aerosol speciesadv_chem: true # Horizontal and vertical advection for chem/aerosol species incloud_conv: true # Aerosol In-cloud scavenging from convective cloudsbcloud_conv: true # Aerosol below-cloud scavenging from convective cloudsincloud_strt: true # Aerosol in-cloud scavenging from stratiform cloudsbcloud_strt: true # Aerosol below-cloud scavenging from stratiform clouds conv_trans: true # Convective adjustment of aerosol arraysvdiff_chem: true # vertical diffusiondry_dep: true # Aerosol dry depositionnewsedim: true # new aerosol sedimentation nsedim: 4 # Adjustment steps between new sedimentation calls

8

Configure file (not exhaustive) For RRTM radiation

iaer: 22 # flag for aerosols scheme selection (all options work for N12B) # - 3-digit aerosol flag (volc,lw,sw) # = 0: turn all aeros effects off (sw,lw,volc) # = 1: use clim tropspheric aerosol for sw only # = 10: use clim tropspheric aerosol for lw only # = 11: use clim tropspheric aerosol for both sw and lw (def. NCEP) # =100: volc aerosol only for both sw and lw # =101: volc and clim trops aerosol for sw only # =110: volc and clim trops aerosol for lw only # =111: volc and clim trops aerosol for both sw and lw # = 2: gocart/BSC-Chem tropspheric aerosol for sw only # = 20: gocart/BSC-Chem tropspheric aerosol for lw only # = 22: gocart/BSC-Chem tropspheric aerosol for both sw and lw # =102: volc and gocart/BSC-Chem trops aerosol for sw only # =120: volc and gocart/BSC-Chem trops aerosol for lw only # =122: volc and gocart/BSC-Chem trops aerosol for both sw and lw

fhdust: 100000. # = 0 dust determined from gocart clim # > 0. and <99999. determined from nmmb fcst and gocart clim # >99999. determined from fcst

fhsalt: 0. fhom: 0. fhbc: 0.fhso4: 0. fhash: 0.

4 soil particle size populations and 8 dust transport particle sizes

GcFS

ii

tta su

u

u

uu

gG

2*

2*

*

*3* 11

Vertical flux

Horizontal fluxWhite (1979)

a vertical to horizontal flux ratioc tuning parametersi relative surface area of each soil particle fraction

Dust (emission)

68.0'** )(21.11 wwuu tdryt

ClayClayw %17.0%0014.0' 2 Soil moisture effectsFecan et al. (1999)

Threshold frictionvelocity of dry andsmooth soilIversen and White (1982)Marticorena and Bergametti (1995)

A landuse fraction (desert mask) [0-1]P preferential source probability [0-1]VF Vegetation fraction [0-1]

Source functionGinoux et al. (2001)

5

minmax

max

zz

zzS i

Pérez et al. (2011 ACP)

10

Dust (results)

Global (Year 2000, 1.4x1 deg), surface concentration

Pérez et al. (2011 ACP)

Dust (results)

Global (Year 2000, 1.4x1 deg), Aerosol Optical Depth

Pérez et al. (2011 ACP)

12

Dust (results)

Regional daily (Year 2006, 0.25x0.25 deg)

R ranges between0.6 and 0.8 withoutdata assimilation inNorthern Africa, MiddleEast and Europe)

AERONET

Pérez et al. (2011 ACP)

• Simulations between 2002 and 2006

• 5 online emission schemes implemented and tested (dependent on U10 and/or SST)

• Jaegle et al. (2011) best for surface concentrations globally (U. Miami stations) but tends to overestimate coarse AOD in tropical stations

Sea-salt (global) Spada et al. (2013 ACP)

• Stations that unaffected by local surf conditions may no be considered representative of open ocean conditions from a meteorological point of view

• Enhancing resolution 1 deg to 0.1 deg over

New Zeland: Sea-salt conc biases corrected in U. Miami stations surrounded by topography (factor of 2!)

Sea-salt (regional)Spada et al. (2015 AtmEnv)

• Anthropogenic and biomass burning emissions: ACCMIP (annual)• Fires' inj. height: IS4F (monthly) • Simulated years: 2002–2006 (monthly means eval.)

OA / BC / SO4Spada et al. (2015 GMD in prep)

Clear sky AOD

Monthly data from 240 AERONET stations between 2002 and 2006

Spada et al. (2015 GMD in prep)

Strong overestimation over fire regions

Good agreement over polluted areas.

Need to implement attenuation of radiation due to aerosols in photolysis scheme.

• Anthr. and BB emissions: ACCMIP• Biogenic emissions: MEGAN• No lightning emissions• 1 year spin-up• 2004 simulation Comparison with MOPITT (v5) at 800 hPa

Carbon Monoxide (CO)Badia et al. (2015, GMD in prep)

NO2 Vertical Tropospheric ColumnBadia et al. (2015, GMD in prep)

19

rural WDCGG, CASTNET and EMEP stations

O3 surfaceBadia et al. (2015, GMD in prep)

Period: Run one year simulation (2010).

Domain: European simulations: 30W- 60E, 25N-70N

Chemical BC: MACC (IFS-MOZART)

Meteorological BC: NCEP/FNL 1ºx1º

Emissions: TNO-MACC; Biogenics: MEGAN; No Fire Emissions

Horizontal Resolution: 0.2º x 0.2º

Vertical Resolution: 24 (and 48) top 50hPa

Gas Chemical mechanism: CB05 Blue: model domainRed: AQMEII domain (to submit)Green: BC domain

Regional Experiment configuration – AQMEII-Phase2

Badia and Jorba (2014, AtmEnv)Ulas et al. (2014)

O3

• Captures high NO2 over the most polluted regions.• Over land: Overestimates in big cities and underestimates in rural regions.• Over sea: Overestimation in Mediterranean (Italy) and North seas -> shipping emissions or

stability of marine boundary layer?

r=0.65

r= 0.87

r=0.84

r=0.84

(Badia and Jorba, 2014)

NO2 Vertical Tropospheric Column vs OMI

CO mixing ratio against MOPITT

• The pattern of emissions in central EU is well-captured.• Over land: satellite evaluation confirms that there is a general trend to underestimate

surface CO• Summer underestimation due to no fire emissions (important fires in Russia and Portugal)

r=0.91

r= 0.87

r=0.82

r=0.90

(Badia and Jorba, 2014)

24

Ongoing and future developments and research activities• Complete modularity of the BSC-CTM code within NMMB code

• Test NMMB/BSC-CTM with online nesting (global-regional/regional-regional)

• Replace current WRF/CMAQ/DREAM air quality forecasts for Europe (12 km) and Spain (4 km) at BSC (CALIOPE System)

• Test regional model for North America

• Updating dust emission scheme and dust sources for high resolution NEMS dust forecasts in the US (P. Ginoux, A. Deroubaix, S. Basart)

• Coupling of chemistry gas-phase with a detailed SOA scheme for high-resolutions applications (D. Dabdub, M. Spada)

• Evaluate nitrate and ammonium (M. Spada)

• Online coupling of a volcanic ash module (Fall3D model, A. Marti and A. Folch)

• Aerosol-radiation feedbacks: effects upon weather forecasts (A. Gikkas), and regional climate simulations (M. Gonçalves)

• Aerosol data assimilation data assimilation using the Local Ensemble Transform Kalman Filter (LETKF) (Enza DiTomaso)

Participates in the ICAP global-model intercomparison project http://www.nrlmry.navy.mil/aerosol/icap.1087.php

Mineral dust forecasts at the WMO Sand and Dust Storm Warning System DS-WAS North Africa, Middle East and Europe portal

http://sds-was.aemet.es/

AQMEII on-line Air Quality model intercomparison project

Charmex Chemistry-Aerosol Mediterranean experiment

BSC aerosol and chemistry modeling collaborations

25

Barcelona Dust Forecast Center: http://dust.aemet.es/

26

First Specialized Center for Mineral Dust

Prediction of the World Meteorological

Organization

Numerical forecasts based on the NMMB/BSC-CTM

Dust component

www.bsc.es

Thank you!

oriol.jorba@bsc.escarlos.perezga@nasa.gov

27

Work funded by grants CGL2006-11879, CGL2008-02818, CGL2010-19652, CSD00C-06-08924, CGL2013-46736-R and Severo-Ochoa of the Spanish

Ministry of Economy and Competitiviness

www.bsc.es/projects/earthscience