Yingying Yan 燕莹莹, Jintai Lin 林金泰, Xiong Liu, Jinxuan Chen School of Physics, Peking University Tropospheric Ozone in Two-way Coupled Model of GEOS-Chem

Un-even terrain Small-scale meteorology Variability in land use, vegetation, etc. Small-scale horizontal & vertical transport Small-scale variability in chemistry & emissions modelreality  r AB < 0 : Model has an overestimation  r AB > 0 : Model has an underestimation 2 Limited by Resolution Models Misrepresent Small-scale Processes

Developments by regional models Ground-level NMVOCs/NOx ratio and ozone diff.  Emissions Small-scale variability smoothed Resolution-dependent natural emission (6.5% higher in biogenic NMVOCs)  Ozone production regime: 20  Negative diff. : East China East America Europe  Positive diff. : Tibet Plateau Rocky Mountains  Topography smoothed

High-res regional nested simulations ‘correct’ global model Global and multiple regional models interact simultaneously High computation efficiency and low model complexity Differences can be transported to outside the nested domains and accumulate during the lifetime of chemical species 4 Global model : ~ 200km res. Regional models: ~ 50 km res. Yan Y.-Y. et al., 2014, ACP Global-multi-regional Two-way Coupled Modeling

Simulation flowchart 5 Global v.s. regional models: –Different horizontal resolutions –Consistent data inputs Multi models run independently Models ‘talk’ via simple I/O files Low CPU and I/O burdens Fast calculation Yan Y.-Y. et al., 2014, ACP Global ↔ Regional Model Interaction Scheme

6 Files in the new PKUCPL subdirectory: 1. Twoway.compile.sh: compile 2-way modeling 2. PKUCPL.F90: modulate the normal DO-Loop, to coordinate all models 3. PKUCPL.sh: provide utility to run models in the right directories 4. run:./run 2cne (couple global model (2x2.5) with three nested models) Steps: 1. Set up the run directories 2. Compile for two-way coupled model./Twoway.compile.sh 3. submit the batch job script 'run_twoway' for two-way simulation qsub run_twoway (See Development in the future: Easy to include the newly built and higher resolution nested models Coupled simulation setup steps Easy O(∩_∩)O Flexible \(^o^)/ Cool to run all models coupled in parallel

Better simulation of ground-level ozone  Mean R increases from 0.51 to 0.65  Mean bias decreases by 4.8 ppb Yan Y.-Y. et al., 2015, in prep

Tropospheric ozone profile Obs Two-way Global

Tropospheric Column Ozone MAM DJF SON JJA  OMI_MEAN=(OMI/MLS + OMI/LIU)/2  2-way minus global: 3.2 DU (8.9%)  2-way – OMI_MEAN: MAM: 2.2 DU (6.5%) JJA: -0.7 DU (-1.7%) SON: 1.6 DU (5.6%) DJF: -0.8 DU (-2.1%)

10 2-way ~ (Global + CO emis) HIPPO-1: + 15% HIPPO-2: + 25% HIPPO-3: + 15% HIPPO-4: + 25% HIPPO-5: + 35% Mean : + 25% It is an important model dependence on resolution that is largely unaccounted for CO emission constraints. Yan Y.-Y. et al., 2014, ACP Improvements in CO Simulation

11 Yan Y.-Y. et al., 2014, ACP; 2015, in prep * Greater than its interannual variability (2.3%) # Greater than the change from 2000 to 2100 under RCP6.0 & Equivalent to a 25% increase in global CO emissions Global ModelTwo-way Model‘Observation’ OH (10 5 cm -3 ) ( – 5% * )10.4 – 10.9 MCF lifetime (yr) ( + 5.2% )6.0 – 6.3 CH 4 lifetime (yr) ( + 5.1% )10.2 – 11.2 O 3 (DU) ( – 8.7% )31.1 ± 3 (OMI/MLS) O 3 (Tg) ( – 9.4% # ) NOx (TgN) ( + 4.1% ) CO (Tg) ( % & ) NMVOC (TgC) Improvements in tropospheric Simulation