11th CMAS Meeting, RTP October 15 – 17, 20121

Fine Resolution Air Quality Forecasting Capability for limited-area domains

– tested over Eastern Texas

Pius Lee, Hyuncheol Kim, Li Pan, Daniel Tong, and Tianfeng ChaiAir Resources Laboratory, NOAA

Shobha Kondragunta, Pubu Ciren, Qiang Zhao, Chuanyu Xu, and Xiaoyang Zhang

National Environmental Satellite, Data, and Information Service

Jianping Huang, Sarah Lu, and Jeff McQueenNational Centers for Environmental Predictions, NOAA

Ivanka StajnerOffice of Science and Technology

2

NMMB grids: CONUS (green)*, nested Discover-AQ (red); in relation to 5x (blue)

*1326*795=1.05 million horizontal grid cells @ 22 vertical layers

11th CMAS Meeting, RTP October 15 – 17, 2012

Need for fine resolutionLimited-area domain:• affordability• Study finer scale processes

Talk (Tues): Ivanka Stajner -- NAQFC Talk (Wed): Tianfeng Chai – Chemical D.A.

311th CMAS Meeting, RTP October 15 – 17, 2012

Next generation operational AQ modeling – consensus among many operational centers (e.g. Lee and Ngan 2011)

Better describes complex terrain & land/sea interface processes

Helps field campaign planning (e.g. NAQFC for DISCOVER-AQ 2010)

Imperative to capture certain pollution scenarios, e.g. wild fire induced hazards -- smoke (Hu et al. 2008 )

Fire weather forecasting a minimum of 24 hours lead time (e.g. NOAA 2012)

Some applications of fine resolution AQ modeling

1.3 km fire weather on-demand nestCourtesy: mmb/emc/NCEP

Poster: Christopher Loughner – DISCOVER-AQ Baltimore-Washington

Poster: Yongtao Hu

411th CMAS Meeting, RTP October 15 – 17, 2012

An on-demand event-based chemistry and air quality forecasting system Limited-area domain within the Conterminous U.S.A (CONUS)

Parent domain:12 km horizontal spacing resolution CONUS

Child domain:4 km resolution limited-area domain

Emission generated at 4 km back- aggregated to 12 km domain

Tested for Eastern TX

Configuration of 4km AQ limited-area forecasting domain within CONUS

NMMB

NMMB-Post & AqmPrdgen

PREMAQ

CMAQ

GRiB products & graphics

Verification

ICON

BCON from

CONUS results

SMOKE

MOBILE6

Poster: Li Pan – Emission upgrades NAQFC 2012

511th CMAS Meeting, RTP October 15 – 17, 2012

Versatility of selecting a limited-area domain of interest

Limited-area domain forecasts are heavily influenced by boundary conditions and their derivation is criticale.g. exo-domain wild fire emissions

~21x

~12x

5x

Agricultural burningprevails in the monthsof March and Aprilin Mexico

HMS wildfire detections during Apr. 2010Talk (Wed): Hyuncheol Kim – Fire Emission

Poster: Yongtao Hu – Wildfire Impact

Poster: Ka-wa Chan – Concept model

611th CMAS Meeting, RTP October 15 – 17, 2012

Lowest: one to one lowest 5 layers

PBL: Next 6 CMAQ levels corresponds to 12 NAM levels up to about 1350 m AGL,

Free Troposphere: Next 7 CMAQ levels corresponds to 21 NAM levels to 5000 m AGL,

Tropopause: Next 4 CMAQ levels corresponds to 8 NAM levels embedding the transition @ 200 hPa

The remaining CMAQ full-interface level caps the model at 100 hPa. It corresponds to 3 NAM levels.

Vertical coupling between NMMB and CMAQ: Layer correspondence

61

6059

12

n

NMMB Layer

200hPa

15 layers

PD

711th CMAS Meeting, RTP October 15 – 17, 2012

5x

Obtain CO concentration at lateral boundaries for 5x from predicted PM concentration

e.g. PM/CO ratio:Forest: 0.13±0.05Savanna: 0.08±0.03Grass: 0.07±0.03(Vicente et al. 2011;Janhäll et al. 2010)

Obtain NOx, VOC, NH3, SO2

Concentrations scaled from CO using EPA SPECIATE data 2010

HMS-Bluesky-SMOKE for wild fire emissions within 5x

5x provides IC and BC for HOU Limited-area 4km domain

HYSPLIT

CMAQ

CMAQ

811th CMAS Meeting, RTP October 15 – 17, 2012

Modeled O3 difference between 12km and 4km domains at 13UTC August 26, 2012

At 4km resolution model simulates higher ozone concentrations aloft and resolves smaller features at all levels: interaction of emissions and meteorology at finer scales

Surface

Surface Surface

900 m 900 m 900 m

12-km (cut from 5X CONUS) 4-km Houston domain

Finer descriptions helped

12-km (cut from 5X CONUS) 4-km Houston domain

Finer descriptions helped

Study other metrics?

Metrics Paired (4km)Paired

(12km)

RMSE (ppb) 15.65 14.94

NME (%) 25.36 24.29

MB (ppb) 1.86 2.43

NMB (%) 3.92 5.13

R 0.55 0.61

The performance measures over the 4 km resolution may not be necessarily better than over the coarser (12 km) resolution; it may be even worse if it is evaluated using the traditional evaluation metrics based on paired obs-mod data

Courtesy: Daiwen Kang, CMAS 2011

12

Statistical metrics for high resolution AQ model evaluation -- Need New paradigm !

11th CMAS Meeting, RTP October 15 – 17, 2012

Preparing the Air Quality Community for GOES-R Advanced Baseline Imager (ABI) Aerosol Products

GOES-R ABI Air Quality Proving Ground project successfully deployed a web-based dissemination of aerosol proxy data to air quality forecasters from different states in preparation for day one readiness of GOES-R products.

GOES-R Aerosol Optical Depth (AOD), aerosol type, RGB image on the hour for every hour of the day are streamed via a website located on a STAR computer.

ABI aerosol proxy retrievals are generated using ABI radiances generated by Community Multiscale Radiative Transfer Model (CRTM). The 24-hr aerosol forecast fields from 00Z run of WRF-CMAQ (Weather Research and Forecasting-Community Multiscale Air Quality) model are provided as inputs to CRTM.

13

http://www.star.nesdis.noaa.gov/smcd/spb/aq/aqpg/

• Key instrument on GOES-R is the Advanced Baseline Imager (ABI):– 16 bands with spatial resolution of 2 km (nadir) and 5 min refresh rate for CONUS– AOD with accuracy similar to MODIS (multi-channel aerosol retrieval)– Aerosol type information (smoke vs. dust)– Fire/hot spot characterization– Visible, IR, water vapor imagery

Huff et al., The NOAA Air Quality Proving Ground: Preparing the Air Quality Community for Next Generation Products from The GOES-R Satellite, EM, accepted, 2012.

Courtesy: S. Kondragunta

1411th CMAS Meeting, RTP October 15 – 17, 2012

Courtesy: S. Kondragunta and C. Xu, NESDIS STAR

Flow Chart : Process for Generating Proxy ABI Aerosol Products

12km and 4 km results were input to CRTM – attempted for ABI imagery

3. CRTM Run on NOAA STAR computer orbit006l with 12GB memory and 16 CPUs. Code based on FORTRAN, IDL, and shell

scripts

5. ABI AOD algorithm run on

orbit006l for daytime scenes

4. GOES-R ABI Synthetic Radiances

(6 bands: 0.47, 0.66, 0.86, 1.36, 1.6, 2.25 m)

1. Hourly outputs of aerosol and met fields from 48-hour

-nmmb forecast run at 12-km resolution

(ftp.nsstc.org/outgoing/yes)

36 min

6 hrs

7. Post-processing of ABI aerosol products using IDL and ImageMagic to

generate display imagery files

2. Re-formatting of CMAQ outputs

to prepare for CRTM run

2 min 50 min

1 min

6. Simulated ABI aerosol products

8. ABI aerosol imagery

9. Web display: http://www.star.nesdis.noaa.gov/smcd/spb/aq/aqpg/

30 min

1511th CMAS Meeting, RTP October 15 – 17, 2012

Examples of Near Real-Time ABI Proxy Aerosol Products for July 30, 2011, 14:00 UTC: AOD (left), Aerosol Type (center), RGB

(right).

Courtesy: S. Kondragunta and C. Xu, NESDIS STAR

12km and 4 km results were input to CRTM – attempted for ABI imagery

RGB at 15:00 2012-08-13 RGB at 17:00 2012-08-13

RGB at 19:00 2012-08-13 RGB at 21:00 2012-08-1311th CMAS Meeting, RTP October 15 – 17, 2012

ABI Proxy Aerosol Products for 12 km results

AOD at 15:00 2012-08-13 AOD at 17:00 2012-08-13

AOD at 19:00 2012-08-13 AOD at 21:00 2012-08-1311th CMAS Meeting, RTP October 15 – 17, 2012

ABI Proxy Aerosol Products for 12 km results -- cont’d

18

Summary

11th CMAS Meeting, RTP October 15 – 17, 2012

4km AQ forecasting system for a limited-area domain has been tested:o Houston domain was tested for multiple dayso The fine resolution window over HOU showed more local

sfc O3 maxima and stronger concentration gradiento Inner-nest domain averaged time series showed mixed

results. The grid-to-obs paired statistics may not be adequate

Multiple application of 4m limited-area domain QA forecasting:o Intensive measurement campaignso Pre-launch calibration of retrieval algorithms.

Looking ahead: o CMAQ5.0 Options: lightning NOx, Windblown-dust o Explore metrics for 4km domain verification and evaluation

Poster: Daniel Tong – Windblown dust emission module

Acknowledgement:

This work is partially supported through NASA Air Quality Applied Sciences Team (AQAST) Tiger Team project.