Supermodel, Supermodel, Can I Breathe Tomorrow? Talat Odman* and Yongtao Hu Georgia Institute of Technology School of Civil & Environmental Engineering Georgia Air Quality & Climate Summit May 4, 2006

Air Quality Forecasting Increasing interest in day-to-day air quality –Public awareness –Short-term local management strategies Forecasts are produced using various techniques –Persistence –Simple empirical “rules-of-thumb” –Statistical regression –Complex heuristics In Atlanta, since 1996 –Panel of experts produce a forecast Recently, forecasts based on numerical models have emerged

Numerical Forecasting Efforts NOAA/EPA –Eta-CMAQ modeling system MCNC/BAMS –MM5-MAQSIP-RT modeling system NCAR/NOAA –WRF-Chem modeling system Canadians –GEM-CHRONOS modeling system

Goal To provide accurate “fine-scale” local forecasts sufficiently in advance for planning purposes NOAA/EPA’s target is to issue nationwide 2-day forecasts with 2.5-km resolution 10 years. –Davidson, P. M. et al., “National Air Quality Forecasting Capability,” February 14, We want to get there (and beyond) locally much faster. –Longer periods –Finer resolution –Viability of strategies to avoid bad episodes

WRF for meteorology SMOKE for emissions CMAQ for chemistry and transport Our Modeling System

Modeling Domain and Grids Three grids: –36-km (72x72) –12-km (72x72) –4-km (99x78) Horizontal domains are slightly larger for WRF 34 vertical layers used in WRF 13 layers in CMAQ

Current Operation WRF is driven by NAM (formerly ETA) data –3 ½ -day NAM data available every 6 hours (00, 06, 12, 18Z) Tomorrow’s forecast by 10 a.m. today –Friday’s operation started on Tuesday night We simulate: –3 days over the 36-km grid using 00Z NAM and IC from previous cycle and “clean” BC –2 ½ days over the 12-km grid using 12Z NAM and IC/BC from 36-km –28 hours over the 4-km using 12Z NAM and IC/BC from 12-km Mostly automated, employ 2 people and 6 CPUs The product is a 24-hr forecast once per day

Emission Forecasting Our goal is to use most up-to-date emissions inventories We projected the NEI-2002 emissions to 2006 using growth and control factors –EGAS model –NO x SIP controls We use monthly-averaged data for major point sources and wild fires We forecast mobile emissions –Emission factors use the episode (3, 2 ½ or 1 day) average temperature We forecast biogenic emissions using summertime leaf indexes

Today’s & Tomorrow’s Forecasts

Metropolitan Atlanta Today: –Peak 1-hr ozone will be 65 ppb at Gwinnett at 2 p.m. –Peak 1-hr PM2.5 will be 29.6  g m -3 at Gwinnett at 8 a.m. Tomorrow: –Peak 1-hr ozone will be 65 ppb at Yorkville at 2 p.m. –Peak 1-hr PM2.5 will be 32.0  g m -3 at South DeKalb at 8 a.m.

Yesterday’s Forecast

Peak 1-hr Ozone & PM 2.5 Predicted peak ozone –Conyers –4 p.m. –72 ppb Predicted peak PM 2.5 –South DeKalb –8 a.m. –23.6  g m -3 Observed –Conyers  –4 p.m.  –66 ppb  (9% over prediction) Observed –Confederate Ave.  (nearest monitor) –10 a.m.  (2 hrs in advance) –29.5  g m -3  (20% under prediction)

12-km Ozone at 4 p.m.

Ozone at Conyers on 5/3/2006

Ozone in Metro Atlanta

PM 2.5 at South DeKalb

PM 2.5 in Metro Atlanta

Other Places in Georgia

Athens

Augusta

Columbus

Macon

Next Steps Find support –Many thanks to Georgia Tech Forecasting Group and GA DNR for seeding this effort –Need at least 5x in FY-07 Objectives: –Continue the operation –Set up a user-friendly web site –Archive the data for future use –Extend the domain of coverage –Increase the resolution –Elongate the forecast period –Issue a daily update –Start an evaluation program –Improve the accuracy