timdye

Pilot PM2.5 Forecasting Projects for 21 USA Today Cities: 

Pilot PM2.5 Forecasting Projects for 21 USA Today Cities Sonoma Technology, Inc. Petaluma, CA (707) 665-9900 www.sonomatech.com tim@sonomatech.com Presented at EPA’s National Air Quality Conference: Mapping and Forecasting San Antonio, TX February 2-5, 2003 Acknowledgments: U.S. Environmental Protection Agency 902461-2322 David E. B. Strohm Clinton P. MacDonald Timothy S. Dye Craig B. Anderson Presented by

Objective and Outline: 

Objective and Outline Objectives: Develop PM2.5 forecasting methods/tools and operationally forecast PM2.5 for 21 cities Transfer knowledge and forecasting tools to each city Outline: Background Approach and schedule Conceptual model development Forecasting tool development Operational forecasting and results Forecasting software Next steps

Background (1 of 2): 

Background (1 of 2) Several driving issues Health effects PM2.5 year-round pollutant Dominant wintertime pollutant Media’s (USA Today) desire for year-round air quality forecasts Challenges PM is different than ozone Primary and secondary pollutant Seasonal changes in PM Regional differences in PM Local hot spots Very limited historical data Uncertainties in continuous real-time data PM chemistry is complex

Background (2 of 2): 

Background (2 of 2) Honolulu USA Today Cities

Approach and Schedule: 

Approach and Schedule General approach Select cities (Oct) Acquire historical data (focus on FRM, 24-hr average measurements) (Oct-Jan) Quality control data (Oct-Jan) Develop conceptual model understand weather and air quality in each city (Oct-Jan and Feb-Mar) Develop forecasting tools (Dec-Feb) Conduct operational forecasting (Jan-April) Document tools and lessons learned (April-May) Evaluate forecasts (May-June) Transfer tools and knowledge to staff in each city (June-July)

Conceptual Model Development: 

Conceptual Model Development Components of a Conceptual Model Climatology Analysis: Examine past air quality conditions to help characterize PM2.5 episodes and provide guidelines for forecasters Weather Characterization: Examine the large-scale weather patterns that influence PM2.5 concentrations Case Studies: Conduct detailed analysis of specific episodes to understand local processes that affect PM2.5 Forecasting tools are based on this conceptual understanding. Phenomenological forecast tables Statistical tool development

Conceptual Model – Climatology (1 of 2): 

Conceptual Model – Climatology (1 of 2) Developed a climatology for each city: Examined the frequency and characteristics of PM2.5 episodes Monthly frequency Day-of-week frequency Holiday frequency

Conceptual Model – Climatology (2 of 2): 

Conceptual Model – Climatology (2 of 2)

Conceptual Model – Weather Characterization: 

Conceptual Model – Weather Characterization Common large-scale weather features associated with high PM2.5 Aloft ridge of high pressure or approaching trough Prefrontal conditions Temperature inversion and stable conditions Local stagnation or transport between cities (East) Holiday effect Small-scale subtle features are important and are becoming evident in daily forecasting

Conceptual Model – Case Studies: 

Conceptual Model – Case Studies * * * * *

Forecasting Tool Development (1 of 2): 

Forecasting Tool Development (1 of 2) Phenomenological Tables

Forecasting Tool Development (2 of 2): 

Forecasting Tool Development (2 of 2) Statistical tools Regression and CART Initial assessment is more difficult than for ozone Rare event forecasting Multiple pathways to form PM High emissions variability Limited historical data Use some stability and transport parameters Washington D.C. PM2.5 Winter Climatology

Operational Forecasting: 

Operational Forecasting NOTE: These forecasts are not being sent to the public and are only intended for internal use by your agency during this pilot PM2.5 forecasting project. Regional PM2.5 Forecast for Baltimore Today's forecast - Monday, January 27: Good (32 AQI) Tomorrow's forecast - Tuesday, January 28: Moderate (52 AQI) Yesterday's observed maximum - Sunday, January 26: Moderate (60 AQI) Discussion: A strong trough of low pressure aloft moved through the area late yesterday, resulting in strong vertical mixing and lower PM2.5 concentrations. Surface high pressure is building into the area today, but moderate northerly winds will keep PM2.5 levels Good. Tomorrow, there will be zonal flow aloft, and the surface high pressure system will be centered right over the Baltimore area, resulting in light winds. These conditions will allow PM2.5 levels to reach the low end of Moderate. This forecast was issued on Monday, January 27, 2003 at 12:36:37 PM PST Forecaster: Dianne Miller Sonoma Technology, Inc. office: 707-665-9900 cell phone: 707-338-0304 dianne@sonomatech.com Sample Daily E-mail

Initial Forecasting Results: 

Initial Forecasting Results Forecasting since January 2 for 10 cities Using phenomenological tables (modifying as we gain more experience) Weather conditions have generally not been conducive to high PM2.5 except in the West Next-day Forecasts Hit - Good Hit - Moderate and above Miss

Forecasting Software – Overview: 

Forecasting Software – Overview Centralized, web-based approach to forecasting Ingests, processes, stores, and integrates large amounts of information Weather observations and forecasts Air quality observations and forecasts Air quality links Verification statistics Climatologies Customizable by forecasters in each agency Expandable to other cities, pollutants, and forecast tools

Forecasting Software – Example: 

Forecasting Software – Example Screenshots from Forecaster Software

Next Steps: 

Next Steps Start the process for the remaining 11 cities Conceptual model development Tool development Daily forecasting Document the results Conduct regional workshops to transfer knowledge and products to the 21 cities

Contacts: 

Contacts U.S. EPA Pat Dolwick, (919) 541-5346, dolwick.pat@epa.gov John White, (919) 541-2306, white.johne@epa.gov Sonoma Technology, Inc. Tim Dye, (707) 665-9900, tim@sonomatech.com Clint MacDonald, (707) 665-9900, clint@sonomatech.com Craig Anderson (707) 665-9900, craig@sonomatech.com