This presentation was given as the foundation for the EPA Student Working Group problem lab. Working group members used this information to begin researching their problem and later went on to present their findings to their peers.

1. (Hybrid) Splicing on Multi-Scale
Downscaler Air Quality Surfaces
Elizabeth Mannshardt
Barron Henderson
Brett Gantt
US EPA
Brian Reich, NCSU Faculty Adviser

2. 2
EPA
What is the EPA?
The US Environmental Protection Agency
is tasked with protecting human health and
the environment.

3. 3
EPA
EPA Offices
Office of the Administrator
Office of Administration and Resources Management
Office of Air and Radiation
Office of the Chief Financial Officer
Office of Chemical Safety and Pollution Prevention
Office of Enforcement and Compliance Assurance
Office of Environmental Information
Office of General Counsel
Office of Inspector General
Office of International and Tribal Affairs
Office of Land and Emergency Management
Office of Research and Development
Office of Water
Regional Offices (10)

4. Looking Back
Donora, PA: Noon on October 29, 1948.
20 people died, thousands sickened
1970: Clean Air Act
• The Clean Air Act is a US federal law
designed to control air pollution on a
national level. One of the US' first and
most influential environmental laws, and
one of the most comprehensive in world.
• Congress designed the CAA to combat a
variety of air pollution problems, and to
tackle emerging pollution threats
• Under the CAA, health-based air quality
standards are based on latest science.
https://en.wikipedia.org/wiki/Clean_Air_Act_%28United_States
https://www.epa.gov/clean-air-act-overview

5. Clean Air Progress: 1970-2016
Office of Air and Radiation (OAR)
Pollution Down While Growth Continues

6. 6
Office of Air Quality Planning and Standards (OAQPS)
• What pollutants are in the air?
– Right now, and in the recent past
– By hour, day, and longer periods
– At monitoring sites; everywhere else
• Where does the air pollution come from?
– What sources
– What locations
• How will things change in the future?
– Potential impact of our policy options
– Meeting air quality & program goals
“Air Quality Assessment”

7. Office of Air Quality Planning and Standards (OAQPS)
NAAQS: National Ambient Air Quality Standards
https://www.epa.gov/air-trends/air-quality-national-summary
In 2016, the US EPA estimated that 122.5
million people live in counties where air
quality concentrations were above the
levels of one or more of the EPA’s NAAQS
The World Health Organization reported in
2018 that ambient air pollution is
estimated to have caused 7 million
premature deaths in 2016
http://www.who.int/gho/phe/air_pollution_mortality/en/

8. BenMAP: Open-Source Platform to Quantify Health
Impacts and Economic Value of Stressors – Neal Fann, HEID
Magnitude
of impacts
Millions
Thousands
Tens of
Thousands
Proportion of population affected
>90% of
monetized
benefits
Severity of
Effects
Ln(y) = Ln(B) + ß(PM)
Incidence
(log scale)
PM concentration
Ln(B)
∆Y =Yo (1-e -ß∆ PM) * Pop
ß = Effect estimate
Yo = Baseline Incidence
Pop = Exposed population
Health impact function:
Pollutant change
Effect
estimate
Health
impact
Population Baseline incidence

9. 9
National Downscaler
Currently EPA provides a fused data product (downscaler) for air quality surfaces on a national
spatial scale. Broad aim - assess ways to improve the performance of the downscaler (DS) in
certain areas of the continental US, particularly in areas with sparse monitor representation.
First step: consideration of predictions run on a regional scale (e.g., NOAA climate regions)

10. 10
NOAA Climate Regions
https://www.ncdc.noaa.gov/monitoring-references/maps/us-climate-regions.php

11. Apply downscaler to 9 NOAA climate regions using overlap
Regional Downscaler

12. 12
Regional Downscaler
Predictions on a regional scale (NOAA climate regions)
Goal: Smooth National Surface

13. 13
Splicing
Critical step: splicing regions back together along the regional boundaries to create
smooth national surface.

14. 14
This could possibly be achieved through
• a stochastic statistical model
• a deterministic model
• a combination
Uncertainty quantification is an important component.
Splicing

15. 15
The spliced boundary area could be assessed via comparisons
to monitor data, CMAQ, and DS.
Important considerations include assessment of the
discrepancies within the boundary overlap including:
• scale, scope and range of differences;
• gradient changes within the boundary area;
• magnitude of discrepancy at interior versus outer boundary
edge.
Splicing

16. Q1

17. 17
Goal:
A methodological framework/algorithm for splicing regional estimates to
create a smooth national surface.
Methodology should include consideration of scalability, as the
downscaler is often run on a daily temporal scale.
The 9 NOAA climate regions may not be selected as the optimal spatial
regions from which to develop a national surface, but the assessments
performed and the methodology established will be applicable to future
spliced surfaces.
Splicing

18. 18
Hybrid Spliced Multi-Scale Downscaler
Ultimately will combine different regional/local scales with
additional data source(s) (IMPROVE network; remote
sensing) for “Hybrid” Spliced Multiscale Downscaled
Surface

19. 19
EPA – RTP

20. 20
EPA
How can you work with EPA?
Internships: http://www.epa.gov/careers/student-internships
Students and Recent Graduates – Federal Pathways Program:
https://www.usajobs.gov/StudentsAndGrads
Careers: https://www.usajobs.gov/
Careers at EPA: https://www.epa.gov/careers

21. 21
EPA
Thank you!
References
AirTrends Report, EPA 2016: https://gispub.epa.gov/air/trendsreport/2016/
BenMAP, EPA 2015: https://www.epa.gov/benmap
Berrocal, V. J., Gelfand, A. E. and Holland, D. M. (2010a). A spatio-temporal downscaler for
outputs from numerical models. J. Agric. Biol. Environ. Stat. 15 176–197.doi:10.1007/s13253-
009-0004-z