Integrated Assessments: Integrated Assessments: Using CMAQ/CAMx to Evaluate Using CMAQ/CAMx to Evaluate

Health Impacts of Air Pollution in Health Impacts of Air Pollution in the United States and Chinathe United States and China

Denise L. MauzerallDenise L. MauzerallXiaoping WangXiaoping WangQuansong TongQuansong Tong

Science, Technology and Environmental Policy programScience, Technology and Environmental Policy programWoodrow Wilson SchoolWoodrow Wilson School

Princeton UniversityPrinceton University

Presented at Models-3 Workshop, RTP, NC, Oct. 28, 2003

Overall ObjectivesOverall Objectives

Utilize science to inform air quality policyUtilize science to inform air quality policyMethodology: Methodology: – Use atmospheric modeling and available data to Use atmospheric modeling and available data to

describe air qualitydescribe air quality– Evaluate impacts of air pollution on human health and Evaluate impacts of air pollution on human health and

agriculture using exposure-response functions from agriculture using exposure-response functions from literatureliterature

– Monetize the costs of the impactsMonetize the costs of the impacts– Examine alternative energy/pollution control Examine alternative energy/pollution control

technologies and policies for optimal cost-effective air technologies and policies for optimal cost-effective air quality control strategies quality control strategies

Ongoing Integrated Assessment Projects Ongoing Integrated Assessment Projects

at Princeton Using Models-3/CMAQat Princeton Using Models-3/CMAQ

Evaluate effectiveness of the current NOEvaluate effectiveness of the current NOxx emissions cap-and-trade program in the North-emissions cap-and-trade program in the North-Eastern United States on reducing surface OEastern United States on reducing surface O33 levels and resulting health effects. levels and resulting health effects.

Evaluate impact of air pollution on health in the Evaluate impact of air pollution on health in the Shandong region of China.Shandong region of China.

Evaluate long-term impacts of air pollution on Evaluate long-term impacts of air pollution on health and the environment in the United States.health and the environment in the United States.

Integrated Assessment ApproachIntegrated Assessment Approach

Technology

Pollutant Emissions

Ambient Concentrations

Human Exposure

Health Effects

Social Benefits

SMOKE

Epidemiological Exposure-Response

Functions

MM5 / RAMS

CMAQ / CAMx

Population Distribution

Economic Analysis

Policy ControlOptions

Charging NOx Emitters for Health Damages:Charging NOx Emitters for Health Damages:An Exploratory AnalysisAn Exploratory Analysis

Denise L. Mauzerall, Denise L. Mauzerall,

Babar Sultan, David BradfordBabar Sultan, David Bradford

Princeton UniversityPrinceton University

OutlineOutline

Description of NODescription of NOxx Emissions Trading in NE United States Emissions Trading in NE United States

Evidence that NOEvidence that NOxx Cap-and-Trade program has failed to Cap-and-Trade program has failed to reduce surface Oreduce surface O33 concentrations concentrations

Free trades permitted between May – September Free trades permitted between May – September regardless of: regardless of: – temperature variability temperature variability – Biogenic hydrocarbon emissionsBiogenic hydrocarbon emissions– Population densityPopulation density

Modeling to answer questions of how OModeling to answer questions of how O33 production and production and mortalities changes with:mortalities changes with:– high / low temperatures on the days of NOhigh / low temperatures on the days of NOxx emission? emission?– Regions of high / low isoprene emission? Regions of high / low isoprene emission? – Regions of high / low population density?Regions of high / low population density?

OTC NOx Budget Cap and Trade ProgramOTC NOx Budget Cap and Trade Program

Goal:Goal: Reduce summer NOx emissions within 13 north-eastern Reduce summer NOx emissions within 13 north-eastern U.S. states in order to attain NAAQS for surface OU.S. states in order to attain NAAQS for surface O3 3 (85 ppb (85 ppb over 8-hour average).over 8-hour average).

Regulatory Approach:Regulatory Approach: – Limit total Limit total NOxNOx emissions from stationary sources such as emissions from stationary sources such as

power plants and industrial boilers, but permit trading power plants and industrial boilers, but permit trading among emission sources.among emission sources.

– Program capped summertime NOx emissions in 1999 at Program capped summertime NOx emissions in 1999 at less than half of the 1990 baseline emissions of 490,000 less than half of the 1990 baseline emissions of 490,000 tons.tons.

Question:Question: Do O Do O33 concentrations in the region decrease after concentrations in the region decrease after the emissions cap-and-trade program is put in place in the emissions cap-and-trade program is put in place in summer 1999?summer 1999?

EPA-AIRS O3 Data

In 13 eastern U.S.OTC states May - September

Black = 1995-1998 (pre-cap)

Red = 1999-2001 (post-cap)

There is no significant reduction in O3

concentrations after the cap-and-trade program isin place!

Modeling AnalysisModeling Analysis

For each scenario, two CAMx simulations are For each scenario, two CAMx simulations are conducted for July 7-18, 1995:conducted for July 7-18, 1995:– Standard simulation with all regional emissions.Standard simulation with all regional emissions.– Perturbation simulations (NOx emissions from Perturbation simulations (NOx emissions from

individual power plants are reduced by a fixed amount) individual power plants are reduced by a fixed amount) during conditions of:during conditions of:

high/low temperature, high/low temperature, high/low isoprene, high/low isoprene, High/low population.High/low population.

CalculateCalculateOO33 resulting from difference between resulting from difference between standard run and each perturbation run standard run and each perturbation run Estimate change in mortality resulting from Estimate change in mortality resulting from change of Ochange of O33 distribution and exposed distribution and exposed population. population.

NOx emissions = 1.77 x 106 moles from one power plant in 24-hours results in the following change in O3 concentration and mortalities.Mean Temp = 302 K +/- 3.6 K (83.5 F)

High Temperature Period

Total O3 increase = 36 ppb Total Increase in Mortality = 0.17

NOx emissions = 1.77 x 106 moles from one power plant in 24-hours results in the following change in O3 concentration and mortalities.Mean Temp = 296 K +/- 4.3 K (72 F)

Low Temperature Period

Total O3 increase = 22 ppb Total Increase in Mortality = 0.0

NOx emissions = 1.77 x 106 moles NOx from one power plant in 24-hours results in the following change in O3 concentration and mortalities.

High Isoprene Emission Region

Total O3 increase = 170 ppb Total Increase in Mortality = 0.13

Change in O3 concentration andresulting change in Mortality for unit change in NOx emissions from a single power plant in 24-hours.

NOx = 1.77 x 106 moles NOx.

Low Isoprene Emission Region

Total O3 increase = 46 ppbTotal Increase in Mortality = 0.09

Change in O3 concentration andresulting change in Mortality for unit change in NOx emissions from a single power plant in 24-hours.

NOx = 1.77 x 106 moles NOx

High Population Region

Total O3 increase = 40 ppb Total Increase in Mortalities = 0.21

Change in O3 concentration andresulting change in Mortality for unit change in NOx emissions from a single power plant in 24-hours.

NOx = 1.77 x 106 moles NOx

Low Population Region

Total O3 increase = 121 ppb Total Increase in Mortalities = 0.19

Conclusions on Emissions TradingConclusions on Emissions Trading

NOx emissions in locations of high temperature NOx emissions in locations of high temperature and high isoprene emissions result in higher Oand high isoprene emissions result in higher O33 concentrations.concentrations.

NOx emissions near regions of high population NOx emissions near regions of high population result in greater mortalities.result in greater mortalities.

Controls on temporal and spatial location of NOx Controls on temporal and spatial location of NOx emissions are critical, even within the May-emissions are critical, even within the May-September OSeptember O33 season, to reduce damage season, to reduce damage resulting from increased Oresulting from increased O33 concentrations. concentrations.

Future WorkFuture Work

Estimate economic costs of mortalities and Estimate economic costs of mortalities and morbiditymorbidityEvaluate feasibility of: Evaluate feasibility of: – charging emitters for the damage they cause charging emitters for the damage they cause – adjusting number of pollution permits available in adjusting number of pollution permits available in

locations of high ozone production potential and large locations of high ozone production potential and large population density.population density.

Examine ability of chemical weather forecasting Examine ability of chemical weather forecasting system to predict future damage from emissions system to predict future damage from emissions so that emitters may adjust their emissions to so that emitters may adjust their emissions to minimize their total costs.minimize their total costs.

Evaluating Adverse Health Impacts of Air Evaluating Adverse Health Impacts of Air Pollution in Eastern China-- Pollution in Eastern China--

the Price of Clean Airthe Price of Clean Air

Xiaoping Wang, Denise MauzerallPrinceton University

Yongtao Hu, Armistead RussellGeorgia Institute of Technology

Questions to be addressedQuestions to be addressed

How much health damage does current air pollution How much health damage does current air pollution cause in eastern China based on year 2000 emissions cause in eastern China based on year 2000 emissions using conventional energy technologies? using conventional energy technologies?

To what extent can alternative energy and pollution To what extent can alternative energy and pollution control technologies mitigate air pollution damage to control technologies mitigate air pollution damage to human health? human health?

What are the conditions and constraints for China to What are the conditions and constraints for China to adopt a coal gasification-based energy system in both adopt a coal gasification-based energy system in both the near- and long- term? the near- and long- term?

Integrated Assessment ApproachIntegrated Assessment Approach

Link emissions from a particular energy end-use sector/activity Link emissions from a particular energy end-use sector/activity to the health effects. to the health effects.

Four parts: Four parts:

I. Develop high-resolution regional anthropogenic emission I. Develop high-resolution regional anthropogenic emission inventory inventory

II. Simulate ambient concentrations of PM and gaseous species, II. Simulate ambient concentrations of PM and gaseous species, including secondary PM using Models-3/CMAQincluding secondary PM using Models-3/CMAQ

III. Estimate physical health impacts associated with air pollution III. Estimate physical health impacts associated with air pollution exposureexposureIV. Quantify costs resulting from health effects of ambient air IV. Quantify costs resulting from health effects of ambient air pollution exposurepollution exposure

PART I

Emission Inventory Development

Emission Inventory (1)

8 species(j): CO, NH3, SO2,NOx,PM10,PM2.5,NMVOC,CO2 87 municipalities (k), 72 source categories (l*m*n)

EFAE nmlkjnmlkjnmlkj * ,,,,,,,,,,,, E—emissions A—activity rateEF—emission factor j—species k—municipality l—sector m—fuel or activity typen—abatement technology

Emission Inventory (2)

Emissions by SectorEmissions by Sector

II Model Simulations

Model Boundary and Location of the Case Region

CMAQ domains (solid squares) and MM5 domains (dashed squares)

Core region

Model configurationModel configuration

Map projectionMap projection: Lambert Conformal, central meridian: 116: Lambert Conformal, central meridian: 116ºº, , latitude of origin: 35latitude of origin: 35ºº, 1st standard parallel: 25, 1st standard parallel: 25ºº, 2nd standard , 2nd standard parallel: 47parallel: 47ºº

Model domainsModel domains: two with 12km and 36km resolutions: two with 12km and 36km resolutions

MM5MM5: 54 * 54 grid cells for the 36km domain, 60*72 grid cells : 54 * 54 grid cells for the 36km domain, 60*72 grid cells for the 12km domain, 34 vertical layers from surface to 100mbfor the 12km domain, 34 vertical layers from surface to 100mb

SMOKE and CMAQSMOKE and CMAQ: 48*48 grid cells for the 36km domain, : 48*48 grid cells for the 36km domain, 54*66 grid cells for the 12km domain, 13 vertical layers54*66 grid cells for the 12km domain, 13 vertical layers

Carbon Bond 4 – ae3 – aq chemical mechanism Carbon Bond 4 – ae3 – aq chemical mechanism

Running periods: January, April, July, and October 2-18, 2000.Running periods: January, April, July, and October 2-18, 2000.

Midnight 1pm

1) Mobile

2) Area

3) Total

Emissions 4/8/2000

Air Temp

Wind

ASO4

Midnight 1pm4/8/2000

Comparison of Simulated Concentrations with Observations

PART III Health Impact AnalysisPART III Health Impact Analysis

Methods for Health Impact AnalysisMethods for Health Impact Analysis

cases = Icases = Iref ref * Pop* [exp (* Pop* [exp (**c)-1] c)-1]

Cases = annual change in mortalities or morbidities Cases = annual change in mortalities or morbidities resulting from air pollution exposureresulting from air pollution exposure

IIrefref = annual baseline mortality /morbidity rate = annual baseline mortality /morbidity rate

Pop = size of affected populationPop = size of affected population

= relative risk per unit change in concentrations = relative risk per unit change in concentrations

c = annual change in ambient concentrationsc = annual change in ambient concentrations

Changes in PM2.5 and mortality associated with anthro emissions

Population PM2.5 in 2000

Chronic mortality associated with PM2.5

Domain Population: 281 million

Total PM2.5-related chronic mortality: 840,000 deaths

PM2.5-related chronic mortality by sectorPM2.5-related chronic mortality by sector

Further WorkFurther Work

A.A. Conduct an economic analysisConduct an economic analysis

B.B. Construct two alternative energy technology Construct two alternative energy technology and emission scenarios for the year 2020 in and emission scenarios for the year 2020 in the study regionthe study region

C.C. Conduct a sensitivity analysis for health Conduct a sensitivity analysis for health impacts to various input variables such as impacts to various input variables such as emissions and exposure-response coefficientsemissions and exposure-response coefficients

SummarySummary

Developed an anthropogenic emission inventory for the Developed an anthropogenic emission inventory for the Shandong region of China.Shandong region of China.Conducted MM5/SMOKE/CMAQ simulations and comparison Conducted MM5/SMOKE/CMAQ simulations and comparison of calculated and observed pollutant concentrations for the of calculated and observed pollutant concentrations for the regionregionCalculated health damages associated with the year 2000 Calculated health damages associated with the year 2000 pollution levelspollution levelsDemonstrated the benefit of using an integrated approach for Demonstrated the benefit of using an integrated approach for examining the environmental externalities associated with examining the environmental externalities associated with human activities human activities

Will examine potential gains in health benefits by adopting Will examine potential gains in health benefits by adopting advanced, cleaner energy technologies in 2020advanced, cleaner energy technologies in 2020

Stay Tuned!

An Integrated Assessment Model:Architecture, Development and

Application

Quansong Tong and Denise MauzerallPrinceton University

Robert MendelsohnYale University

Presented at CMAS Models-3 Workshop, RTP, NC, Oct. 28, 2003

Overall Objectives

• Quantify the damage caused by air pollution over the continental U.S.

• Monetize these damages so that their impact can be included in national economic accounting.

• Develop a user-friendly integrated assessment model for community use.

Strategy Design Module

Emission Module

Health ModulePopulation

& Dose Response

Chemistry Transport Module

Meteorology Module

Socio-economic Module

Reanalysis Visualization

Model Framework

Integrated Assessment Model:Model Flow Chart

Strategy

Design

Module

Meteorology

Module

Emission

Module

Chemical

Transport

Module

Health

Module

Economic

Valuation

Module

Input Output

For streamline above,

Provide interface to utilize state-of-the-art models

Develop needed additional modules based on literature review

• Emission Module: SMOKE version 1.5

• Meteorology module: MM5 version 3.5

• Chemical transport module: CMAQ 4.3

• Health module: Under construction

• Economic module: Yale Microeconomics

• Others: SAS/S-Plus, arcGIS, JSP/ Servlet container, etc.

Current Model Components

Health Module

Current version includes an exposure response function for O3;

PM2.5 and SO2 exposure-response functions are being added;Plan to include estimates of visibility losses, material damage, ecosystem losses, and crop losses in future versions.

Chemical Transport Module Output

Population Census

Mortality and

Morbidity

Exposure

Dose Response

GIS Information

32-node Beowulf Linux cluster 2 x 2.4 GHz / 512K cache Xeon; 2GB DDR, 200 MHz RAM 1 TB SCSI hard disks; 1 fast-Ethernet and 2 Gigabit switches

Portland and Intel Fortran/C/C++ compiler

MPICH/PVM compiled with PGF/Intel

openPBS job scheduler

MPIEXEC workload manager

Hardware & Software

Scalability Analysis

436

250

600

430

538

657

736

845

1081

1326

0

200

400

600

800

1000

1200

1400

1 2 3* 4 6 8 10 12 16 20

Number of Processors

Ru

ntim

e (s

ec)

0

5000

10000

15000

20000

25000

30000

Ru

ntim

e x

# o

f C

PU

s

Runtime Runtime x NP

Distributed to 3 nodes

1 node

3 nodes

Integrated Assessment Model

Hardware/Software Approach

Validation

Filter

Jobs Submission

Online Strategy Design

HTTP Server

Master Node

Dynamic Workload Generator

User Database

Scheduler &

Workload Manager

Bulk Computation

Beowulf Cluster

Model

Outpu

t

Analysis Tools

User Database

Intermediate Final Results

Integrated Modeling System

Potential Benefits to the Community

A platform to bridge different scientific disciplines with end-users outsides the community;

Make use of state of the art models provided from different scientific communities;

Provide the full functions to users without locally installing the system.

User friendly and transparent to those not model experts;

Centralized data storage and better utilization of model data;

Enhanced concept of “One-Community” modeling!

Summary / Future DirectionsSummary / Future Directions

We have incorporated the latest research from We have incorporated the latest research from atmospheric science, epidemiology and economics into atmospheric science, epidemiology and economics into an integrated framework useful for informing policy.an integrated framework useful for informing policy.

Plan to make the integrated assessment modeling Plan to make the integrated assessment modeling system more user-friendly and flexible to facilitate system more user-friendly and flexible to facilitate additional applications.additional applications.

Potentially make the model available for use by the Potentially make the model available for use by the community.community.

Challenge is to minimize uncertainties involved in Challenge is to minimize uncertainties involved in individual model components in order to maximize the individual model components in order to maximize the utility of the coupled integrated assessment model.utility of the coupled integrated assessment model.