WHITE PAPER:

INTEGRATION OF AIR QUALITY AND CLIMATE CHANGE -
MODELING CONNECTIONS FROM GLOBAL TO REGIONAL

SCALES

(8 October 2010)

Climate Team of the Atmospheric Modeling and Analysis Division

Atmospheric Modeling and Analysis Division
National Exposure Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711

-------
List of Acronyms

AMAD

Atmospheric Modeling and Analysis Division

AMIP

Atmospheric Model Intercomparison Project

AOGCM

Atmosphere-Ocean General Circulation Model

AR4

(IPCC) Assessment Report 4

AR5

(IPCC) Assessment Report 5

BC

Black Carbon

BEIS

Biogenic Emissions Inventory System

CAM

Community Atmospheric Model

CAP

Criteria Air Pollutant

CCN

Cloud Condensation Nuclei

CCSM

Community Climate System Model

CDC

Centers for Disease Control

CESM

Community Earth System Model

CIRAQ

Climate Impacts on Regional Air Quality

CLM

Community Land Model

CM

(NOAA GFDL) Coupled Model

CMAQ

Community Multi scale Air Quality (Model)

CORDEX

Coordinated Regional Climate Downscaling Experiment

DOE

Department of Energy

DMS

Dimethyl Sulfide

DOD

Department of Defense

EDGAR

Emissions Database for Global Atmospheric Research

EPA

Environmental Protection Agency

ESMF

Earth System Modeling Framework

FORE-SCE

Forecasting Scenarios of Land Use Change

GEOS-Chem

Goddard Earth Observing System - Chemistry (model)

GFDL

Geophysical Fluid Dynamics Laboratory

GHG

Greenhouse Gas

GISS

Goddard Institute for Space Sciences

GLIMPSE

GEOS-Chem LIDORT Integrated with MARKAL for the Purpose of
Scenario Exploration

ITR

Integrated Transdisciplinary Research

IPCC

Intergovernmental Panel on Climate Change

LIDORT

Linearized Discrete Ordinate Radiative Transfer

LSM

Land-Surface Model

LW

Longwave (radiation)

MAGIC C

Model for the Assessment of Greenhouse Gas Induced Climate Change

MARKAL

Market Allocation (model)

MEGAN

Model of Emissions of Gaseous and Aerosols from Nature

MODIS

Moderate Resolution Imaging Spectroradiometer

NARCCAP

North American Regional Climate Change Assessment Program

NARR

North American Regional Reanalysis

NASA

National Aeronautics and Space Administration

-------
NCAR

National Center for Atmospheric Research

NCEP

National Centers for Environmental Prediction

NCER

National Center for Environmental Research

NEI

National Emissions Inventory

NERL

National Exposure Research Laboratory

NLCD

National Land Cover Database

NNRP

NCEP/NCAR Reanalysis Project

NO A A

National Oceanic and Atmospheric Administration

NRMRL

National Risk Management Research Laboratory

NSF

National Science Foundation

OAP

Office of Atmospheric Programs

OAQPS

Office of Air Quality Planning and Standards

OAR

Office of Air and Radiation

ORD

Office of Research and Development

PBL

Planetary Boundary Layer

PM

Particulate Matter

PoBNS

Programs of Broad National Significance

RCCM

Regional Climate Chemistry Model

RCM

Regional Climate Model

RCP

Representative Concentration Pathway

RRTMG

Rapid Radiative Transfer Model for GCMs

SLCF

Short-Lived Climate Forcers

SST

Sea-Surface Temperature

STAR

Science to Achieve Results

SW

Shortwave (radiation)

USD A

United States Department of Agriculture

USGS

United States Geological Survey

VOC

Volatile Organic Compound

WRF

Weather Research and Forecasting (model)

-------
1. Introduction

As part of the EPA Office of Research and Development (ORD), the National Exposure
Research Laboratory (NERL) is in the process of developing Integrated Transdisciplinary
Research (ITR) programs to better address Problems of Broad, National Significance (PoBNS).
These programs are intended to provide strategic research directions in high priority topics that
would benefit from more integrated collaborative research implementation across ORD's
Laboratories and Centers. The nexus between air quality, climate change, and the energy
system, with a focus on environmental impacts and the effects of climate change mitigation and
adaptation strategies, has been identified as one of these ITR programs. The purpose of this
white paper is to establish the roles of AMAD in the emerging research areas of climate change
and environmental impacts by identifying key questions and research approaches, which include
applications as well as model and methods development to address these questions. A focused
research plan for climate and energy will aid AMAD in organizing effective collaborations with
other NERL Divisions, other ORD Laboratories and Centers, EPA program and regional offices,
and external groups to accomplish the required research.

We begin by identifying the environmental problem, outlining other Federally-funded climate
research programs, and setting general goals for AMAD's involvement. Section 2 describes the
science and policy drivers for the Agency that can be broadly divided into two categories: risk
assessment and adaptation, and risk mitigation. Section 3 presents AMAD's conceptual
approach and plans for collaborations, some of which have already been initiated. AMAD's
research approach, presented in Section 4, describes plans for development of modeling tools for
assessment of climate change impacts for the United States and for assessment of mitigation
strategies. Finally, Section 5 is dedicated to describing expected products.

Environmental Problem

Consumption of fossil fuels contributes to global warming and degrades air quality, both of
which profoundly impact human and ecosystem health. Atmospheric levels of carbon dioxide
(CO2) and other greenhouse gases have increased dramatically since the Industrial Revolution,
and emissions from fossil fuel combustion have been linked to human disease since the London
smog event in 1952. The Intergovernmental Panel on Climate Change (IPCC) Fourth
Assessment Report (AR4) (Solomon et al., 2007) concluded that the continued rise in
greenhouse gases from human activity is the primary cause of the temperature increases
observed over the 20th century and that global warming is likely to continue over the next
century even with significant mitigation of greenhouse gas emissions. These temperature
increases also lead to changes in other climatic conditions such as changes in precipitation
intensity and duration. Extreme weather conditions could become more frequent which, in turn,
can adversely affect human and ecosystem health. Anthropogenic combustion also represents
the largest emissions source of primary particulate matter (PM), as well as oxides of nitrogen and
sulfur, which react in the atmosphere to form ozone and secondary PM. These pollutants are
known to contribute to respiratory and cardiovascular effects in human populations as well as
ecological effects to aquatic (acidification and eutrophication) and terrestrial ecosystems
(damage to agricultural and other vegetation).

-4 -

-------
Climate research at other agencies

The Department of Energy (DOE) funds the development of climate models, develops
technologies to reduce emissions from energy systems, and collects data on the Earth's radiation
budget. Specifically, the DOE is "particularly interested in developing models that better define
interactions between climate change and decadal modes of natural climate variability, simulate
climate extremes under a changing climate, and help resolve the uncertainties of the indirect
effects of aerosols on climate."

As a science funding agency, the National Science Foundation (NSF) has many activities
relevant to climate science. Current topics of greatest interest include "developing models that
will produce reliable predictions of 1) climate change at regional and decadal scales; 2) resulting
impacts; and 3) potential adaptations of living systems to these impacts. Related research may,
for example, include studies of natural decadal climate change, regional aspects of water and
nutrient cycling, and methods to test predictions of climate change."

DOE and NSF are the main funders of the National Center for Atmospheric Research (NCAR)
Community Earth System Model (CESM). This model seeks to comprehensively simulate the
complex interplay of physical, chemical, and biological processes of the atmosphere, ocean, and
land surface. The first version was released in June 2010. In order to facilitate participation
from a broad, inter-disciplinary community of scientists, CESM uses the Earth System Modeling
Framework (ESMF), which is designed to support the interconnection of geophysical models.
ESMF is funded by NO A A, NASA, NSF, and DoD.

NOAA also plays an important role in climate science. The Geophysical Fluid Dynamics
Laboratory is developing an Earth System Model, CM3, the Coupled Atmosphere Ocean Land
Ice model version 3. Future plans include "Our vision of global modeling is an integrated ESM
[Earth System Model], projecting not only climate variability on seasonal to centennial
timescales, but also biogeochemical and ecosystem cycling and biospheric feedbacks on the
climate system. This is a comprehensive effort, requiring incorporation of climate dynamics,
ecological processes and human activity." On the regional scale, NOAA's Earth System
Research Laboratory has developed WRF-Chem, a coupled interactive chemistry-meteorology
model. Current efforts are focused on developing WRF-Chem as a regional climate model. In
addition to developing models, NOAA also runs the National Climatic Data Center to archive
weather data.

NASA's contribution to climate science includes the Goddard Institute for Space Studies (GISS)
ModelE, a coupled Atmosphere-Ocean-Land-Ice model. NASA's earth observing satellites are
using remote sensing techniques to track the composition of the atmosphere, understand the
impact of aerosols on clouds, and quantify the fluxes of greenhouse gases.

The Centers for Disease Control (CDC) is using its prevention expertise to anticipate potential
impacts of climate change on human health. Also, the U.S. Department of Agriculture (USDA)
is particularly interested in developing climate models that can be linked to crop, forestry and
livestock models. Such models will be used to help assess possible risk management strategies
and projections of yields at various spatial and temporal scales.

- 5 -

-------
Goals

There is a large established national and international community of scientists who have studied
global climate change issues for many years. AMAD, as a relative newcomer in the climate
science community, must determine which scientific questions that it is uniquely positioned to
help answer and for which it can create a niche in the climate change arena. AMAD's broad
goals include:

•	To lead scientific investigations of the interactions between climate change and air
quality.

•	To be recognized internationally for our contributions to better understanding climate
science on regional scales.

•	To develop partnerships across the Agency to quantify impacts of climate change on
human health, water resources, and ecosystem resources.

To achieve these goals AMAD scientists must become expert users of climate models and
atmospheric chemistry models at both the global and regional scales to perform scientific
investigations as well as assessments. Traditionally AMAD has developed, evaluated, and
applied meteorological and atmospheric chemistry models on local to continental scales. The
climate change research arena requires an expansion of our toolkit to the global scale. AMAD's
primary modeling expertise will remain on the local and regional scales, and AMAD will use
global models primarily to downscale results to finer regional scales to address the Agency's
concerns from health and ecological perspectives. Thus one of AMAD's focal areas will be the
development and comparison of dynamical and statistical downscaling methods from global to
regional climate. AMAD will also develop methods to link global and regional atmospheric
chemical models and will refine the Community Multiscale Air Quality (CMAQ) model for full
hemispheric air quality simulations. AMAD will further develop, test, and refine its integrated
meteorological/air quality model to assess the impacts of the short-lived gas and aerosol species
on direct radiative forcing, cloud radiative properties, and precipitation. Concurrently, AMAD
will develop screening tools to more quickly assess the radiative effects of various greenhouse
gas (GHG) and particulate matter (PM) control scenarios that will be used to select focused
analyses with more comprehensive modeling tools on the principal scenarios of interest from a
climate change perspective. AMAD scientists will perform modeling assessments of the effects
of climate change and related adaptation and mitigation actions on regional and local air quality,
and water availability and quality.

2. Science and Policy Drivers

The scientific issues of global climate change have been studied for decades. However, in
response to the 2007 U.S. Supreme Court decision Massachusetts us. EPA and the recently
finalized Endangerment Finding under the Clean Air Act, the EPA must rapidly prepare for
rulemaking regarding control of greenhouse gas emissions. Additionally, the EPA Office of Air
and Radiation (OAR) recognizes that some "traditional" pollutants regulated under the Clean Air
Act also have radiative forcing properties (e.g., the black carbon component of PM2.5, ozone) that
impact climate. The Agency has a need for integrated policy approaches to both mitigate climate
change and manage air quality.

-6-

-------
The urgency has also greatly increased to understand the implications of climate change on local
and regional scales given the numerous anticipated impacts on environmental protection and
regulatory responsibilities of the Agency. For example, research has already demonstrated that
future climate conditions will very likely increase air quality risks and decrease the effectiveness
of emission control efforts (e.g., Nolte et al., 2008; Weaver et al., 2009). In addition to enhanced
photochemical production in a generally warmer environment, greater frequency of stagnation
events could exacerbate air quality problems. It is also anticipated that increases in extreme
precipitation events could lead to additional water-borne disease outbreaks and degradation of
water quality conditions. Anticipated damages to aquatic and terrestrial ecosystems could
include impacts such as encroachment by invasive species and biome migration.

The Agency's needs fall into two broad questions from the air quality, climate, and energy ITR
plan that will be used to organize and focus the AMAD research activities described in this white
paper:

A.	Risk Assessment and Adaptation: How will climate change impact air and water
quality, water availability, and ecosystems?

AMAD has initiated a series of research investigations that are focused on risk
assessment, specifically asking how climate change might impact air quality in the future.
The first phase of this study, which considered air quality under current emissions with a
future climate, was completed during 2007 (Nolte et al., 2008) and work is underway
with a second phase that considers potential future emission scenarios as well. The latter
effort will contribute to a second assessment of future scenario (both climate and
emissions) impacts on ozone and PM2.5 The primary client for this work is EPA OAR's
Office of Air Quality Planning and Standards (OAQPS) which uses these assessment
results to inform its policy decisions regarding the effectiveness of regulations in light of
future climate change and how to design policies that can adapt to increased air quality
risks from climate change.

In addition to this ongoing work on air quality impacts from climate change, the Agency
has additional risk assessment needs to consider how a changing climate will impact
other regulated endpoints, namely water quality and ecosystem resources. Regional
climate scenarios similar to those used to assess climate impacts on air quality are needed
for these assessments; however, water and ecological assessments have finer spatial scale
requirements that further challenge the uncertainty of regional climate scenarios and how
future temperatures and precipitation may change. The client for assessment information
on these topics is ultimately the EPA Office of Water which includes the Office of
Ground Water and Drinking Water and Office of Wetlands, Oceans, and Watersheds.

B.	Risk Mitigation: How best to contribute to climate change mitigation through
U.S. controls of both long-lived greenhouse gases and short-lived pollutants?

While the central theme of the AMAD climate and air quality research has historically
focused on air quality assessments, the introduction of climate change mitigation into the
Agency mission creates a critical need for decision-making tools that can be used to

-7-

-------
recommend optimal U.S. policy choices that mitigate climate change. Given that some
traditional criteria pollutants have radiative forcing properties that also affect climate, the
optimal policy solutions are likely to be a combination of greenhouse gas emission
reductions and controls on black carbon, ozone, and methane (an ozone precursor and
greenhouse gas itself). If air quality management options are chosen that both improve
air quality and support climate mitigation, an additional benefit is that positive impacts
will be realized for climate trends in the nearer term while greenhouse gas mitigation
impacts on climate will be evident only after several decades due to their much longer
chemical lifetimes (Levy et al., 2008).

To characterize potential climate and air quality impacts from various greenhouse gas and
short-lived radiatively active air pollutants, modeling tools are needed that consider
global impacts from various emission mitigation scenarios and that can translate these
impacts to regional scales. The EPA OAR can use this suite of tools to determine policy
options that improve air quality and help mitigate climate change.

3. Conceptual Approach

To address the key scientific and policy drivers identified above, a system of models is needed
that can assess the regional-scale impacts of future climate change and quantify the effects of
mitigation options. The conceptual approach for such a system of models is shown in Figure 1.
The first step is to develop scenarios that describe the trajectory of emission and land use
changes for a potential future realization or for a given policy option. This includes GHGs,
short-lived climate forcers (SLCF), and criteria air pollutants (CAPs) identified by the Clean Air
Act. On the right side of Figure 1, these emission and land use change scenarios are used to
drive the complex numerical models of global and regional climate. Atmosphere-Ocean General
Circulation Models (AOGCMs) are needed to project the impact of anthropogenic forcing on
global climate. AOGCMs that include online chemistry and aerosol processes are preferred
because they include the impact of SLCF as well as the well-mixed GHGs. To understand the
regional impacts of climate change on finer spatial and temporal scales, "time slices" from the
AOGCM simulations (with higher-temporal frequency output for time periods on order of a
decade) are downscaled using the Weather Research and Forecasting (WRF) model as a regional
climate model. AMAD's goal is to better resolve the impacts of global change on the United
States, particularly changes in average conditions, variability, and extreme events of near-surface
temperature and precipitation. To improve AMAD's understanding of the impacts of SLCF on
regional climate, AMAD will use its two-way coupled WRF-CMAQ model to simulate regional
climate, chemistry, and aerosol processes in an integrated fashion. In partnership with other
scientists within the Agency, the results from AMAD's regional climate simulations will form
the basis for assessments of the impacts of regional climate change on air quality, human health,
water resources, and ecosystems. Examples of climate impacts could include changes in storm
tracks and frequency that increase the frequency of stagnation events and reduce air quality,
early onset of spring that disrupts ecosystems, and extreme precipitation events that lead to
drought or flooding and efficient transfer of atmospheric contaminants to land and water.

- 8 -

-------
The system of complex numerical models in Figure 1 requires considerable computational
resources which limits the number of scenarios that can be realistically examined. Yet, large
uncertainties in future emission projections require consideration of a large number of scenarios.
A parallel set of reduced-form models, on the left side of Figure 1, form a suite of screening tools
that can be used to rapidly assess the impacts of emission changes on global climate change and
regional air quality. These screening tools will be derived using an adjoint version of a global
chemical-transport model, GEOS-Chem, and the regional chemical-transport model, CMAQ.
These can be used in conjunction with the Model for the Assessment of Greenhouse Gas Induced
Climate Change (MAGICC), a tool that approximates the global change in temperature due to a
change in GHG emissions using a statistical parameterization based on the AOGCMs used in the
IPCC AR4.

To address questions relevant to the assessment of climate change risks, AMAD initially will use
a subset of the Representative Concentration Pathway scenarios of the IPCC Fifth Assessment
Report (AR5) to drive the system of complex numerical models in Figure 1 that lead from
emission scenarios to quantitative assessments. To understand the impacts of climate mitigation
policy options for simultaneously addressing climate change and air quality, the screening tools
can be used to identify options that reduce radiative forcing while improving air quality, which
can be developed into more comprehensive scenarios to drive the global and regional climate
models in order to more accurately and credibly quantify the effects of a mitigation action on
human health, air quality, water resources, and ecosystems (see Figure 1).

-9-

-------
Emission & Land Use

Scenarios
Reflecting U.S. Policy
Options for GHGs,
SLCF, and CAPs

Atmosphere-Ocean
General Circulation Model

GISS ModelE & other models
with online chemistry
and aerosols

It

Screening Tools

MAGIC C

GHG
A Global T

Adjoint
GEOS-Chem

SLCF ->
ARad. Forcing

Adjoint
CMAQ

CAPs

AAir Quality

I

GHG & SLCF A Climate
(Temperature, precipitation, etc.)
2° x 2.5° horizontal scale

Weather Research and
Forecasting Model (WRF)

WRF-CMAQ

with online
chemistry
and aerosols

Downscaling to
regional climate
at finer horizontal
scale

Assessments: Human Health, Air Quality,
Water Resources, Ecosystem Impacts

Figure 1: Integrated framework of modeling tools for addressing the key science and policy drivers for
AMAD's climate and energy research.

Collaborators

Constructing a comprehensive system of models and modeling applications requires AMAD to
broaden its skills and capabilities by collaborating and partnering with other parts of ORD,
Federal agencies, and academic researchers. First, AMAD's simulations require the
development of future scenarios of emissions and land use change. AMAD is continuing to
work with the developers of ORD National Risk Management Research Laboratory's (NRMRL)
MARKet ALlocation (MARKAL) energy-economics model to provide consistent scenarios of
both GHGs and SLCF for assessing and understanding the changes in emissions due to policy
actions. An important area of collaboration is to develop scenarios that consistently capture
changes in both emissions and land use in a way that can be included in AMAD's climate
modeling. For non-anthropogenic emissions, the Science to Achieve Results (STAR) grants
program within the EPA National Center for Environmental Research (NCER) has recently
funded several projects to quantify the impacts of climate change on biogenic and geogenic
emission sources such as wind-blown dust, soils, the biosphere, wildfires, and lightning. AMAD
will stay abreast of these continual developments and integrate these approaches in the future as
they become critical to the model assessments and client needs.

- 10 -

-------
Second, to become advanced users of the AOGCMs, AMAD will partner with the developers of
these models. There are three U.S. institutions with AOGCMs that participated in the IPCC
AR4: the National Aeronautics and Space Administration (NASA) Goddard Institute for Space
Studies (GISS), the National Oceanic and Atmospheric Administration (NOAA) Geophysical
Fluid Dynamics Laboratory (GFDL), and the National Center for Atmospheric Research
(NCAR). AMAD has established a formal collaboration with GISS. AMAD has made initial
contact with scientists at GFDL and NCAR, and AMAD is seeking to solidify collaboration in
regional climate research with both groups. It will be important to consider several AOGCMs in
order to increase the robustness of the assessment of regional climate change impacts.

Since regional climate modeling is relatively new, AMAD will leverage its existing
collaborations with the WRF community to adapt and improve regional climate capabilities.

This includes consultation on appropriate physics options, methods for data assimilation, and
integrating chemistry and aerosol processes using WRF-CMAQ.

The theoretical and practical development of adjoint models is also relatively new. To extend
AMAD's capabilities in these areas, AMAD is collaborating with the developers of the CMAQ
adjoint and GEOS-Chem adjoint models with projects jointly funded by EPA and NASA.

Finally, as AMAD develops and refines its regional climate simulations for use in assessments of
climate change impacts to water resources and ecosystems, it is critical to form collaborative
partnerships with the EPA Laboratories and Centers that will be leading these assessments. It
will be important that AMAD's ecosystem research group be involved and work with the
Ecological Services Research Program collaborators within the EPA to better understand how
the regional scenarios are used and what improvements would be most critical to the
assessments. To address impacts of climate change on water quality and quantity, AMAD will
partner with the Laboratories and Centers designated as part of the ORD emerging program on
Sustainable Waters.

AMAD's growing expertise in regional climate modeling can help improve these assessments.
Also, better understanding of the climate metrics most important for critical Agency applications
will help AMAD focus its development efforts and improve our modeling tools. However,
cultivating and sustaining productive collaborations can be time-consuming. Therefore, AMAD
will judiciously pursue collaborative partnerships focusing on research areas that are high-
priority to the Agency.

4. Research Approach

4.1 Assessing Global to Regional Climate Impacts

Climate is typically defined as the long-term average (several decades) of meteorological
conditions in a given location. Climate can change over longer time scales due to changes in the
Earth's radiation budget, known as radiative forcing. These changes can have myriad causes,
including physical factors such as variations in the Earth's orbit or the amount of solar radiation

- 11 -

-------
emitted by the Sun, as well as changes in the chemical composition of the atmosphere. Certain
atmospheric constituents known as GHGs (e.g., water vapor, C02, 03, CH4, and N20) as well as
some aerosol species absorb energy radiated from the Earth's surface and radiate a portion of
that energy back to the surface, resulting in warming of both the Earth's surface and atmosphere.

All state-of-the-science global climate models (GCMs) are coupled AOGCMs. The ocean
components of the models simulate oceanic heat uptake and transport via currents, as well as sea
ice melt and formation. Because the ocean is a vast thermal reservoir compared to the relatively
thin atmosphere, these dynamic processes are critical to simulating climate change over
centuries. For some "time slice" (on the order of a decade) applications, a climatic average
representation of the ocean is used with the atmospheric component, allowing the model to be
run at higher spatial and/or temporal resolution.

An AOGCM simulation must span a long period of time because conclusions drawn based on
too short a simulation may be invalid if the modeling period happens to be anomalous. Because
of the complexity of the physical system being simulated and the requirement for long simulation
periods, computational resource constraints necessitate that AOGCMs be run at relatively coarse
spatial and temporal resolution. The IPCC AR4 AOGCMs had horizontal spatial resolutions
ranging from -1.1° x -1.1° to 4° x 5° (Randall et al., 2007, Table 8.1). These large spatial scales
are insufficient to resolve geographical features, such as mountains and lakes that can have
significant impacts on regional climate. In addition, many applications of interest to the Agency
are episodic in nature, such as the change in frequency and spatial distribution of extreme events,
and a higher temporal resolution is required of climatic fields than is typical of AOGCMs.
Therefore fields from AOGCMs will be inadequate to fully understand the regional climate
change issues that are of interest to the Agency. In response, AMAD will acquire fields from
AOGCMs to drive regional climate simulations to address the Agency's needs and to study the
relevant scientific questions.

Obtaining global model fields

The IPCC AR4 (Solomon et al., 2007) included results from 23 different AOGCMs. Based on
the results presented in AR4, no single model can be considered to be the best for all aspects of
global climate modeling. Though it would be ideal to conduct regional climate modeling studies
using an ensemble of models, as was done in the AR4, this is not feasible with resources
currently available to AMAD. Accordingly, the question arises as to which AOGCM(s) to
choose for regional climate downscaling. The perceived scientific quality of the model and the
modeling group's scientific reputation and publication track record are two important factors for
selection of a particular AOGCM. Other criteria worth considering are the availability and
quality of the model's documentation, the size of the community using the model, and whether
training or help is available for prospective users. It is also desirable to establish collaboration
with the AOGCM developers, in which they conduct the AOGCM simulations and/or provide
training and guidance on how to set up and use the model and interpret model results.

AMAD scientists previously conducted a pilot project, Climate Impact on Regional Air Quality
(CIRAQ), in which partners in academia performed the AOGCM simulation of future global
climate under one future GHG scenario, and collaborators at the Department of Energy's Pacific
Northwest National Laboratory used these AOGCM fields to simulate regional climate over the

- 12 -

-------
continental United States. Currently, collaborators at GISS are conducting future climate
simulations for the IPCC AR5, and they will provide highly time-resolved fields (i.e., time
slices) from those simulations to AMAD for regional climate downscaling. However, ultimately
EPA will need the capability to conduct additional global climate simulations under alternative
GHG emission scenarios. Accordingly, AMAD plans to develop expertise at using AOGCMs,
rather than merely being consumers of the data generated by them.

AOGCMs are typically evaluated for mean temperature and precipitation over large spatial areas
(e.g., continental-scale) and over long time periods (e.g., decades). In order to determine the
validity and to quantify the benefit of downscaling the AOGCM fields with the regional climate
model (RCM), one key challenge for AMAD will be to develop evaluation methods for both
global and regional climate modeling. Of particular interest for regional climate modeling is the
frequency of extreme events such as heat waves and droughts. Careful evaluation of the
performance of the AOGCM under current climate will be necessary to evaluate the efficacy of
the regional climate downscaling. Of particular importance for regional climate modeling are the
AOGCM biases over the United States and the extent to which the AOGCM accurately captures
synoptic circulation and teleconnection patterns reflecting changes in atmospheric waves and the
jet stream.

Assessing Regional Climate Change

AOGCMs are used to predict gridded future climate conditions with world-wide coverage. To
investigate and understand regional effects of climate change (and particularly to capture
extreme events), higher temporal and spatial frequency of the gridded fields is required. One
broad category of methods to create regional climate fields that are influenced by AOGCM
simulations is "downscaling." Downscaling involves using the fields from an AOGCM in
combination with additional information about the regional scale (e.g., topography, land use,
historical regional climate data) and physical, mathematical, and/or statistical models to extend
the AOGCM simulation to finer spatial and temporal granularity. With increased horizontal
texture, the downscaled fields are more sensitive than the AOGCMs to local spatial
heterogeneities in weather and climate that result from topographical changes, land/water
interfaces, vegetation, and population. Downscaled climate fields can then be used to predict the
regional impacts of climate change on water quality and availability, agriculture, ecosystems,
human health, and air quality resulting from emissions control strategies, and energy demand.
Through the CIRAQ project, downscaled meteorological fields were generated by external
collaborators for use in the CMAQ modeling system to assess predicted changes in ozone and
particulate matter resulting from predicted changes in climate at 2050 (Nolte et al., 2008). Based
on the lessons learned from the CIRAQ project and the desire to tailor downscaled fields to
better capture extreme events, AMAD will be leading research to assess regional climate change
based on downscaled AOGCM fields. The AOGCM fields will be captured at a much finer
temporal output increment (e.g., six-hourly) to accommodate regional downscaling methods.
AMAD's primary focus will be on extreme events (e.g., heat waves, droughts, flooding,
stagnation events) and the frequency of such events, in addition to changes in local mean
temperatures and precipitation. For example, increased periods of stagnation during the summer
can adversely affect air quality, which may necessitate additional regulation of air pollutants. In
addition, changes in drought and/or flooding patterns over time can affect water quality and

- 13 -

-------
availability for the general population as well as agriculture. As with all of the research in
AMAD, the assessment of regional climate change using downscaling techniques will have
components of research and development as well as policy-relevant application.

AMAD will investigate the use of both dynamical and statistical downscaling techniques for
regional downscaling of AOGCM simulations. Both dynamical and statistical downscaling are
somewhat small but growing research areas. Initial research in AMAD for dynamical and
statistical downscaling will follow parallel paths to develop a greater understanding of the
published research with each technique thus far, to determine which methodologies (if any) are
appropriate to AMAD's research goals, and to extend and/or refine those techniques to
downscale AOGCM predictions. The first AOGCM fields that will be used to evaluate
downscaling techniques will be from the IPCC AR5 fields from the NASA GISS ModelE.
Follow-on work in downscaling will be conducted using fields from the NOAA GFDL Coupled
Model (CM3) and the NCAR Community Climate System Model (CCSM) or the Community
Earth System Model (CESM) when collaborations with those research groups are formalized and
as those AOGCM fields become available. Additional partnerships will be established in the
regional climate modeling community, as appropriate, to leverage and improve the most current
scientific approaches in regional climate modeling.

Dynamical Downscaling

Regional climate change simulations are critical for impact assessment because AOGCMs are
computationally too expensive for long-term high resolution simulations. Regional climate
models (RCMs) can be used to investigate environmental issues that are important at regional
and local scales. RCMs are high-resolution numerical models that can be used to provide
detailed representation of physical processes in response to complex topography, land-sea
contrasts, and sharp land-use gradients (Dickinson et al., 1989; Giorgi and Bates, 1989). Unlike
AOGCM simulations where the non-linear atmospheric processes are sensitive to perturbations
in the initial conditions and it is possible for the models to diverge after only a few days of
simulation, the RCM is constrained by the lateral boundary forcing. The RCM's simulated
climate represents a synthesis of the global climate information provided at the lateral boundaries
and the physical and dynamical processes in the RCM. The ultimate goal is to improve the
representation of regional climate by improving physics and/or increasing model resolution.
However, error can be introduced at the RCM lateral boundaries due to errors in the AOGCM
fields or inconsistencies in the atmospheric circulations suggested by the RCM and the AOGCM.
It is difficult to distinguish the errors related to physics and resolution from those arising at the
lateral boundaries, but it has been demonstrated that a common problem in using RCMs is that
the simulated state tends to deviate from the driving state at large scales (von Storch et al., 2000;
Giorgi and Bi, 2000). Thus an active area of research for AMAD will be to try to establish
regional climate modeling techniques that will moderately constrain the RCM to the AOGCM's
forcing while enabling the RCM to develop the finer-scale physical and dynamical processes.

For AMAD's dynamical downscaling, the WRF model will be used initially. Dynamical
downscaling presents several technical challenges to define a credible methodology. The
following list includes several of the technical decisions related to WRF model simulations that
need to be addressed and justified before moving forward.

Domain definitions (sizes, spatial extents, etc.)

Horizontal grid spacing (108-km, 36-km, 12-km, other?)

- 14 -

-------
•	Nesting options (if any, e.g., 108-36 km one-way, no feedback; 108-36 km two-way, with
feedback)

Vertical layers (number and specification),

Model top (100 hPa, 50 hPa, or further into the stratosphere?)

Physics options

Four-dimensional data assimilation - Nudging (if, where, how, and how strong)

Length of spin-up time to achieve physical and dynamical balance
Initialization of specific fields (e.g., soil moisture and temperature)

•	Level of constraint of the AOGCM forcing on the WRF simulation

How to evaluate downscaled runs to develop downscaling methodology with confidence in
future climate scenarios

How to characterize and quantify the uncertainty in the WRF simulations

As the first step in the development and evaluation of a dynamical downscaling methodology, a
series of regional climate simulations will be conducted using present-day global reanalysis
fields at a comparable spatial and temporal resolution to the AOGCMs that AMAD will acquire
via partnerships. For example, the 2.5° x 2.5° version of the NCEP/DOE Atmospheric Model
Intercomparison Project (AMIP) Reanalysis 2 (R-2) data sets will be used in the WRF model as
initial and lateral and surface boundary conditions to mimic AOGCM fields. The WRF model
will then be run as an RCM (i.e., continuously, without updating the simulation with additional
observational knowledge of the atmosphere ) to downscale the R-2. The R-2-WRF simulations
will be run using multiple physics, nudging, and gridding configurations with the goal of
selecting a primary dynamical downscaling configuration. The resultant WRF model
simulations will then be compared with the higher resolution (i.e., 32-km) North American
Regional Reanalysis (NARR) data set to evaluate the efficacy of the dynamical downscaling
techniques. Some creativity must be employed in the evaluation methods such that the focus is
on evaluating climate rather than weather. Though it is acknowledged that success with an
R-2/WRF dynamical downscaling technique may not be analogous to coupling an unrelated
AOGCM with WRF, this approach to evaluating and establishing the validity of the dynamical
downscaling techniques is widely accepted in the literature (e.g. Xue et al., 2007; Gustafson and
Leung, 2007).

Concurrent with the R-2/WRF evaluations, an initial dynamical downscaling linkage will be
established using the NASA GISS ModelE and WRF. A one-year data set is used to examine the
initial annual cycle fields, and that data set will be expanded to include multiple consecutive
years and future time periods. As NASA GISS develops ModelE fields for the IPCC AR5, those
data will be made available at high temporal frequency (i.e., six-hourly) for "time slice intervals"
of the AOGCM simulation. Using knowledge gained from the R-2/WRF testing, a base
dynamical downscaling technique will be employed to create regional downscaled climate fields.

As part of the efforts with the reanalysis fields and the AOGCM fields, dynamical downscaling
work will examine more closely the method and strength of constraining the RCM (i.e., WRF)
toward the AOGCM. Specifically, both spectral (Waldron et al., 1996) and analysis nudging
(Stauffer and Seaman, 1990) will be tested to determine the suitability of each method to force
the RCM to adopt the prescribed large scales over the entire domain. Analysis nudging
approximates the fields at each grid point while spectral nudging uses the large-scale waves of

- 15 -

-------
the driving field to keep the interior model solution consistent with the driving fields. Interior
nudging is a "hot" research topic in the regional climate modeling community with great
potential for AMAD to lead the science. Two scientific questions with respect to nudging will
help guide our experiments. Rockel et al. (2008) identified one of the higher-order questions:
Should interior nudging be usedfor AOGCM lateral boundary forcing driven simulations, since
real-world observational constraints on the simulations are absent in the AOGCM unlike
reanalysis driven simulations? The second question is derived from in-house expertise on the
importance of the implementation of nudging: What is the relative contribution of the nudging
coefficients to the regional climate modeling simulations? It will be important to balance the
benefits of using nudging techniques to improve the RCM simulations with the potential
unintended consequence of reducing the variability (and extremes) in the solution. In addition,
special attention will be given to the nudging techniques for aerosol-radiation-cloud-chemistry
interactions in the integrated regional climate-chemistry modeling system (see Section 4.2.2),
which could be very sensitive to small changes in the atmosphere. Evaluation techniques such as
spectral power (Castro et al., 2005) and isotropic digital filter (Feser and von Storch, 2005) will
be applied to understand if the RCM is improving the regional-scale variability. Lessons learned
from these simulations will then be used to help determine the robustness of the sensitivity using
longer-term integrations (-ten years). A WRF simulation from the North American Regional
Climate Change Assessment Program (NARCCAP) with no nudging may serve as a baseline for
comparison of added value of the nudging when using the same physics options. AMAD will
also strive to participate and contribute to regional climate modeling intercomparison activities
such as NARCCAP and the Coordinated Regional Climate Downscaling Experiment
(CORDEX).

A major area of interest for regional climate modeling is the ability of the RCM to simulate
physical processes important for climate, including radiation, cloud cover, and surface heat and
moisture fluxes. Biases in WRF's representation of these physical processes may have a larger
effect on long-term simulated climate than would be evident from integrations conducted for a
shorter period or when nudging is used to constrain the RCM to observations. The ability to
simulate these physical processes more accurately will be critical for understanding the coupled
climate-air quality system.

One initial area of focus in this area will be analyzing certain of the radiation schemes
implemented in WRF. Although the radiation schemes in WRF have been adapted from global
climate models and are tuned for the total global forcing, they may need to be improved for
regional climate modeling purposes. In addition, the treatment of partially cloudy conditions
may need to be adjusted within the radiation schemes. A research activity will be conducted to
evaluate the most advanced radiation schemes in WRF against satellite-based observations to
determine the strengths and target areas of improvement in those schemes.

Another area of interest in defining the dynamical downscaling technique is the selection of the
land-surface model (LSM) for the RCM. The existing LSM options in the WRF system are
designed for short-term forecasting or retrospective simulation. They are intended to model the
effects of vegetation and soil conditions on the atmosphere through parameterizations of surface
fluxes including soil evaporation, evapotranspiration, and evaporation from wet canopies.
Multilayer soil moisture is a key parameter for many of these processes. Thus, some of these
LSMs rely on sophisticated land data assimilation systems to initialize soil moisture using a
variety of observations such as solar radiation, precipitation, ground level temperature and

- 16 -

-------
humidity, etc. Alternatively, soil moisture can be nudged through an indirect data assimilation
scheme according to model biases in ground-level temperature and moisture compared to
observation based analyses. In either case, extensive use of observed data is needed to
frequently update soil moisture conditions because these LSMs do not include the complete
hydrological cycle and therefore are incapable of long-term simulation without observational
constraint.

For climate modeling, more comprehensive land models are necessary that are designed for long-
term simulation without the need for observation-based data assimilation. For example, the
Community Land Model (CLM) (Bonan et al., 2002), which has been developed as part of the
Community Climate System Model (CCSM), has comprehensive treatment of the physical,
chemical, and biological processes by which terrestrial ecosystems affect and are affected by
climate across a variety of spatial and temporal scales. The CLM includes detailed
representations of processes such as interception, throughfall, canopy drip, snow accumulation
and ablation, infiltration, surface and sub-surface runoff, soil moisture, and surface moisture
fluxes by canopy evaporation, transpiration, and soil evaporation. Thus, the CLM or similar
climate land models should be adapted for use in the RCM system. A partnership with modelers
at the University of California at Berkeley, who have developed a preliminary version of CLM
coupled with WRF, has been informally established to pursue this research activity.

It is anticipated that the dynamical downscaling will have a continuously evolving research
component. Several aspects of the dynamical downscaling can be expected to be refined as the
climate program grows. For example, there will be opportunities to improve the LSM
component in WRF to add capabilities and/or tailor the model for climate modeling (either by
improving the coupling with specific AOGCMs or by employing generic methods of
scientifically credible linkage with any AOGCM). In addition, nudging methodologies (which
are somewhat controversial in the dynamical downscaling community) can be further developed
and refined for AMAD's dynamical downscaling with a goal of extending techniques back to the
retrospective WRF-CMAQ simulations that are central to the regulatory development efforts in
AMAD. Furthermore, developing evaluation techniques and the quantification of uncertainty
may become an active area of research.

Statistical Downscaling

Statistical downscaling methods use correlations among observed and modeled meteorological
variables to predict regional and/or local patterns and events that are likely to occur based on the
broader-scale AOGCM simulations. Typically, these approaches do not use the same detailed
information that is used in dynamical downscaling, such as physical equations, orographic data,
or extensive land-use information. The advantages of statistical downscaling methods lie in their
efficiency and speed, and these methods could be particularly attractive if numerous climate
scenarios need to be investigated. Statistical methods are not limited by the resolution
achievable by the nested regional dynamical model. Thus, statistical methods could possibly be
used to gain a better understanding of fine-scale variability, even down to point locations.

It has been reported in the literature that the performances of dynamical and statistical
downscaling are comparable for current climatic conditions. However, it is questionable whether
statistical models can perform as well under future conditions (Wilby et al., 2002) because
statistical downscaling methods rely on associations among meteorological variables. These
relationships do not explain all of the inherent variability in atmospheric phenomena; in fact, the

- 17 -

-------
choice of variables to be used as the "predictors" in such approaches is a difficult part of the
statistical downscaling process. Also, once a statistical model has been developed for a
particular time period (e.g., using current climate), it is unclear whether the relationships it
incorporates will remain the same under different climatic conditions (e.g., in future decades).
While this issue of stationarity is discussed by Schmith (2008) in some detail, it is rarely
addressed in individual applications, leaving some doubt as to how accurate the results of
statistical downscaling will be for future time periods.

AMAD began its work in statistical downscaling by investigating the techniques typically used
in the literature. Many of these are regression-based, modeling either relationships between
modeled and observed variables (e.g. Wilby et al, 2002; Spak et al, 2007) or spatial relationships
between fine-scale data and its coarse-scale aggregates (e.g. Hoar and Nychka, 2008). AMAD is
in the process of investigating problematic issues associated with statistical downscaling, such as
violations of the stationarity assumption, systematic differences (biases) between models, and
calibration of statistical estimates of uncertainty. Based on these results, AMAD will determine
if it is reasonable to utilize statistical downscaling techniques and will identify the situations in
which these can be utilized most efficiently and accurately. Particular areas of application may
include (again, depending on the research findings previously mentioned):

Implementing appropriate statistical downscaling approaches using the AOGCM fields

Continuing to improve our understanding of how uncertainty affects estimates, and
particularly how the uncertainty may change when applied to future-year AOGCM
simulations.

Evaluating the performance of statistical downscaling methods in estimating the frequency,
duration, and/or intensity of extreme meteorological events.

Identifying the relative strengths/weaknesses of the dynamical and statistical approaches to
downscaling.

Determining whether hybrid downscaling approaches may be able to capitalize on the
strengths of both methods.

4.2 Modeling tools for mitigation of climate change

4.2.1 Global Climate and Chemistry Modeling

For our assessment work described in Section 4.1, the AOGCMs will be driven with the
Representative Concentration Pathways that prescribe changes to national-level GHG and SLCF
emissions. A critical question is what further climate change mitigation is possible from U.S.
controls on SLCF? In particular, our goal is to determine the relative impact of emission
reductions of SLCF compared to GHGs.

Using GISS ModelE, we will perform a series of century-long sensitivity studies varying the
emissions of SLCF and GHGs. We will quantify the timing and impact of these emissions to
global climate change and from the present to 2100. With this understanding, we will work with
our colleagues in NRMRL to develop U.S. emission scenarios that properly account for the
technology options and economic interactions between sectors, such as electricity generation,

- 18 -

-------
industrial activity, and transportation. The product from this work will be a series of policy-
relevant climate mitigation scenarios, a series of global model simulations, and a description of
appropriate analysis methods to quantify the impacts of SLCF and GHGs emission reductions.

To fully understand the air quality, water quality, human health, and ecosystem impacts of these
climate mitigation efforts, the results of these global modeling simulations can be downscaled as
described in Section 4.1. The next section describes further regional modeling development
efforts to better understand the impacts of SLCF on regional climate.

4.2.2 Regional Climate Chemistry Modeling

While the primary agents of climate change are the long-lived greenhouse gases (GHG), short-
lived gases and aerosol likely play an important role. The IPCC AR4 Report (Solomon et al.,
2007) estimated that tropospheric ozone and the direct and indirect effects of aerosols contribute
significant portions of the mean global radiative forcing. The AR4 also suggests that there is a
large uncertainty associated with these forcings and that their spatial scales are continental to
global. The major sources of this uncertainty are related to inaccurate characterization of
atmospheric loading of aerosols, their chemical composition, and source attribution, all of which
are highly variable both spatially and temporally. Accurate characterization of the spatial
heterogeneity in aerosol composition and size distribution is critical for estimating their optical
and radiative properties and thus in quantifying their impacts on radiation budgets of the earth-
atmosphere system. Large amounts of the atmospheric loading of these constituents are created
by anthropogenic emissions of precursors and complex atmospheric photochemistry. Although
globally much of the secondary aerosols and ozone results from biogenic emissions of volatile
organic compounds (VOCs) such as isoprene and terpenes, the productivity of these "natural"
emissions is greatly enhanced by the anthropogenic emissions that create a more oxidizing
environment (Kanakidou et al., 2005, Matsunaga et al., 2005, Guenther et al., 2006). There are
also potentially important interactions between emissions of photochemically active species such
as NOx and CO with long-lived GHGs such as CH4. Furthermore, because short-lived species
respond more quickly to controls than long-lived GHGs, they may represent the most effective
way to alter trends in global warming in the near future (Jacobson, 2002). Thus, current and
future emissions of air pollutants that lead to elevated levels of aerosols and ozone must be
considered by policy makers for mitigation strategies for both climate change and air quality.
Linkage of climate models with atmospheric chemistry models that include detailed treatment of
emissions, photochemistry, aerosol microphysics, atmospheric transport, and wet and dry
deposition, such as are used for air quality research and impact assessment, are needed to fully
understand the interactions between GHGs and short-lived gases and aerosols.

The WRF and CMAQ models have been integrated into a single executable computer program
with 2-way meteorological and chemical data exchange. Direct effects of aerosols on shortwave
radiation and tropospheric ozone on longwave radiation have already been included while
implementation of the indirect effects of aerosols on the microphysical and radiative properties
of clouds is underway. Direct effects of aerosols on LW radiation, especially BC which is a
strong absorber in both SW and LW parts of the spectrum, is also being developed for
implementation in WRF-CMAQ. The 2-way coupled WRF-CMAQ model can be configured for

- 19 -

-------
use as a Regional Climate Chemistry Model (RCCM) with a coarse resolution hemispheric grid
domain (AX ~ 50-100 km) and higher resolution nests over North America (AX ~ 12-36 km).
The hypothesis is that by coupling a Regional Climate Model (RCM), such as WRF, modified as
described in Section 4.1 to be more suitable for climate application, to a comprehensive
atmospheric chemistry and transport model (CMAQ), regional climate modeling can be
improved because of the inclusion of the radiative forcing of short-lived gases and aerosols and
the effects of aerosols on cloud microphysics, radiation properties, and precipitation.
Conceptually, by superimposing the regional aerosol radiative effects on the large scale GHG
forcing, the coupled WRF-CMAQ system on a hemispheric domain driven by AOGCM time-
slices can provide process-level insights into the spatial heterogeneity in radiative forcing, their
regional climate effects, and their aggregate influence on the earth's radiation budget. An
important component of this demonstration is the verification of the simulated radiative effects
for retrospective periods. Once the WRF-CMAQ is established as a credible RCCM by assessing
retrospective simulations of a series of nested domains downscaled from global reanalysis data, it
can be used to investigate many issues related to the interaction of climate and chemistry.

Questions to be addressed using the WRF-CMAQ RCCM:

•	How important is radiative forcing from short-lived gases and aerosols relative to GHG
forcing, particularly on the regional scale?

•	How important is the characterization of the spatial heterogeneity in radiative forcing of
short-lived species and how does this heterogeneity influence regional climate
variability?

•	What is the sensitivity of regional climate to changes in biogenic and anthropogenic
emissions?

•	What are the regional climate effects of Asian air pollution in Asia and in North
America?

•	Can we recognize the climate effects of the dramatic increase of Asian air pollution
associated with the economic boom from 1990s to 2000s in our model simulations?

•	How will climate change affect biogenic emissions and how will the resulting change in
gases and aerosols further affect climate?

•	How will future emission change scenarios by region and sector affect near future climate
change?

Development needed to use WRF-CMAQ as an RCCM

Further development of the WRF-CMAQ system will be needed to establish its capabilities for
hemispheric to mesoscale atmospheric chemistry and transport modeling. The expansion to
hemispheric scales will necessitate the expansion of the chemistry represented in the model and
also the examination of the suitability of existing chemical schemes for the free troposphere.
The domain expansion will also necessitate a closer examination of processes in the marine
boundary layer since large portions of the modeled troposphere will be over marine
environments. For instance dimethyl sulfide (DMS) is an important marine source of
atmospheric sulfate. Efforts are already underway to improve the representation of O3 gradients
in the mid-upper troposphere using a potential vorticity scaling and this will be further tested in
the hemispheric applications. On a global basis, O3 is formed predominantly over the continental
regions and is typically lost in marine regions through photolysis. Further reductions in the
tropospheric 03 burden through bromine and iodine emitted from open oceans has also been

-20 -

-------
recently postulated. Another area for further improvement will be the inclusion of emissions
from aircraft, lightning, and fugitive dust outbreaks.

The direct effects of aerosols on shortwave radiation and the direct effects of tropospheric ozone
on longwave (LW) radiation have been implemented in two of the radiation schemes available in
WRF that were designed for climate modeling: the Community Atmosphere Model radiation
scheme (CAM) (Collins, et al., 2004) and the Rapid Radiative Transfer Model for GCMs
(RRTMG) (Iacono et al., 2008). A new Mie scattering algorithm for a wider range of
wavelengths including LW has been developed and implemented in the RRTMG scheme. New
mixing state treatments for aerosols containing black carbon and other constituents such as
sulfate and organic carbon are also being developed and tested. New model simulations of the 2-
way WRF-CMAQ using the latest versions of both models have been evaluated for a summer
month in the eastern US and an outbreak of wildfires in California in 2008. Comparisons
between runs with and without direct feedbacks show significant impacts on solar radiation, 2-m
temperature, PBL height, and ozone and PM2.5 concentrations, especially in areas affected by
smoke plumes. For example, the reduced SW radiation caused much cooler surface air
temperatures (up to 4-5 K in the smoke plumes), lower PBL heights (up to 400 m), and
significantly higher ground level concentrations of ozone and PM2.5 (Mathur et al., 2009).

The 2-way WRF-CMAQ model also includes an experimental implementation of indirect effects
where aerosols from CMAQ are activated as cloud condensation nuclei which determine the
droplet number concentration for a 2-moment cloud microphysics model. The resulting effective
droplet radius is used in the radiation model to compute cloud optical properties. The indirect
effects are being tested by evaluation of cloud radiative forcing compared to satellite
measurements.

Climate and Chemistry Downscaling

Many of the same developments needed to configure WRF for use as an RCM (Section 4.1) will
be needed to configure the WRF-CMAQ as an RCCM. The main difference between the
downscaling described above and the WRF-CMAQ modeling described here is that it will
include a hemispheric domain on a polar stereographic grid as an intermediate step between the
AOGCM and the regional model. There are advantages to this method for both the
meteorological downscaling from the AOGCM and chemical downscaling from the global
chemistry model. The meteorology on the hemispheric domain can be made to closely replicate
the AOGCM (albeit at higher resolution) through spectral and/or grid nudging, sea surface
temperature (SST), sea ice, and snow fields from the AOGCM. In addition, much of the physics
in WRF can be harmonized with the AOGCM. For example, we have been using the CAM
radiation scheme which is also used in NCAR's Community Climate System Model (CCSM)
and recently we have implemented aerosol feedback effects in the new RRTMG radiation
scheme which is being used in the the Community Atmospheric Model (CAM5), which is the
latest atmospheric component of the Community Earth System Model (CESM1). Also, we plan
to use the Community Land Model (CLM), which is a component of the CESM1, as a land
surface model in WRF. A general problem with regional downscaling from an AOGCM is that
it is often ambiguous whether the differences between the higher resolution regional model and

-21 -

-------
the global model are caused by inconsistencies in physics, dynamics, grid structure, and
numerical techniques or the result of greater resolution. Thus, by configuring WRF with
compatible physics and using data assimilation and surface forcing from the AOGCM, and
confirming similar model results on the hemispheric domain, we can have more confidence that
results from the regional nested domains reflect more the benefits of higher resolution and are
less artifacts of model inconsistencies.

Evaluation

Much of the evaluation of the WRF-CMAQ when applied as an integrated regional climate
chemistry model would be similar to the evaluation steps of a downscaled RCM as described
above in Section 4.1. For example, an important step in the evaluation of either system would be
retrospective experiments where the regional model is applied to a historic period driven by
global reanalysis data and observed datasets for SST and sea ice (Randall et al., 2007). Evaluation
by comparison to observations and high resolution observation-based analyses would effectively
assess the model's ability to accurately downscale from observation-based global information, as
a surrogate for an essentially perfect AOGCM. Sensitivity studies where various feedback
effects are turned off or emission sectors are zeroed out could be used to investigate the
importance of the short-lived gas and aerosol feedbacks to the regional simulations. The role of
the hemispheric domain could also be assessed through comparison to RCM simulations that are
downscaled directly from global to regional domains. The next phase in evaluation of any
downscaling model system should be to replicate the historic simulations described above using
an AOGCM model rather than global reanalysis. These experiments should address the
question: Are the downscaling model capabilities sufficiently retained when driven by an
AOGCM?

As a development and evaluation study, the 2-way coupled WRF-CMAQ modeling system will
be used to simulate the effects of changing anthropogenic emissions over the past two decades
on spatial and temporal variability in tropospheric aerosol loading and resultant radiative forcing
over the continental United States. Measurements of ambient aerosols and radiation will be
analyzed before and after the implementation of the Title IV Clean Air Act controls (SO2), from
early 1990s to the present, to discern trends in radiative forcing (brightening) resulting from
reductions in sulfate aerosol. The 2-way coupled WRF-CMAQ model should be able to replicate
the observed trends in aerosol concentrations and radiative forcing over this period. Thus, this
will be a critical test of the integrity of the model's chemistry and transport, aerosol chemistry
and dynamics, and direct and indirect radiation feedback effects.

The synergistic applications of global-scale and regional-scale chemistry-climate models provide
unique opportunities to inter-compare the relative merits of the two approaches and to address
current uncertainties in climate science. For instance, comparison of simulated radiative forcing
from short-lived species for selected regions by the two approaches, with appropriate
measurements, would provide valuable guidance on the level of spatial heterogeneity that must
be resolved to adequately represent regional climate variability (e.g., effects of resolving aerosol
composition and size distributions at finer resolutions). Analyses and inter-comparison of results
could also help develop region-specific metrics to assess impacts of climate change on various
environmental endpoints (e.g., changes in hurricane intensity in the Atlantic, changes in
precipitation in the Pearl River Delta) and air quality-climate interactions (e.g., modulation of
arctic snow albedo due to black carbon deposition and its source attribution).

-22 -

-------
4.2.3 Decision support tools for scenario assessment and comparison

The impacts of climate change accumulate over many decades. Emissions, land use change, and
other forcing agents on the climate system cannot be reliably forecast over these time scales.
The analyses of the Intergovernmental Panel on Climate Change have instead relied on
scenarios, rough sketches of potential futures. From scenarios that lead to mitigated climate
change, we can learn more about what emission reduction goals are needed. Scenarios also
include assumptions about future changes that can only be partially controlled by policy options,
such as availability of natural resources, the pace of technological change, and global population
migration. Robust policy options are those that achieve our climate mitigation goals across a
wide range of assumptions about the future. A key need is a set of modeling tools that can
rapidly screen large numbers of scenarios to uncover robust policy options, isolate determining
assumptions, and present trade-offs to decision makers.

Process-based, first-principle models alone are not ideally suited for this task. Such models are
comprehensive but computationally expensive. They do, however, strive to accurately capture
the complex, non-linear relationship between anthropogenic activities and the environmental
harms. Climate models and chemical-transport models are no exception.

However, some first-principle models have been augmented to include on-line sensitivity
calculations. Adjoint models allow the user to define a receptor and to calculate the sensitivity
of the receptor to each model input. For example, an adjoint chemical transport model can
calculate the influence of each emission source in the domain on the aerosol concentration in one
non-attainment area. In adjoint models, the receptor is defined by a cost function. This function
can be as simple as the concentration in a single location, or it can be arbitrarily complex, such as
the contribution to locations and times that exceed a concentration threshold.

Model development

Our current model development activities are focused on two different adjoint models. First, we
are focusing on ensuring that the CMAQ Adjoint (Hakami et al., 2007) contains the most up to
date, state of the science modules as the current released version of CMAQ. Second, along with
our collaborators, we are developing a GEOS-Chem Adjoint (Henze et al., 2009) that includes an
adjoint of the radiative transfer model LIDORT (Spurr et al., 2001). GEOS-Chem is a global
chemical transport model (http://acmg.seas.harvard.edu/geos/index.html). In addition to
calculating the sensitivity of concentrations, this model can also calculate the sensitivity of direct
radiative forcing due to emission changes. Currently, this is only implemented for shortwave
radiation and black carbon; our development activities will extend this to scattering aerosols,
ozone, and longwave radiation.

Model application

Using results from these models, we will develop screening tools to rapidly calculate the impact
of a given emission reduction strategy on both air quality (for example, PM concentration in non-

-23 -

-------
attainment areas) as well as global or regional change in radiative forcing. We will work closely
with our colleagues in NRMRL to ensure the development of economic emission models that are
sufficiently comprehensive to explore the relationships between different policy options. We
will also work closely with our colleagues in EPA OAQPS and OAP Climate Change Division to
ensure the metrics calculated by these adjoint models are relevant to the environmental indicators
that the Agency needs to assess the success of regulatory actions.

It is critical to note that the efforts described here are to complement, not to replace the
physically based models. We need continued improvement in the physically based models to
ensure that the adjoint is correctly representing the interactions between the processes. The
reduced-form model is intended only to be a screening tool for rapid scenario discovery. Change
in global radiative forcing is a useful indicator of anthropogenic impact of climate change, but it
is not sufficient to understand the regional-scale impacts such as extreme heat events and
variable precipitation that causes droughts and flooding. The goal of the adjoint models is to
illuminate the coarse differences between scenarios, understand the impacts of key assumptions
that drive future outcomes, and discover policy options that are robust across these assumptions.
Once a subset of robust policy options is identified, this more computationally tractable subset of
scenarios must be assessed using the more complete, physically based models described in the
other sections of this white paper including the global chemistry and climate models and the
regional WRF-CMAQ system both for downscaling of global climate and for further assessment
of regional changes in radiative forcing and air quality.

4.2.4 Emissions and Land Cover

Land Cover

Land cover changes are interrelated with many of the anthropogenic activities affecting
emissions to the air, all of which interact as a part of climate change. A combination of
enhanced information based on current sources and new land cover and emission modeling
projection tools will be required as a part of the development of modeling regional climate
change and effects. Land model components of AOGCM systems, such as the Community Land
Model (CLM) (Gibbard et al., 2005; Lawrence and Chase, 2007), typically characterize
vegetation types according to global land-use data such as the MODIS satellite land cover
product (available at approximately 1 km spatial resolution). While 1 km spatial resolution is
more than sufficient for the grid resolution of current global models, our regional modeling
system (WRF-CMAQ) has benefitted from the greater resolution and accuracy of the National
Land Cover Database (NLCD) (Fry et al., 2009) which has 30 m spatial resolution for the U.S.
based on Landsat Thematic Mapper imagery. This high resolution allows for much more
accurate classification of land use categories and better accounting of subgrid spatial
heterogeneity. Outside the U.S., MODIS land cover data (Giri et al., 2005) and a new global
landuse database available at 300 meter resolution known as the GlobCover data have recently
been made available from the European Space Agency (Arino et al., 2007). However, the
classification system of the MODIS land cover data is more compatible with NLCD data.
Adaptation of the CLM to use these higher resolution landuse databases, which would require
significant code modification in collaboration with the CLM developers at NCAR, would result
in consistent modeling of dynamic natural land surface processes, chemical surface fluxes (dry
deposition and bidirectional surface fluxes), and biogenic emissions in our WRF-CMAQ RCCM.

-24 -

-------
Projection of land cover for regional climate modeling will require development and application
of modeling tools. There are some promising tools that can be built upon in a collaborative
effort. On a global scale, the CLM has the advantage of being actively supported by many
contributors for use with global climate models. As a dynamic model, it accounts for the major
biophysical processes and feedbacks affecting vegetation and hydrology looking forwards many
hundreds of years.

Scenarios of projected regional land cover change must also be compatible with scenarios of
anthropogenic emissions as well as vegetation cover change. On a global scale, the IPCC is
assessing the differences in land cover use and projections applied by the four different
Representative Concentration Pathway (RCP) emission scenario teams for the fifth assessment
report (AR5) by means of analysis of the effects of the differences (Hurtt et al., 2009). The
USGS Forecasting Scenarios of Land Use Change (FORE-SCE) model (Sohl et al, 2007)
presents a promising approach for scenario-based land cover projections to approximately 2050,
with stochastic components using NLCD-scale land cover in conjunction with econometric
model drivers (for example MARKAL, below). FORE-SCE is usually applied on the basis of
hydrologic basins, which is consistent with the needs of ecological programs. The spatial scale
of application is variable from NLCD-pixels to potentially national scale. Thus far it has been
applied to entire regions such as the U.S. Southeast and high plains (Sohl and Sayler, 2008).
Substantial work is required to statistically define the landscape drivers for the balance of the
United States, in order to apply FORE-SCE nationally. To take full advantage of both FORE-
SCE and CLM for future periods of approximately 50 years or more, it will be necessary to
establish a comparison between their land cover projections at specified future times. It is not
expected that they will provide the same results, but rather different realizations reflecting their
different strengths. Initially, collaboration between USGS, NCAR and EPA will be pursued to
result in a mutually-agreed description of a common baseline of assumptions and time periods
that can be used for model comparisons. Since FORE-SCE is the shorter term and more
anthropogenically-focused model, a next step would be to nudge aspects of CLM toward FORE-
SCE land cover at approximately 2050, project further (say 2100) with CLM and compare with
CLM results without the nudging.

Emissions

In order to project anthropogenic emissions, either on a global or regional scale, sound global
and regional inventories are required. The most complete existing global emission inventories
for greenhouse gases and air pollutants are maintained online through the Emission Database for
Global Atmospheric Research (EDGAR) system (http://www.mnp.nl/edgar/). The inventories
are the result of compiling national inventories focused on the year 2000, including those of the
United States and Canada, in addition to new emission estimate contributions from individual
investigators. Currently, efforts are underway to make 0.5 x 0.5 degree resolution the common
spatial resolution of EDGAR data sets. Enhanced EDGAR data sets are the basic components of
the 2000 reference year being used for the IPCC emission inventory, again at a spatial resolution
of 0.5 x 0.5 degrees (Lamarque et al., 2009). The base IPCC emission data sets are posted on the
internet at ftp://-ipcc.fz-iuelich.de/pub/emissions. The IPCC data sets should be spatially
resolved enough to form the basis for regional climate modeling at hemispheric or continental
scales. For purposes of more detailed regional climate modeling of non-greenhouse gases

-25 -

-------
within North America, more spatially and temporally resolved emission inventories than the
global AR5 inventories may be required. It will be necessary to substitute the county-level EPA
National Emission Inventory (NEI) for EDGAR data, to address the relatively short-lived criteria
pollutants for the United States and Canada for the base periods. This is not to prejudge the AR5
scenarios as "truth", only for inter-study comparison. EPA will require additional scenarios to
reflect plausible policy directions and additional science questions.

Prioritized future years and policy-relevant scenarios of interest will be selected in conjunction
with EPA clients. The number of scenarios examined will be limited by the computational
resources and program time limitations. A prioritized list of future scenarios will be developed
by NRMRL in collaboration with AMAD using the GEOS-Chem LIDORT Integrated with
MARKAL for the Purpose of Scenario Exploration (GLIMPSE) screening tool (under
development). The MARKet Allocation (MARKAL) model (Fishbone and Ablilock, 1981), an
optimizing econometric energy use and production model, has been applied regionally by the
EPA for carbon dioxide, sulfur dioxide, nitrogen dioxide, and PM10. For comparison with AR5
global emission scenarios, the inputs to MARKAL could be scaled to the hemispheric and
continental scale assumptions used in the AR5 scenarios. There will also be regional scale
scenarios that will vary within the bounding influences of the AR5 scenarios, or be independent
from them. Developments to extend the range of gaseous and particulate emission species and
economic sectors addressed by MARKAL are underway, with the lead group being NRMRL.
For example, addition of volatile organic compound species, methane, ammonia and carbon
monoxide are now planned, as well as increased spatial resolution beyond the nine U.S. census
divisions now addressed. Spatially dispersed or "nonpoint" sources require indirect data or
"surrogates" to spatially allocate their emissions from geographic reporting units such as
counties to model grid cells. This has historically been done using detailed surrogate data such
as census and land cover information. For future years, additional work is needed to refine initial
work to allocate projected emissions spatially using relatively simplified projected information,
such as population and land cover combinations. For anthropogenic emission projections
beyond 2050 to 2100, research is needed to determine the best approach. Input data for
MARKAL and modeling assumptions are much less reliable after 2050. Alternate viable
emission projection methods might range from scaling the IPCC AR5 scenarios to extended
MARKAL assumptions from 2050 based on more simplified assumptions, or some combination
of approaches. It is also likely that additional approaches may be gleaned from the developing
literature in the next two to three years.

Sources of naturally occurring emissions, including biogenic emissions, fluxes from oceans,
lightning, wildfires, and anthropogenic biomass burning, are areas requiring additional
information, to be gained primarily from EPA grant program results and current research
updating CMAQ, with the effects on regional climate and feedbacks evaluated in AMAD.
Emissions from these sources are currently roughly estimated in the EDGAR global data sets
used by AR5, except for biomass burning, which is being addressed by a composite of different
models (Lamarque, 2009). However, additional spatial and temporal information is required for
regional scale climate modeling. Inclusion of feedback effects as a part of defining and
modeling regional emission scenarios is increasingly important for these sources. Biogenic
emissions may be addressed by ongoing development of the NCAR-supported Model for
Emissions of Gases and Aerosols in Nature (MEGAN) model (Guenther et al., 2006), rather than

-26 -

-------
the Biogenic Emission Inventory System (BEIS) model currently used in EPA emission
modeling (Schwede et al, 2005). MEGAN is being used as a part of global climate modeling and
is currently being evaluated by AMAD. There are ongoing discussions with NCAR on reaching
consensus on grouping of forest species in the land cover database used by MEGAN. AMAD is
evaluating the effect on national and regional climate emissions and feedbacks relative to the
older BEIS model results and different climate scenarios.

Fluxes of important chemical substances from the ocean, such as mercury (Hg), dimethyl sulfide
(DMS), and ammonia (NH3) are subject to substantial uncertainty, both in magnitude and sign
(Liss and Lovelock, 2007; Sunderland et al., 2009; Johnson and Bell, 2009). This is an area of
ongoing research from which AMAD will need to glean measurement information from other
researchers and collaborate on modeling the fluxes. Emissions of nitrogen species from
lightning are possibly a major contributor to nitrogen species emissions globally (Bond et al.,
2002). This is the subject of ongoing collaborative work between AMAD investigators and
NASA. Because climate change is likely to affect the amount of convective activity in the
atmosphere, lightning events will also vary and generate more or less emissions (Reeve and
Toumi, 1999). Modeling of emissions from lightning will need to be coupled with the modeling
of storm convection within the regional climate applications of WRF and CMAQ. The modeling
of future wildfire and biomass burning emissions is related to the modeling of future lightning
events in conjunction with regional climate, and regional land cover scenario conditions (Brown
et al., 2004; Ramanathan et al., 2001).

5. Research Plan

The research projects described in Section 4 cover a wide range of approaches to climate related
issues including the development and application of a suite of tools that can be used to determine
the impacts of U.S. climate mitigation policies. These tools span global and regional scales as
well as short-term and century-scale temporal impacts. This section first outlines a roadmap for
the development of a Global-to-Local scale climate and chemistry modeling system that will be
essential for understanding the regional and local impacts of climate change and the relative
importance of GHGs and SLCF. Next, milestones for anticipated products for the next five years
are listed.

5.1 Development Plan for Regional Climate Chemistry System

There are four concurrent research projects that are building the components needed for the
development of a comprehensive Global-to-Local climate chemistry model system.

These are:

1.	Global climate model simulations with online chemistry and aerosols for various
emission scenarios including time slices with high time resolution output (6 hourly) for
downscaling. Initially, we will use the GISS Model-E in collaboration with NASA GISS
with the goal of also working with the NCAR CESM and the GFDL AM3.

2.	Downscaling regional climate modeling using WRF. This project is focused on testing
and evaluating data assimilation techniques and physics components for optimal regional
climate representation.

-27 -

-------
3.	Development of the WRF-CMAQ coupled meteorology-chemistry model. This project
includes the development and implementation of direct and indirect aerosol and short
lived gas effects on radiation and cloud microphysics.

4.	Hemispheric modeling using WRF and WRF-CMAQ. Hemispheric applications of WRF
and WRF-CMAQ will provide a bridge between the global models and the regional and
local scale models thereby improving downscaling fidelity by use of nested grids within a
consistent modeling framework.

Although, each of these four projects have their own research goals, some of which are related to
other environmental issues, they are also necessary steps toward the development of the RCCM
system. We envision that these efforts will be integrated in a staged approach such that the
methods can be thoroughly evaluated for value added at each step.

The developmental stages are:

1.	Regional Downscaling - AOGCM to WRF

Experiments and evaluation of nudging and physics options.

2.	Add hemispheric WRF to downscaling experiments

Does the hemispheric domain improve regional downscaling results?

3.	Use WRF-CMAQ for all grids

Assess impact of chemistry feedbacks at all scales.

This approach will lead to a multiscale integrated regional climate chemistry model capable of
much more highly resolved treatment of emissions, photochemistry, aerosol dynamics, and
interactions between short-lived gases and aerosols, and radiation and cloud microphysics, than
are possible in the coarse grid AOGCMs. A schematic of the model system is shown Figure 2
with anticipated horizontal grid resolution ranges for each WRF-CMAQ model domain.

-28 -

-------
Emissions Scenarios
Reflecting Policy

options for
GHG, SLCF, and
CAP

Contir

le

kX

ntal WRF-C
= 12-36 krr

IV

1AQ







Local/Regional WRF-CMAQ
AX = 4-12 km

Figure 2: Global-to-Local climate chemistry modeling system. The blue arrows indicate the downscaling of
meteorological information while the gold arrows indicate chemical downscaling.

5.2 Anticipated Products

2010

•	Develop advanced methods for downscaling global climate simulations to the regional
scale

•	The 2-way coupled WRF-CMAQ model will be tested with a preliminary indirect aerosol
feedback capability

2011

•	Downscale AR5 climate simulations to create regional climate simulations that can be
used to assess the impacts of regional climate change on human health, air quality, water
resources, and ecosystems

•	The 2-way coupled WRF-CMAQ model will be publically released

AOGCM

NASA GISS Model-E
CCSM, GFDL
With online chemistry and aerosols

I

r





Hemispheric WRF-CMAQ
AX = 50 - 100 km

-29 -

-------
•	Installation of GlobCover 300 m land cover data for use with the Community Land
Model and for regional climate modeling (CMAQ) on a hemispheric and continental
basis. This would require mapping of the land cover classification between that used by
NLCD and GlobCover. Classification mapping by 2011.

2012

•	Development of GEOS-Chem Adjoint for assessing the impact of emission changes on
global radiative forcing due to ozone and aerosols

•	Development of decision support tools using adjoint model results for rapidly screening
scenarios, understanding key assumptions, and discovering robust policy options for both
mitigating climate change and managing air quality

•	An analysis of policy-relevant climate mitigation scenarios, focusing on the timing and
impact of climate change mitigation due to emission reductions of SLCF compared to
GHGs

•	The 2-way coupled WRF-CMAQ model will demonstrated for regional climate
assessment

•	Improved modeling of biogenic and geogenic emission sources including lightning,
wildfires, biomass burning, and wind-blown dust emissions within CMAQ - hemispheric
and/or regional - with method tested and documented

•	Tune and apply the FORE-SCE land cover projection model for the United States, couple
with MARKAL as an economic driver, and compare with CLM projections in
collaboration with the USGS. FORE-SCE addresses anthropogenic land cover drivers as
well as vegetation changes. This would be accomplished for two or three time frames,
with the latest being 2050. Multiple climate scenarios could be run using the land cover
data set for each time period. Initial methodological comparisons documented.

2013-2014

•	Experiments will be conducted using the 2-way coupled WRF-CMAQ model, driven by
AOGCM time slice data, using a series of nested grids from hemispheric to mesoscale to
assess relative impact of SLCF and GHGs on regional climate

•	Installation of GlobCover 300 m land cover data for use with the Community Land
Model and for regional climate modeling (CMAQ) on a hemispheric and continental
basis - Installation and testing in climate and air quality model components (would need
to be done with NCAR involvement)

•	Tune and apply the FORE-SCE land cover projection model for the United States, couple
with MARKAL as an economic driver, and compare with CLM projections in

-30-

-------
collaboration with the USGS - harmonized system producing land cover results usable in
the regional climate model tested and documented.

•	Expand the pollutants, source categories and spatial resolution (to state-level) addressed
by MARKAL to include additional criteria and greenhouse substances. NRMRL has
begun this effort, and they would do most of work on it.- project completion and testing
in phases as sources and pollutants added - some by 2011, the balance of greenhouse
pollutants by 2013.

•	In collaboration with USGS and NCAR, compare FORE-SCE and CLM land cover
projects for agreed time period (e.g. 2050). Document similarities and differences. Use
FORE-SCE projected land cover to update projected CLM land cover and then project
land cover further in time (2100) using CLM. These results would be analyzed in
comparison to CLM land cover not informed by FORE-SCE.

•	Procedures for development of new future emission scenarios must be developed and
tested. For the period between now and 2050, AMAD will collaborate with USGS to
develop MARKAL based scenarios with a set of base scenarios tuned to the new AR5
continental or national projections. MARKAL drivers will be varied to determine how
results might vary from the AR5 projections. For the time beyond 2050, different
approaches will be developed; possibilities include a model other than MARKAL - an
updated MARKAL - or simply percentage variations of the AR5 assumptions. Updating
MARKAL to match AR5 will occur by 2013. New approaches for beyond 2050 will be
documented by 2014.

6. References

Arino, O., Leroy, M., Renera, F., Gross, D. Bicheron, P., Nino, F., Brockman, C., Defournay, P.,
Vancutsem, C., Achard, F., Durieux, L., Bourg, L., Latham, J., DiGregorio, A., Witt, R.,
Herold, M., Sambale, J., Plummer, S., Weber, J-L., Goryl, P., and Houghton, N. The
Globcover Initiative, Proceedings of the Envisat Symposium, 2007, Montreaux, Switzerland,
April 23-27, 2007, European Space Agency report SP- 616, July 2007.

Bonan, G.B., K.W. Oleson, M. Vertenstein, and S. Levis, "The land surface climatology of the
Community Land Model coupled to the NCAR Community Land Model" Journal of
Climate, Vol. 15, pp. 3123-3149, Nov. 15, 2002.

Bond, D.W., S. Steiger, R. Zhang, X Tie, and R.E. Orville, The importance of NOx production
by lightning in the tropics, Atmospheric Environment, Vol 36, pp. 1509-1519, 2002.

Brown, T.J., B.L. Hall, and A.L. Westerling, The impact of twenty-first century climate change
on wildland fire danger in the western United States: an application perspective, Climatic
Change, Vol 62, pp. 365-388, 2004.

Castro, P.G., A. F. Heavens, and T. D. Kitching, 2005: Weak lensing analysis in three
dimensions. Phys. Rev. D, DOI: 10.1103/PhysRevD.72.023516

Collins, W.D. et al., 2004: Description of the NCAR Community Atmosphere Model (CAM3.0),
NCAR Technical Note, NCAR/TN-464+STR, 226pp.

Dickinson, R. E., R. M. Errico, F. Giorgi, and G. T. Bates, 1989: A regional climate model for
the western United States. Climatic Change, 15 (3), 383-422.

-31 -

-------
Feser, F., and H. von Storch, A spatial two-dimensional discrete filter for limited-area-model
evaluation purposes, Monthly Weather Review, Vol. 133, 1774-1786, 2005.

Fishbone, L.G. and Ablilock, H., MARKAL: A linear-programming model for energy-systems
analysis: technical description of the BNL version, Journal of Energy Research, Vol. 5, pp.
353-375, 1981.

Fry, J. A., Coan, M. J., Meyer, D.K., and Wickham, J.D., Completion of the National Land Cover
Database (NLCD) 1992-2001 Land Cover Change Retrofit product, U.S. Geological Survey,
Open-File Report 2008-1379, 18 pp, 2009.

Gibbard, S., Caldiera, K., Bala, G., Phillips, and Wickett, M., Climate effects of global land
cover change, Geophysical Research Letters, Vol. 32, L23705,doi:10.1029/2005GL024550,
4pp., 2005.

Giorgi, F., and G. T. Bates, 1989: The climatological skill of a regional model over complex
terrain. Mon. Wea. Rev., 117, 2325+.

Giorgi, F. and X. Bi, 2000: A study of internal variability of a regional climate model. J.
Geophys. Res., 105, 29 503-29 522.

Giri, C., Zhu, Z, and Reed, B., A comparative analysis of the Global Land Cover 2000 and
MODIS land cover data sets, Remote Sensing of Environment, Vol 94, No. 1, pp. 123-132,
2005.

Guenther, A., T. Karl, P. Harley, C. Wiedinmyer, P.I. Palmer and C. Geron, Estimates of global
terrestrial isoprene emissions using MEGAN (Model of Emissions of Gases and Aerosols
from Nature), Atmospheric Chemistry and Physics, Vol, 6, pp. 3181-3210, 2006.

Gustafson, W., Jr., and L. R. Leung, 2007, Regional downscaling for air quality assessment,

Bull. Am. Meteorol. Soc., 88, 1215-1227.

Hakami, A., Henze, D.K., Seinfeld, J.H., Singh, K., Sandu, A., Kim, S., Byun, D., Li, Q. The
Adjoint of CMAQ. Environmental Science and Technology 41 (22), pp 7807-7817, 2007.

Henze, D.K., Shindell, D.T., Spurr, R. J. D. Attributing direct radiative forcing to specific

emissions using adjoint sensitivities, GEOS-Chem User's Meeting, Cambridge, MA, April 7-
10, 2009 (available at http://acmg.seas.harvard.edu/geos/meetings/2009/ppt/Tuesday/
T ueDAerosol s2_henze_ 1 mac. pdf)

Hoar, T., D. Nychka, 2008. Statistical downscaling of the Community Climate System Model
(CCSM) monthly temperature and precipitation projections, (white paper available at
http://www.gisclimatechange.org/Downscaling.pdf)

Hurtt, G.C., Chini, L.P., Frolking, S., Betts, R., Feddema, J., Fischer, G., Goldesijk, K.K.,
Hibbard, K., Janetos, A., Jones, C., Kindermann, G., Kinoshita, T., Riahi, K., Shevliakova,
E., Smith, S., Stehfest, E., Thomson, A., Thornton, P. VanVuuren, D. and Wang, P. W.
Hamonisation of global land-use scenarios for the period 1500-2100 for IPCC-AR5, ILEAPS
Newsletter, 7, pp. 6-8, June 2009.

Iacono, M.J., J.S. Delamere, E.J. Mlawer, M.W. Shephard, S.A. Clough, and W.D. Collins,
Radiative forcing by long-lived greenhouse gases: Calculations with the AER radiative
transfer models, J. Geophys. Res., 113, D13103, doi:10.1029/2008JD009944, 2008.

Jacobson, M. Z., 2002: Control of fossil-fuel particulate black carbon and organic matter,

possibly the most effective method of slowing global warming, J. Geophys. Res., 107 (D19),
4410, doi: 10.1029/2001JD001376.

Johnson, M. and T. Bell, Concept: Coupling between DMS emissions and the ocean-atmosphere
exchange of ammonia, Environmental Chemistry, Vol 5, no. 4, pp. 259-267, 2009.

-32-

-------
Kanakidou, M., Seinfeld, J.H., Pandis, S.N., Barnes, I., Dentener, F.J., Facchini, M.C., Van

Dingenen, R., Ervens, B., Nenes, A., Nielsen, C.J., Swietlicki, E., Putaud, J.P., Balkanski, Y.,
Fuzzi, S., Horth, J., Moortgat, G.K., Winterhalter, R., Myhre, C.E.L., Tsigaridis, K., Vignati,
E., Stephanou, E.G., Wilson, J., 2005: Organic aerosol and global climate modeling: a
review. Atmospheric Chemistry and Physics, 5, 1053-1123.

Lamarque, J.F., C. Granier, T. Bond, V. Eyring, A Heil, M. Kainuma, D. Lee, C. Liousse, A.
Mieville, K Riahi, M. Schultz, S. Smith, E. Stehfest, D. Stevenson, A. Thomson, J.
VanAadenne, and D. VanVuuren, Gridded emissions in support of IPCC AR5, IGACtivities,
Issue No. 41, pp. 12-22, May 2009.

Lawrence, P.J. and Chase, T.N., Representing a MODIS consistent land surface in the

Community Land Model (CLM3.0), Journal of Geophysical Researach, Vol 112, G01023,
doi: 10.1029/2006JG000168, 17 pp, 2007.

Levy II, H., D. Shindell, A.B. Gilliland, M.D Schwarzkopf, L.W. Horowitz, T.Wigley, R.

Stouffer, and A. Waple, 2008: Climate Projections Based on Emission Scenarios for Long-
lived and Short-lived Radiatively Active Gases and Aerosols, U.S. Climate Change Science
Program Synthesis and Assessment Product (SAP) 3.2.

Liss, P.S. and J.E. Lovelock, Climate change: the effect of DMS emissions, Environmental
Chemistry, Vol. 4, pp. 377-378, 2007.

Mathur, R., J. Pleim, D. Wong, T. Otte, R. Gilliam, S. Roselle, J. Young, F. Binkowski, and A.
Xiu, 2009. The WRF-CMAQ Integrated On-line Modeling System: Development, Testing,
and Initial Applications, D.G. Steyn and S.T. Rao (eds), Air Pollution Modeling and Its
Application XX, Springer, 155-159.

Matsunaga, S., Wiedinmyer, C., Guenther, A., Orlando, J., Karl, T., Toohey, D.W., Greenberg,
J.P., Kajii, Y., 2005: Isoprene oxidation products are a significant atmospheric aerosol
component. Atmospheric Chemistry and Physics Discussions, 5, 11143-11156.

Nolte, C. G., A. B. Gilliland, C. Hogrefe, and L. J. Mickley, 2008: Linking global to regional
models to assess future climate impacts on surface ozone levels in the United States. Journal
of Geophysical Research, 113, D14307.

Ramanathan, V., P.J. Crutzen, J.T. Kiehl, and D. Rosenfeld, Aerosols, climate, and the
hydrological cycle, Science, Vol, 294, no. 5549, pp. 2119-2124, December 7, 2001.

Randall, D.A., R.A. Wood, S. Bony, R. Colman, T. Fichefet, J. Fyfe, V. Kattsov, A. Pitman, J.
Shukla, J. Srinivasan, R.J. Stouffer, A. Sumi andK.E. Taylor, 2007: Cilmate Models and
Their Evaluation. In: Climate Change 2007: The Physical Science Basis. Contribution of
Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on
Climate Change [Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt,
M.Tignor and H.L. Miller (eds.)]. Cambridge University Press, Cambridge, United Kingdom
and New York, NY, USA.

Reeve, N. and R. Toumi, Lightning activity as an indicator of climate change, Quarterly Journal
of the Royal Meteorological Society, Vol. 125, pp. 893-903, 1999.

Rockel, H., C. L. Castro, R. A. Pielke, Sr., H. von Storch, and G. Leoncini, 2008: Assessment of
model system dependent retained and added variability for two different regional models. J.
Geophys. Res., 113, D21107.

Schmith, T., 2008. Stationarity of regression relationships: Application to empirical
downscaling. Journal of Climate, 21, 4529-4537.

Schwede, D., Pouliot, G. and Pierce, T., Changes to the Biogenic Emissions Inventory System
Version 3 (BEIS3), Fourth Community Modeling and Analysis System (CMAS) Conference,

-33 -

-------
Chapel Hill, NC, University of North Carolina, October 2005, 6 pp. Available on-line at:
http;//www.cmascenter.org/conference/2005/abstracts/2_7.pdf

Sohl, T.L. and Sayler, K.L., Using the FORE-SCE model to project land-cover change in the
southeastern United States, 219, pp. 49-65, 2008, doi:10f 1016/j.ecolmodel.2008.08.003.

Sohl, T.L., Sayler, K.L., Drummond, M.A. and Loveland, T.R., The FORE-SCE model: a
practical approach for projecting land cover change using scenario-based modeling, Journal
of Land Use Science, Vol 2, No. 2, pp. 103-126, 2007, doi: 10:1080/17474230701218202.a

Solomon, S., D. Qin, M. Manning, R.B. Alley, T. Berntsen, N.L. Bindoff, Z. Chen, A.

Chidthaisong, J.M. Gregory, G.C. Hegerl, M. Heimann, B. Hewitson, B.J. Hoskins, F. Joos,
J. Jouzel, V. Kattsov, U. Lohmann, T. Matsuno, M. Molina, N. Nicholls, J. Overpeck, G.
Raga, V. Ramaswamy, J. Ren, M. Rusticucci, R. Somerville, T.F. Stocker, P. Whetton, R.A.
Wood and D. Wratt, 2007: Technical Summary. In: Climate Change 2007: The Physical
Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the
Intergovernmental Panel on Climate Change [Solomon, S., D. Qin, M. Manning, Z. Chen, M.
Marquis, K.B. Averyt, M. Tignor and H.L. Miller (eds.)]. Cambridge University Press,
Cambridge, United Kingdom and New York, NY, USA.

Spak, S., T. Holloway, B. Lynn, R. Goldberg, 2007. A comparison of statistical and dynamical
downscaling for surface temperature in North America. Journal of Geophysical Research,
112, D08101.

Spurr, R.J.D., Kurosu, T.P., Chance, K.V. A linearized discrete ordinate radiative transfer model
for atmospheric remote-sensing retrieval. Journal of Quantitative Spectroscopy & Radiative
Transfer 68 689-735, 2001.

Stauffer, D. R., and N.L. Seaman, 1990: Use of four-dimensional data assimilation in a limited-
area mesoscale model. Parti: Experiments with synoptic-scale data. Mon. Wea. Rev., 118,
1250-1277.

Sunderland, E.M., D.P. Krabbenhoft, J.W. Moreau, S.A. Strode, and W.M.Landing, Mercury
sources, distribution and bioavailability in the North Pacific Ocean - Insights from data and
models, Global Biogeochemical Cycles, 2009.

von Storch, H.. H. Langenberg, and F. Feser, 2000: A spectral nudging technique for dynamical
downscaling purposes. Mon. Wea. Rev., 128,3664-3673.

Waldron, K. M., J. Paegle, and J. D. Horel, 1996: Sensitivity of a spectrally filtered and nudged
limited-area model to outer model options. Mon. Wea. Rev., 124, 529-547.

Weaver, C. P., X.-Z. Liang, J. Zhu, P.J. Adams, P. Amar, J. Avise, M. Caughey, J. Chen, R.C.
Cohen, E. Cooter, J.P. Dawson, R. Gilliam, A. Gilliland, A.H. Goldstein, A. Grambsch, A.
Guenther, W.I. Gustafson, R.A. Harley, S. He, B. Hemming, C. Hogrefe, H.-C. Huang, S.W.
Hunt, D.J. Jacob, P.L. Kinney, K. Kunkel, J.-F. Lamarque, B. Lamb, N.K. Larkin, L.R.
Leung, K.-J. Liao, J.-T. Lin, B.H. Lynn, K. Manomaiphiboon, C. Mass, D. McKenzie, L.J.
Mickley, S.M. O'Neill, C. Nolte, S.N. Pandis, P.N. Racherla, C. Rosenzweig, A.G. Russell,
E. Salathe, A.L. Steiner, E. Tagaris, Z. Tao, S. Tonse, C. Wiedinmyer, A. Williams, D.A.
Winner, J.-H. Woo, S. Wu, D.J. Wuebbles, 2009: A preliminary synthesis of modeled
climate change impacts on U.S. regional ozone concentrations. Bull. Amer. Meteor. Soc., 90,
1843-1863.

Wilby, R. L., C. W. Dawson, and E. M. Barrow, 2002. SDSM - A decision support tool for the
assessment of regional climate change impacts. Environmental Modelling & Software, 17,
147-159.

-34-

-------
Xue, Y., R. Vasic, Z. Janjic, F. Mesinger, K. E. Mitchell, 2007: Assessment of Dynamic

Downscaling of the Continental U.S. Regional Climate Using the Eta/SSiB Regional Climate
Model, J. Climate, 20, 4172-4193

-35 -

-------