Transcript WCAIfinal - Weather and Climate Impacts Assessment Science

NCAR Initiative on Weather and Climate
Assessment Science
Linda O. Mearns
Doug Nychka and Jerry Meehl
(acting co-directors)
Overarching Goals of Assessment Initiative
•
Develop new programs that address research gaps
in Impact Assessment Science and that leverage
expertise at NCAR.
•
Emphasize themes that integrate research across
NCAR divisions.
•
Foster NCAR’s leadership role in the IPCC and
other national and international efforts related to
assessment.
•
Create feedback between the impact and
assessment communities and the geophysical
modeling programs
Main Initiative Themes
• Characterizing Uncertainty in Assessment
Work
• Modeling and Assessment of Extreme
Events
• Establishment of a Climate/Human Health
Program
Initiative Management
L. Mearns, Director
G.Meehl and D. Nychka
Acting
Co-directors
Advisory Board
Initiative
Representation
B. Harriss
W. Washington
V. Holzhauer
Administrator
Mentors
G. Bonan, B. Brown, R. Katz, K. Miller, R. Morss, T. Wigley
Cyber/infrastructure, Biogeosciences, Data Assimilation,
GIS, Water Cycle, Wildfire.
Characterizing Uncertainty in Assessment Work
“To know one’s ignorance is the best part of knowledge”
- Lao Tzu
“Doubt is not a pleasant condition, but certainty is an
absurd one”
-Voltaire
Projections of Future Climate
Uncertainty due to spatial scale
T
??P
T
?P
T
P
Temperature
Precipitation
Total Cumulative Carbon Dioxide Emissions (GtC)
3000
2500
A1F1
2000
High > 1800 GtC
A2
Medium High 1450-1800 GtC
A1B
1500
Medium Low 1100-1450 GtC
1000
B2
A1T
Low < 1100 GtC
B1
500
0
1990
IS92
Range
2000
2010
2020
2030
2040
2050
2060
2070
Cumulative Emission 1990-2100, CtC
2080
2090
2100
A Cascade of Uncertainty for Climate Change Research
Characterizing Uncertainty in Assessment Work
• Climate projections and scenarios
• Emissions and land processes
• Impacts models
(e.g., agriculture and ecosystem models)
• Environmental data sets
(e.g., climate observations, climate proxy data, soils)
• Uncertainty and decision making
Uncertainty and Decision Making
(some details)
• Role of different types of uncertainty in different phases
of the policy decision process.
• Uncertainty and multiple decision makers in resource
management.
• Understanding policy makers' needs for quantification of
uncertainty and adapting the analysis of climate
projections and scenarios to address these needs.
• Economic value of reducing uncertainty in weather and
climate information.
Extreme Events
“Man can believe
the impossible.
But man can
never believe the
improbable.”
- Oscar Wilde
Fort Collins Flood, July 1997
Heaviest rains ever documented over an urbanized area in
Colorado (10 inches in 6 hours).
5 dead, 54 injured, 200 homes destroyed, 1,500 structures
damaged.
These locations were not in 500-yr floodplain.
Weather and Climate Extremes
Atmospheric Processes
Modeling of Extremes
Extremes toolkit
Trends in
Observations
Weather
Impacts and Vulnerabilities
Extreme Value Methodology
Climate
Change
Extreme Events
Integrate different aspects of research in
weather and climate extremes:
• Atmospheric science (processes and modeling)
• Statistical aspects of extremes
• Societal impacts and vulnerability
Atmospheric Processes and Modeling of Extremes
Modeling of extremes

Regional climate and mesoscale model validation how well models reproduce extremes.

Spatial scaling of extremes - point versus area
average.

Projections of changes in extremes with climate
change.
Research in the physical processes of extremes
(e.g., warm season heavy rainfall – requires partnership with
Water Cycle Initiative)
Application of Extreme Value Methodology
Normal
Max 100 Normals
• Analysis of weather and climate variables in
terms of tail events and their properties at
different spatial scales.
• Trend analysis of extremes (e.g., temperature
and precipitation).
• Spatial dependence of extreme events.
Societal Impacts of and Vulnerability
to Extremes
• Identification of extremes significant to society
• Modeling the impacts of extreme events
• Tools to reduce societal vulnerability to extremes
Understanding vulnerability requires
knowledge of the behavior and interactions
of all systems involved in an extreme event
e.g., town
economics
storm
meteorology
flood
hydrology
Climate/Health Program
Interdisciplinary Community Integrated Approach
Physical
Data Rescue
and Archive
Biological
• Hydrology
• Oceanograph
y
• Climatology
Program
Vulnerability
and
Risk
Assessment
Analytical Studies
Social
• Public
Health
• Demography
• Economics
• Sociology
•
•
•
•
Design
Analysis
Ecology
Entomology
Microbiology
Mammalogy
Modeling
and
Prediction
Observation
Monitoring
Surveillance
Initiative Highlights
Uncertainty
• Uncertainty due to land cover changes
• Sensitivity and scaling of climate model
results
• Combining multi-model ensembles
Extremes
• Change in frost days in climate projections
• 24-hour precipitation extremes and flood
planning
(also a use of the Extremes Toolkit)
Land Cover Forcing from SRES
Scenarios in Climate Models
How do changes in land use and land cover
alter climate projections?
NCAR Team: G. Bonan, L. Mearns, J. Meehl, K. Oleson
External Collaborators: J. Feddema, U. of Kansas,
R. Leemans, M. Schaeffer, RIVM, Netherlands
Potential
(or2.2Natural)
Vegetation
IMAGE
- 1970 Potential
Vegetation
Impact of Croplands on
Climate
IMAGE 2.2
- 1970Land
Land Cover
IMAGE
1970
Cover
5 - Regrowth (tim ber)
11 - Tem perate M ixed F ores t
17 - S avan na
0 - Ocean
6 - Ice
12 - Tem perate Decid Forest
18 - Tropical W oodland
1 - Agri cult ure
7 - Tundra
13 - W arm M ixe d Forest
19 - Tropical Fores t
2 - Exte ns ive grass land
8 - Wooded Tundra
14 - G ras s/S tep pe
No D ata
3 - C plantation - NU
9 - Bo real Fo re st
15 - D esert
4 - Regrowth (aban don)
10 - C ool C o nifer
16 - S crubland
IMAGE 2.2 Land Cover Types
Model Experiments
• Multi-decadal climate model simulations with the
Parallel Climate Model (Washington & Meehl)
• Uses NCAR LSM as land surface model
• One simulation with potential vegetation
• Another with 1970 (present-day) land cover
5 - Regrowth (tim ber)
11 - Tem perate M ixed F ores t
17 - S avan na
0 - Ocean
6 - Ice
12 - Tem perate Decid Forest
18 - Tropical W oodland
1 - Agri cult ure
7 - Tundra
13 - W arm M ixe d Forest
19 - Tropical Fores t
2 - Exte ns ive grass land
8 - Wooded Tundra
14 - G ras s/S tep pe
No D ata
3 - C plantation - NU
9 - Bo real Fo re st
15 - D esert
4 - Regrowth (aban don)
10 - C ool C o nifer
16 - S crubland
IMAGE 2.2 Land Cover Types
Climate Change From Present-Day Croplands
Summer (JJA) Daily Maximum Temperature (40 Year Average)
Present Day Land Cover – Natural Land Cover
• Decreased daily maximum temperature in June-August of present-day
croplands compared to natural vegetation
• Due primarily to higher albedo of croplands, but also to changes in
evapotranspiration
Transient Climate Change Simulations
IMAGE 2.2 - A2: 2100 Land Cover
5 - Regrowth (tim ber)
11 - Tem perate M ixed F ores t
17 - S avan na
0 - Ocean
6 - Ice
12 - Tem perate Decid Forest
18 - Tropical W oodland
1 - Agri cult ure
7 - Tundra
13 - W arm M ixe d Forest
19 - Tropical Fores t
2 - Exte ns ive grass land
8 - Wooded Tundra
14 - G ras s/S tep pe
No D ata
3 - C plantation - NU
9 - Bo real Fo re st
15 - D esert
4 - Regrowth (aban don)
10 - C ool C o nifer
16 - S crubland
IMAGE 2.2 Land Cover Types
Currently In Progress: Transient climate simulations from 1870-2100
using historical and future land cover change
Sensitivity of Climate Models to
Natural and Anthropogenic Forcings
• Paleo-climate simulation using PaleoCSM to test the validity over a long
integration period
• Designed suite of forcings to probe model
sensitivity for the 20th century
• Scaling of models to new forcings
NCAR Team: C. Ammann, G. Meehl, C. Tebaldi, B. Otto-Bliesner, E. Wahl
Outside Collaborators: P. Naveau (CU), N. Graham (Scripps/HRC), M. Mann (UVA), P. Jones
(CRU-UK), H.-S. Oh (UAlberta), F. Joos, C. Casty, J. Luterbacher (UBern), J. Bradbury, R. Bradley
(UMass), K. Cobb (CalTech)
Model Validation with the Proxy Climate Record
20th Century Climate
• Climate models with
only “natural” forcings
(volcanic and solar) do
not reproduce observed
late 20th century
warming
• When increases in
anthropogenic
greenhouse gases and
sulfate aerosols are
included, models
reproduce observed
late 20th century
warming
Years
Scaling of Climate Models
Pilot project uses PCM results from the B2 scenario and
scales to the A2 scenario using a simple linear regression
A22100  B21990   ( B22100  B21990 )
Here the regression coefficient is based on the global
mean temperature estimated (usually by an energy
balance model) for the A2 scenario.
A key statistical challenge is to characterize the
error in this method.
Scaled spring
precipitation field
for A2 scenario
Actual climate
projection
Error field scaled
by natural
variability
Scaling Errors as a Function of Resolution
Relative RMS errors for different grid box sizes
8 Realizations of Error Fields Along with the Actual Errors
How do we combine the results of several
climate models to make inferences about
changes in regional climate ?
NCAR Team: C. Tebaldi, D. Nychka, G. Meehl
External Collaborators: R. Smith (UNC)
Regional
Statistical
Analysis
Regional
Inference
for Climate
Change
Circa
the Third
AssessmentA2
Report
9 AOGCM
Projections
Scenario
Central Asia used as an example
Combining Multi-Model Ensembles
Consider the results of several models as a sample from a
hypothetical super population of models:
X j  m  ej
Yj  n  e j
j= 0: observed data j=1, M: the M models
X: current climate Y: future projection
m and n : true values
The variance in the errors is determined based on principles of
model bias and model convergence
Posterior Distributions for Current and Future
Winter Temperatures (DJF) for Central Asia
°C
Inference for the Mean Temperature Change
°C
Multivariate Model - Pooling Uncertainties
°C
Regions
Extremes from Climate Model
Projections
Number of frost days within a year is a useful
indicator for determining agricultural impacts and
also is a more extreme measure of climate
variability
NCAR Team: C. Tebaldi, G. Meehl
Model vs. Data:
Changes in frost
days in the late
20th century
show biggest
decreases over
the western and
southwestern
U.S. in
observations
and the model
Future changes in frost days from the climate model
show greatest decreases in the western and
southwestern U.S., similar to late 20th century
Large-scale changes in atmospheric circulation
affect regional pattern of changes in future
frost days
Anomalous ridge of
high pressure brings
warmer air to
northwestern U.S.
causing relatively
less frost days
compared to the
northeastern U.S.
where an
anomalous trough
brings colder air
from north
H
cold
L
warm
Influence of Climate Variability and Uncertainty
on Flood Hazard Planning in Colorado
• Extreme policy and decision making
• Precipitation event analysis
• Impacts of flood hazard planning
Standard tool for assessing flooding hazards is the Colorado
Precipitation-Frequency Atlas for the Western US, NOAA (1973), giving
contours of rain rates for various return periods.
The atlas has few measures of statistical uncertainty
NCAR Team: M. Downton, M. Crandell, O. Wilhelmi, R. Morss, U. Schneider, E. Gilleland
External Collaborators: P. Naveau (CU), R. Smith (UNC), A. Grady (NISS)
Extremes Analysis of Boulder Daily
Precipitation
• Simplest analysis is fitting a generalized
extreme value distribution to the annual
maxima
• An exceedance over threshold model can
account for seasonality and other covariates
• A common summary is the return time:
e.g., the size of an event whose average
time to occur is 100 years
Analysis Using Extremes Toolkit
Inference for the 100 Year Return Level for Boulder
Integration Across Uncertainty
• Decision making in resource management –
water resources
• Quantification of uncertainty in regional
climate change projections
• Climate/health issues
American Water Works Research
Foundation Collaboration with NCAR
Project to Identify Impacts of Global Climate
Change on Water Utilities (K. Miller and D. Yates).
Use a “primer on climate change” and a workshop
(Spring ’04) as vehicles to elicit feedback from
managers on multi-model statistical summaries.
These results will be used to modify the statistics.
Global
Regional
Water Resources
Assessment
Sacramento
WNA
ΔTemp, DJF
San Joaquin
%PCP, DJF
30
CC scenar
JJA
C
20
hist
10
DJF
20000
mm
CC scenar
0
%PCP, JJA
30
JJA
20
C
10
DJF
0
2000
1000
mm
Statistical Downscale
(Yates et al. 2003)
ΔTemp, JJA
hist
1000
0
1
11
21
31
41
51
61
71
81
91
Climate/Health
• Climate/health issues integrate well with other
parts of the Initiative
• Both temperature and precipitation extremes are
important contributors to problems in human
health
• Many important issues of uncertainty in attribution
of climate as a cause of health problems
(e.g., vector-borne disease)
Management
Monthly Meetings of Advisory Board – to
discuss topics such as:
• Web site development
• Find and develop points of integration
across projects
• Promote integration with other initiatives