Transcript (I) - GCMs and Climate Change Scenarios

Applied Hydrology
Climate Change and Hydrology
(I) - GCMs and Climate Change Scenarios
Professor Ke-Sheng Cheng
Department of Bioenvironmental Systems Engineering
National Taiwan University
Climate dynamics, climate change and
climate prediction
 Climate: average condition of the atmosphere, ocean,
land surfaces and the ecosystems in them.
• e.g., "Baja California has a desert climate”
 Weather: state of atmosphere and ocean at given
moment.
 Climate includes average measures of weather-related
variability.
•e.g., probability of a major rainfall event occurring in July in
Baja, variations of temperature that typically occur during
January in Chicago, …
Neelin, 2011. Climate Change and Climate Modeling
4/8/2017
Lab for Remote Sensing Hydrology and Spatial Modeling
Dept of Bioenvironmental Systems Eng, NTU
2
 Climate quantities defined by averaging over the weather
•Average taken over January of many different years to
obtain a climatological value for January, many Februaries
to obtain February climatology, etc.
Climatology of sea
surface temperature
for January (15 year
average)
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
 Climate change:
•occurring on many time scales, including those that affect
human activities.
•time period used in the average will affect the climate that
one defines.
• e.g., 1950-1970 will differ from the average from 1980-2000.
 Climate variability:
•essentially all the variability that is not just weather.
• e.g., ice ages, warm climate at the time of dinosaurs,
drought in African Sahel region, and El Niño.
Climate change usually refers to changes in
statistical properties of climate variables.
A stationary climate process can and usually
do exhibit climate variability.
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
 Anthropogenic climate change: due to human activities.
•e.g., ozone hole, acid rain, and global warming.
Data from the Program for Model Diagnosis and Intercomparison (PCMDI) archive.
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
 Global warming: predicted warming, & associated changes
in the climate system in response to increases in "greenhouse
gases" emitted into atmosphere by human activities.
 Greenhouse gases: e.g., carbon dioxide, methane and
chlorofluorocarbons: trace gases that absorb infrared
radiation, affect the Earth's energy budget.
 warming tendency, known as the greenhouse effect
 Global change: human-induced changes more generally
(including ozone hole).
 Environmental change: even more general (including air,
water pollution, deforestation, ecosystems change, …)
 Climate prediction endeavor to predict not only humaninduced changes but the natural variations. e.g., El Niño
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
 Climate Dynamics or Climate Science: studies climate
and climate change processes (older term,
“climatology”).
 Climatology now used for average variables, e.g., “the
January precipitation climatology”.
 Climate models:
• Mathematical representations
of the climate system
• typically equations for temperature,
winds, ocean currents and other climate
variables solved numerically on computers.
 Climate System or Earth System: global,
interlocking system; atmosphere, ocean, land surfaces,
sea and land ice, and biosphere (plant and animal
component).
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
Changes in climate/weather
• Climate extremes or weather extremes?
• Extreme rainfalls are results of severe weather
events.
• Changes in climate can affect occurrences and
frequencies of extreme weather events.
• Studies which evaluate the impact of climate
change on rainfall extremes by comparing to
rainfall climatology may be misleading.
08/22/2013
2013 AsiaFlux, HESSS, GCEER, KSAFM Joint Conference
8
• Climate extremes and weather extremes
08/22/2013
2013 AsiaFlux, HESSS, GCEER, KSAFM Joint Conference
9
• Climate extremes
08/22/2013
2013 AsiaFlux, HESSS, GCEER, KSAFM Joint Conference
10
08/22/2013
2013 AsiaFlux, HESSS, GCEER, KSAFM Joint Conference
11
• Weather extremes
08/22/2013
2013 AsiaFlux, HESSS, GCEER, KSAFM Joint Conference
12
Climate models - a brief overview
 Motions, temperature, etc. governed by basic laws of
physics solved numerically:
•e.g., divide the atmosphere and ocean into discrete grid boxes
•equation for balance of forces, energy inputs etc. for each box.
•obtain the acceleration of the fluid in the box, its rate of change of
temperature, etc.
•from this compute the new velocity, temperature, etc. one time step
later (e.g., twenty minutes for the atmosphere, hour for ocean).
•equations for each box depend on the values in neighboring boxes.
•computation is done for a million or so grid boxes over the globe.
•repeated for the next time step, and so on until the desired length of
simulation is obtained.
•common to simulate decades or centuries in climate runs
 computational cost a factor
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
 Also close relationship to weather forecasting models
 Major differences:
•complexity of the climate system.
•range of phenomena at different time scales.
•“messier”: clouds, aerosols, vegetation, ...
 More attention to processes that affect the long term
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
 The most complex climate models, known as General
Circulation Models or GCMs.
•Once a phenomena has been simulated in a GCM, it is not
necessarily easy to understand.
 Intermediate complexity climate models are also used.
•construct a model based on same physical principles as a
GCM but only aspects important to the target phenomenon
are retained.
• e.g., first used to simulate, understand and predict El Niño.
 Simple climate models:
•e.g., globally averaged energy-balance model, to
understand essential aspects of the greenhouse effect.
 Global warming simulations with GCMs detailed
processes, 3-D response.
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
Global mean surface temperatures estimated since preindustrial times
From the University of East Anglia CRU (data following Brohan et al. 2006; Rayner et al. 2006)
•Anomalies relative to 1961-1990 mean
•Annual average values of combined near-surface air temperature
over continents and sea surface temperature over ocean.
•Curve: smoothing similar to a decadal running average.
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
 Anomaly: departure from normal climatological
conditions.
•calculated by difference between value of a variable at a
given time, e.g., pressure or temperature for a particular
month, and subtracting the climatology of that variable.
 Climatology includes the normal seasonal cycle.
•e.g., anomaly of summer rainfall for June, July and August
1997, = average of rainfall over that period minus averages
of all June, July and August values over a much longer
period, such as 1950-1998.
•To be precise, the averaging time period for the anomaly
and the averaging time period for the climatology should
be specified.
• e.g., monthly averaged SST anomalies relative to 1950-2000
mean.
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
Global Circulation Models (GCMs)
• Computer models that
– are capable of producing a realistic representation
of the climate, and
– can respond to the most obvious quantifiable
perturbations.
– Derived based on weather forecasting models.
4/8/2017
Lab for Remote Sensing Hydrology and Spatial Modeling
Dept of Bioenvironmental Systems Eng, NTU
18
Weather forecasting models
• The physical state of the atmosphere is updated
continually drawing on observations from
around the world using surface land stations,
ships, buoys, and in the upper atmosphere
using instruments on aircraft, balloons and
satellites.
• The model atmosphere is divided into 70 layers
and each level is divided up into a network of
points about 40 km apart.
4/8/2017
Lab for Remote Sensing Hydrology and Spatial Modeling
Dept of Bioenvironmental Systems Eng, NTU
19
• Standard weather forecasts do not predict
sudden switches between stable circulation
patterns well. At best they get some warning
by using statistical methods to check whether
or not the atmosphere is in an unpredictable
mood. This is done by running the models with
slightly different starting conditions and seeing
whether the forecasts stick together or diverge
rapidly.
4/8/2017
Lab for Remote Sensing Hydrology and Spatial Modeling
Dept of Bioenvironmental Systems Eng, NTU
20
• This ensemble approach provides a useful
indication of what modelers are up against
when they seek to analyses the response of the
global climate to various perturbations and to
predict the course it will following in the future.
• The GCMs cannot represent the global climate
in the same details as the numerical weather
predictions because they must be run for
decades and even centuries ahead in order to
consider possible changes.
4/8/2017
Lab for Remote Sensing Hydrology and Spatial Modeling
Dept of Bioenvironmental Systems Eng, NTU
21
• Typically, most GCMs now have a horizontal
resolution of between 125 and 400 km, but
retain much of the detailed vertical resolution,
having around 20 levels in the atmosphere.
• Challenges for potential GCMs improvement
– Modeling clouds formation and distribution
– Tropical storms (typhoons and hurricanes)
– Land-surface processes
– Winds, waves and currents
– Other greenhouse gases
4/8/2017
Lab for Remote Sensing Hydrology and Spatial Modeling
Dept of Bioenvironmental Systems Eng, NTU
22
GCMs
4/8/2017
Lab for Remote Sensing Hydrology and Spatial Modeling
Dept of Bioenvironmental Systems Eng, NTU
23
The parameterization problem
 For each grid box in a climate model, only the average
across the grid box of wind, temperature, etc. is
represented.
 In the observations, many fine variations occur inside,
• e.g., squall lines, cumulonimbus clouds, etc.
 The average of these small scale effects has important
impacts on large-scale climate.
• e.g., clouds primarily occur at small scales, yet the
average amount of sunlight reflected by clouds affects the
average solar heating of a whole grid box.
 Average effects of the small scales on the grid scale must be
included in the climate model.
 These averages change with the parameters of large-scale
fields that affect the clouds, such as moisture and temp.
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
Method of representing average effects of clouds (or other
small scale effects) over a grid box interactively with the
other variables known as parameterization.
Successes and difficulties of parameterization important to
accuracy of climate models.
finer grid implies greater computational costs (or shorter
simulation)
 As computers become faster  finer grids.
But there are always smaller scales.
Scale interaction is one of the main effects that makes
climate modeling challenging.
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
Constructing a Climate Model
Typical atmospheric GCM grid
•For each grid cell, single value of
each variable (temp., vel.,…)
Finite number of equations
•Vertical coordinate follows
topography, grid spacing varies
•Transports (fluxes) of mass,
energy, moisture into grid cell
Budget involving immediate
neighbors (in balance of forces,
PGF involves neighbors)
•Effects passed from neighbor to
neighbor until global
•Budget gives change of
temperature, velocity, etc., one
time step (e.g. 15 min) later
•100yr=4million 15min steps
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
Figure 5.1
Treatment of sub-grid scale processes
Vertical column showing parameterized physics so small
scale processes within a single column in a GCM
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
Resolution and computational cost
Topography of western North America at 0.3° and 3.0° resolutions
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
Topography of North America at 0.5° and 5.0° resolutions
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
Resolution and computational cost
• Computational time = (computer time per operation)
´(operations per equation)´(No. equations per grid-box)
´(number of grid boxes)´(number of time steps per simulation)
• Increasing resolution: # grid boxes increases & time step decreases
• Half horizontal grid size  half time step
 twice as many time steps to simulate same number of years
• Doubling resolution in x, y & z ´´´(# grid cells)
´´(# of time steps)
 cost increases by factor of 24 =16
• Increase horizontal resolution, 5 to 0.5 degrees  factor of 10 in each
horizontal direction. So even if kept vertical grid same, 10´10´(# grid
cells)´10´(# of t steps)= 103
• Suppose also double vertical res.  2000 times the computational time
i.e. costs same to run low-res. model for 40 years as high res. for 1 week
• To model clouds, say 50m res.  10000 times res. in horizontal, if same in
vertical and time  1016 times the computational time … and will still have
to parameterize raindrop, ice crystal coalescence etc.
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
• Why time step must decrease when grid size decreases:
– Time step must be small enough to accurately capture time evolution
and for smaller grid size, smaller time scales enter.
– A key time scale: time it takes wind or wave speed to cross a grid box.
e.g., if fastest wind 50 m/s, crosses 200 km grid box in ~ 1 hour
– If time step longer, more than 1 grid box will be crossed: can yield
amplifying small scale noise until model “blows up”
(for accuracy, time step should be significantly shorter)
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
Numerical representation of atmos. and oceanic eqns.
Finite difference versus spectral models
Finite differencing of a pressure field
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
Spectral representation of a pressure field
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
Climate simulations and climate drift
Climate drift
Examples of model integrations (or runs, simulations or experiments),
starting from idealized or observed initial conditions. Spin-up to
equilibrated model climatology is required (centuries for deep ocean).
Model climate differs slightly from observed (model error aka climate
drift); climate change experiments relative to model climatology.
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
Commonly used scenarios
Radiative forcing as a function of time for various climate forcing scenarios
Top of the atmosphere
radiative imbalance 
warming due to the net
effects of GHG and other
forcings
SRES:
• A1FI (fossil intensive),
• A1T (green technology),
• A1B (balance of these),
• A2, B2 (regional economics)
• B1 “greenest”
from the Special Report
on Emissions Scenarios
• IS92a scenario used in many
studies before 2005
Adapted from Meehl et al., 2007 in in IPCC Fourth Assessment Report
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
SRES emissions scenarios, cont’d
A1 scenario family: assumes low population growth, rapid economic
growth, reduction in regional income differences
A1FI : Fossil fuel Intensive
A1B: energy mix, incl. non-fossil fuel
A2: uneven regional economic growth, high income toward non-fossil,
population 15 billion in 2100
B1: like A1 but switch to information and service economy,
introduction of resource-efficient technology. Emphasis on global
solutions to economic, social, and environmental sustainability,
including improved equity.
•No explicit consideration of treaties
•Natural forcings e.g., volcanoes set to avg. from 20th C.
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
Model names (a sample)
•CCMA_CGCM3.1, Canadian Community Climate Model
•CNRM_CM3, Meteo-France, Centre National de Recherches
Meteorologiques
•CSIRO_MK3.0, CSIRO Atmospheric Research, Australia
•GFDL_CM2.0, NOAA Geophysical Fluid Dynamics Laboratory
•GFDL_CM2.1, NOAA Geophysical Fluid Dynamics Laboratory
•GISS_ER, NASA Goddard Institute for Space Studies, ModelE20/Russell
•MIROC3.2_medres, CCSR/NIES/FRCGC, medium resolution
•MPI_ECHAM5, Max Planck Institute for Meteorology, Germany
•MRI_CGCM2.3.2a, Meteorological Research Institute, Japan
•NCAR_CCSM3.0, NCAR Community Climate System Model
•NCAR_PCM1, NCAR Parallel Climate Model (Version 1)
•UKMO_HADCM3, Hadley Centre for Climate Prediction, Met Office, UK
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
Global average warming simulations in 11 climate models
• Global avg. sfc.
air temp. change
• (ann. means rel.
to 1901-1960
base period)
• Est. observed
greenhouse gas
+ aerosol
forcing, followed
by
• SRES A2
scenario (inset)
in 21st century
• (includes both
GHG and
aerosol forcing)
Data from the Program for Model Diagnosis and Intercomparison (PCMDI) archive.
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
Spatial patterns of the response to
time-dependent warming
scenarios
Response to the SRES
A2 scenario GHG and
sulfate aerosol forcing
in surface air
temperature relative to
the average during
1961-90 from the
Hadley Centre climate
model (HadCM3)
[choosing one model
simulation through the
21st century as an
example; later
compare models or
average results from
several models]
2010-2039
2040-2069
2070-2099
Figure 7.5
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
2010-2039
Response to the SRES
A2 scenario GHG and
sulfate aerosol forcing
in surface air
temperature relative to
2040-2069
the average during
1961-90 from the
National Center for
Atmospheric Research
Community Climate
Simulation Model
(NCAR_CCSM3)
2070-2099
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
January
January and July
surface temperature
from HadCM3
averaged 2040-2069
(SRES A2 scenario)
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
July
January
January and July
surface temperature
from NCAR_CCSM3
averaged 2040-2069
(SRES A2 scenario)
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
July
Comparing projections of
different climate models
GFDLCM2.0
30yr. avg annual surface
air temperature response
for 3 climate models
centered on 2055 relative
to the average during
1961-1990
NCARCCSM3
MPIECHAM5
Figure 7.7
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
Comparing projections of different climate models
•Provides estimate of uncertainty
•Differences often occur with physical processes e.g., shift of jet stream,
reduction of soil moisture, …
•At regional scales (~size of country or state) more disagreement
•Precip challenging at regional scales
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
Comparing projections of
different climate models
GFDLCM2.0
Precipitation from 3
models for Jun.-Aug.
2070-2099 average
minus 1961-90 avg
(SRES A2 scenario)
NCARCCSM3
MPIECHAM5
(mm/day)
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
Comparing projections of
different climate models
GFDLCM2.0
Precipitation from 3
models for Dec.-Feb.
2070-2099 average
minus 1961-90 avg
(SRES A2 scenario)
NCARCCSM3
MPIECHAM5
(mm/day)
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
Precipitation from HadCM3 for Dec.-Feb. 2070-2099 avg. (SRES A2)
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
Precipitation from HadCM3 for Jun.-Aug. 2070-2099 avg. (SRES A2)
Neelin, 2011. Climate Change and Climate Modeling, Cambridge UP
• From GCMs to hydrological process modeling
– Study of hydrological processes requires spatial and
temporal resolutions which are much smaller than
GCMs can offer.
– Downscaling techniques have been developed to
downscale GCM outputs to desired scales.
• Dynamic downscaling
• Statistical downscaling
4/8/2017
Lab for Remote Sensing Hydrology and Spatial Modeling
Dept of Bioenvironmental Systems Eng, NTU
51
What is a control run?
• Control run - A model run carried out to
provide a ‘baseline’ for comparison with
climate-change experiments. The control run
uses constant values for the radiative forcing
due to greenhouse gases and anthropogenic
aerosols appropriate to pre-industrial
conditions.
4/8/2017
Lab for Remote Sensing Hydrology and Spatial Modeling
Dept of Bioenvironmental Systems Eng, NTU
52