THEME – 1 Anticipated dryland expansion in scenarios of global warming
1. Anticipated dryland expansion in scenarios
of global warming
ICARDA Workshop
Christopher T. Simmons
H. Damon Matthews
Concordia University,
Montréal, QC, Canada
2. Outline
• Introduction
– Observed regional changes in precipitation and
temperature in the last century
– Representative Climate Pathways (future scenarios)
– Climate models and uncertainties
• Detailed Results from CMIP5 on Dryland Changes
– Precipitation changes
– Projected P/PET and Aridity Classifications
– Shifts in the Köppen-Trewartha Classification of Climate
Regimes
• Conclusions
3. Observed Regional Precipitation Trends
• Globally, little
evidence of major
precipitation changes
or precipitation
extremes
(floods/droughts)
• Certain regions show
significant trends
toward drier or
wetter conditions
(i.e., drying in W.
Africa and parts of
the Mediterranean
Basin.
+ = significant
(90%
confidence)
Adapted from IPCC Fifth Assessment Report, WG1, Physical Science Basis, Pg. 44
1901-2010 1951-2010
White space
represents
missing data
4. Observed Regional Temperature Trends
• Net observed
warming in datasets
is already evident
across many
portions of the
globe.
• High latitude
warming is
accentuated by
decreases in
seasonal ice and
snow coverage
Adapted from IPCC Fifth Assessment Report, WG1, Physical Science Basis, Pg. 39
+ = significant
(90%
confidence),
white space is
missing data
5. Representative Concentration Pathways
(RCPs)
• Scenarios used by the
Intergovernmental Panel on
Climate Change (IPCC)
• Each pathway represents
different increases in
anthropogenic radiative
forcing, accounting both for
greater GHG concentrations
(warming effect) and
estimated aerosols (cooling
effect).
• Each pathway is named after
the quantity of radiative
forcing increase between
1750 and 2100 (i.e., RCP8.5 is
associated with a +8.5
Watts/m2 increase in radiative
forcing since 1750.
Adapted from IPCC Fifth Assessment Report, WG1, Physical Science Basis, Pg. 94
6. Representative Concentration Pathways
(RCPs) • With ~2.3 Watts/m2 already
obtained, RCP2.6 optimistically
assumes a near-term and
consistent decrease in fossil
fuel emissions.
• RCP4.5, the ‘stabilization
scenario,’ assumes a gradual
increase in emissions to mid-
century, followed by a decrease
• RCP6.0 assumes a longer
gradual increase in emissions,
followed by a decrease in the
last quarter of the century
• RCP8.5 is the ‘business-as-
usual’ scenario, with continued
increasing emissions
accompanying the projected
rise in global population
Adapted from IPCC Fifth Assessment Report, WG1, Physical Science Basis, Pg. 94
7. Climate Models
• Simplified mathematical
representations of the Earth’s
Climate System
– Usual contain a three
dimensional representation of
the atmosphere and/or the
oceans, land-surface
processes, land/sea ice, and
vegetation
• Validated on how well they
reproduce the climate of the
second half of the 20th
century (when meteorological
observations are more
available and abundant
globally)
• Future climate projections are
obtained by running climate
models, starting at the
present climate, with
estimates of how future
greenhouse gas (GHG) and
aerosol concentrations will
evolve (RCP scenarios)
8. Model Uncertainties
Fig. adapted from Knutti et al. (2013), Geophys. Res. Letters, 40, 1194-1199
• Model Performance:
Different models
reproduce historical
(observed)
precipitation and
temperatures with
different accuracies
• Parameterizations
(mathematical
simplifications of
complex processes):
Models have difficulty
capturing and
accurately reproducing
sub-scale processes
related to convective
precipitation and
tropical cyclones (IPCC:
low confidence), both
of which are important
for subtropical
hydrology
9. Model Uncertainties
Fig. adapted from Hawkins and Sutton (2010),
Clim. Dyn., doi: 10.1007/s00382-010-0810-6
• Internal Variability: very
similar initial conditions or
forcing leads to different
results for a single climate
model
• Model Uncertainty:
different models yield
different results for the
same emissions scenario.
• Scenario Uncertainty:
spread of model solutions
created by using different
RCP scenarios, related to
our lack of knowledge on
how future anthropogenic
emissions will evolve.
10. Model Uncertainties
Fig. adapted from Hawkins and Sutton (2010),
Clim. Dyn., doi: 10.1007/s00382-010-0810-6
• For different parameters
(temperature and
precipitation), different
uncertainties are important
• Model uncertainty is most
important in global
decadal mean
precipitation for short and
long timescales, whereas
internal variability is most
important in the first 20
years of the simulation
• Scenario uncertainty is
dominant for long
timescales in temperature
projections
11. Regional Uncertainties (Precipitation)
• For different
regions, sources
of uncertainty are
different.
• Uncertainty in
Sahel summer
precipitation is
mostly tied to
internal
variability, with
virtually no
influence of
scenario
uncertainty
Figs. adapted from Hawkins and Sutton (2010),
Clim. Dyn., doi: 10.1007/s00382-010-0810-6
13. CMIP5
Table adapted from Feng & Fu (2013), Atmos. Chem. Phys., 13, 10081-10094.
• Climate Model
Inter-Comparison
Project, Phase 5
(CMIP5)
• 27 Climate
Models are run
with the same
RCP scenarios
• The results of
these simulations
are averaged
together to
obtain the multi-
model average
(or ensemble
mean)
• Caveat: Not all
climate models
are independent
of each other
14. Projected Precipitation Trends
Fig. adapted from IPCC Fifth Assessment Report, WG1, Physical Science Basis, Pg. 91
Dots indicate 90% model agreement on sign of change and change is more than two standard
deviations away from the models’ internal variability
CMIP5 projected change from 1986-2005 average
greater
warming =
more
substantial
precipitation
decreases in
the
Subtropics
Greater
model
confidence of
precipitation
decreases in
drying regions
for more
extreme
warming
scenarios
15. Annual Climate Changes for RCP8.5
°C
% precip.
change
Precipitation Changes (projected change for 2071-2100 from 1961-1990 average)
Temperature Changes (projected change for 2071-2100 from 1961-1990 average)
From Feng et al. (2014), Glob. Planet. Change, 112, 41-52
16. Seasonal Changes in Precipitation for RCP8.5
Precipitation Changes (projected change for 2071-2100 from 1961-1990 average)
Adapted from Feng et al. (2014), Glob. Planet. Change, 112, 41-52
• Both winter (wet season)
and summer drying in
Mediterranean Basin
• Poleward expansion of
Hadley Cell and longer
Monsoon season leads to
greater precipitation in
Tropical Africa and the
Indian Subcontinent, esp.
during the Boreal
Summer.
• Very small increases in
precipitation result in
changes in
precipitation
percentage in regions
with relatively little
total precipitation (ex.
Sahara, Sahel, Arabian
Peninsula)
% precip.
change
17. Precipitation Changes related to
Monsoons
• Monsoon winds are driven
by broad-scale land-sea
temperature contrast
• Models show a weakening
of monsoon winds, but
overall wet season
lengthens due to greater
heating of land than water
(IPCC, likely)
• This leads to increased
precipitation on the Indian
Subcontinent and the
Tibetan Plateau,
particularly during the
Northern Hemisphere
Summer
18. • Poleward expansion of the descending branch
of the Hadley Cell suppresses precipitation in
regions experiencing greater subsidence,
leading to greater drying poleward from
traditional subtropical desert zones
• The Hadley Cell is expected to broaden,
leading to a wider tropical zones and wider
subtropical drylands, but it might also weaken
in intensity (IPCC Report)
Precipitation Changes related to the
Hadley Circulation
Fig. adapted from Yongyun et al. (2013), Adv. Atmos. Sci., 30, 790-795.
19. P/PET
Dryland Classification P/PET
Dry Subhumid 0.5-0.65
Semiarid 0.2-0.05
Arid 0.05-0.2
Hyper-arid 0-0.05
Fig. adapted from Feng & Fu (2013), Atmos. Chem. Phys., 13, 10081-10094.
• Ratio of Precipitation
to Potential
Evapotranspiration
(PET)
• PET = evaporation
one would get (for a
given air
temperature and
wind speed) over a
completely wet
surface
• Natural values range
from 0 (desert) to
slightly greater than
1 (very moist
regions)
20. Annual P/PET and Relative Humidity in RCP8.5
• Greater warming
over land decreases
relative humidity
• This casues more
evaporation from the
land surface and
increases PET for
much of the globe
• P/PET ratio
decreases in many
regions
% RH
change
CMIP5 Projected change for 2071-2100 from 1961-1990 average
Fig. adapted from Sherwood and Fu (2014), Science,
343, 737-738.
21. • Precipitation does not
increase enough in most
regions to counter the effect
of decreased RH with
increased T
• Gradual drying of many land
surfaces despite increased
precipitation
• Regions influenced by the
stronger monsoon under
greater GHG forcing
experience moistening
(equatorial Africa, Indian
subcontinent)
Annual P/PET vs. Precipitation for RCP8.5
Fig. adapted from Feng & Fu (2013), Atmos. Chem.
Phys., 13, 10081-10094.
CMIP5 Projected change for 2071-2100 from 1961-1990 average
22. Dryland Classification Changes for RCP8.5
Becoming WetterBecoming Drier
Regions with Strong Multi-Model Agreement
(80% consensus) on Drying Trend
CMIP5 Projected change for 2071-2100 from 1961-1990 average
• Expansion arid lands on the northern coast of Africa
to Anatolia, southern Africa, Australia, southwestern
North America.
• Hyper-arid lands expand in northern Africa and Iraq.
• Expansion of semiarid lands to the northern coast of
the Mediterranean, western Africa, Black Sea region,
North American Prairies, southern Africa, western
Africa, and South America.
• Model consensus is much better for regions
experiencing a drying trend (see Fig. to left) than for
regions becoming wetter (not shown).Figs. adapted from Feng & Fu (2013), Atmos. Chem. Phys., 13, 10081-10094.
23. Global Trends
CMIP5 Projected change for 2071-2100 from 1961-1990 average
• Models have
historically
underestimated
dryland expansion.
According to data,
drylands have
increased in size by 4%
since the 1950s.
• Greatest rate of
expansion of drylands
occur in next 50 years
in both RCP scenarios.
Little net change
thereafter in RCP4.5
• Greater warming
(RCP8.5) leads to more
drylands
(thermodynamic
mechanism)
Figs. adapted from Feng & Fu (2013), Atmos. Chem. Phys., 13, 10081-10094.
24. Model Uncertainties
• Averaging between
models eliminates
some of the original
climate signal
• In this example, spread
of precipitation for
multi-model ensemble
mean is nearly 30% less
than for the individual
simulations.
• Drying of more than
15-30% per degree
Celcius (or Kelvin,K)
change in temperature,
as seen in the
individual simulations,
is not captured in
multi-model mean
(multi-model mean
calculated for each grid
cell)
Fig. adapted from Knutti et al. (2010), Journal of Clim., 23, 2739-2758
27. Global Climate Regime Shifts
• 46 % of Earth’s land
surface undergoes a
climate regime shift
• Conversion to dryland
classification in many
parts of presently
subtropical northern
Africa and southern
Spain, Australia, central
Asia, southern Africa,
and southwestern
North America
• Mid-century expansion
of drylands by ~6-9%,
end of century by ~8-
16%
CMIP5 Projected change for 2071-2100 from 1961-1990 average for RCP.8.5
Table and Fig. adapted from Feng et al. (2014), Glob. Planet. Change, 112, 41-52
28. Global Climate Regime Shifts to More
Arid Conditions
Fig. adapted from Feng et al. (2014), Glob. Planet. Change, 112, 41-52
• Semiarid and arid
climate regimes
increase, largely at
the expense of
subtropical climate
regime categories
• Greater GHG forcing
leads to greater
warming and more
global aridification
• Models
underestimated more
recent climate regime
shifts to more arid
climates (1990s)
29. Main Conclusions
• Climate models participating in CMIP5 produce a strong consensus for
drying in subtropical regions for scenarios with greater global warming,
due in part to a broadening of the Hadley Circulation
• Specific regions which show the most consistent trends toward drying
in future scenarios are the Mediterranean and Black Sea basins,
southern Africa, southern Australia, and southwestern North America
• Even where increases in precipitation in many mid-latitude regions, a
decrease in relative humidity is also projected for many of these land
surfaces due to greater heating
• Model and scenario uncertainties have an important influence on the
projected expansion of drylands and must be taken into account for a
realistic perspective on future climate regime and vegetation changes
Slides Available at www.esmg.mcgill.ca/ICARDA.pdf
31. Contributors to Warming Climate
• Carbon dioxide, methane,
halocarbons and nitrous
oxide (greenhouse gases)
trap the radiation absorbed
and re-emitted by the
Earth’s surface, contributing
to warming.
• Emissions of aerosols and
cloud condensation nuclei
lead to less solar radiation
received at Earth’s surface,
and replacement of forests
with croplands leads to less
terrestrial absorption of
solar radiation, contributing
to cooling effect.
• Estimated total radiative
forcing: ~2.29 W m2
contribution of warming
from anthropogenic sources
Adapted from IPCC Fifth Assessment Report, WG1, Physical Science Basis,
Pg. 14
32. Model Uncertainties
• For different parameters
(temperature and
precipitation), different
uncertainties are important
• Model uncertainty is most
important in global decadal
mean precipitation for short
and long timescales, whereas
internal variability is more
important in the first 20 years
of the simulation
• Scenario uncertainty is
dominant for long timescales
in temperature projections
Figs. adapted from Hawkins and Sutton (2010), Clim. Dyn., doi: 10.1007/s00382-010-0810-6
33. Model Uncertainties
• Bias projection to
the ensemble mean
from nearly identical
models
– Difference in results
between models
developed by the
same institution are
often a factor of 3-
10 times smaller
than difference with
other models
Fig. adapted from Knutti et al. (2013), Geophys. Res. Letters, 40, 1194-1199
Climate Model Family Tree,
closeness based on similar
behaviour (correlations) of
temperature and
precipitation between the
individual models, often a
symptom of code sharing
between models
34. Global Climate Regime Shifts
Table adapted from Feng et al. (2014), Glob. Planet. Change, 112, 41-52
• Perspective of
broad-scale
Köppen-Trewartha
Climate
Classification, used
to diagnose
vegetation
changes
• 5 classes, B
represents
drylands (arid and
semi-arid)
35. Other Hydrological Parameters
Dots indicate 90%
model agreement
on sign of change
and change is more
than two standard
deviations away
from the models’
internal variability
Fig. adapted from IPCC Fifth Assessment Report, WG1, Physical Science Basis, Pg. 45
36. Thermodynamic Mechanism
Increasing Global Aridity
• Land warms an average of 50%
more than ocean surfaces;
increased land warming from
GHG forcing decreases relative
humidity.
• Relatively robust mechanism
and seen in most models of
global warming scenarios
e___
es
Clausius-Clapeyron Relationship
Relative
Humidity=
Water Vapour
Pressure
(content) of
atmosphere
Saturation
Vapour
Pressure at a
given
temperature