Transcript Is Climate Really Predictable on 10-50 Year Time Scales?

Is Climate Really Predictable
on 10-50 Year Time Scales?
William R. Cotton
Professor of Atmospheric Science
Colorado State University
• We continually are bomb-blasted with scientific
articles, the news media, and talks like this that
human-produced greenhouse gases will and is
causing global warming
• While IPCC carefully argues that the models are
making “projections” not “predictions” of future
climate, there is still the implication that climate
is inherently predictable on time scales of 10 to
50 years or more; I ask, is it??
Weather and Climate Prediction: A
Humbling Experience
• While I have never tried to make a living
forecasting(thank heavens) I have made
forecasts in support of various field campaigns
as well as soaring forecasts for our glider club
on a weekly basis.
• It is a most humbling experience!
• Anyone who tells you that they can forecast the
climate in the next 10-50 years simply have not
had the opportunity to varify those forecasts and
by really humiliated!
• Let us begin with known climate forcing
factors and assess their predictability
Climate Forcing Factors
• Changes in solar luminosity and orbital
parameters
• Greenhouse gas variability—water vapor, CO2,
Methane.
• Changes in surface properties
• Differential temporal responses to external
forcing by the atmosphere and oceans.
• Natural and human-induced changes in aerosols
and dust--volcanoes, desert dust, pollutants
The Greenhouse Effect
• The major gases that absorb longwave radiation
are CO2, methane, and nitrous oxide. These are
what are referred to as greenhouse gases.
• Water vapor is actually the dominate
greenhouse gas. To obtain substantial
greenhouse warming the oceans must warm
and evaporate more water vapor in the air to
cause a positive feedback.
• Clouds are also major greenhouse warming
agents.
• Clouds also reflect solar radiation(cool)
• Globally clouds contribute to a net cooling as
reflection of solar radiation dominates LW
absorption.
• Because clouds are poorly treated in
General Circulation Models (GCMs) their
influence on climate is a major uncertainty
in climate prediction.
• For example, a 4% change in marine
stratocumulus cloud coverage can
completely negate the influence of
greenhouse gases!
Carbon Dioxide and climate
The solid line depicts monthly concentrations of atmospheric CO2 at Mauna Loa Observatory, Hawaii. The yearly
oscillation is explained mainly by the annual cycle of photosynthesis and respiration of plants in the northern hemisphere. The
steadily increasing concentration of atmospheric CO2 at Mauna Loa since the 1950s is caused primarily by the CO 2 inputs from
fossil fuel combustion (dashed line). Note that CO2 concentrations have continued to increase since 1979, despite relatively
constant emissions; this is because emissions have remained substantially larger than net removal, which is primarily by ocean
uptake. [From Scheraga, Joel and Irving Mintzer, 1990: Introduction. From Policy Options for Stabilizing Global Climate, D.A.
Lashof and D.A. Tirpak, Eds. U.S. Environmental Protection Agency, Office of Policy, Planning and Evaluation. Hemisphere
Publishing Corp. New York. ]
• A diagram that you will rarely see is the
following:
From Max Beran.
• That is a really sobering figure as it
suggests that to inhibit the growth of CO2
we must get our population under control.
Transition to non-fossil fuels is a step in
the right direction but as long as our
population continues to rise, it is likely that
CO2 will continue to rise
• Forecasting decadal and longer climate
requires a forecast of population as not
only do more humans on planet mean
greater changes in CO2, but also aerosols
and land-use.
• Predictability: small
• IPCC estimates greenhouse gases
contribute to 2.3[2.07 to 2.5] W m-2.
• Keep in mind that water vapor is the
dominant greenhouse gas on earth and
that clouds are dominant greenhouse
agents
Changes in solar luminosity and
orbital parameters
Changes in solar luminosity
• There are observed changes in solar luminosity which
account for something like 0.12[-0.4 to 00.0] W m-2 which
is small compared to the 2.3 W m-2 estimated for
Greenhouse gases. These changes are related to
changes in sunspot activity, solar diameter, and umbral
penumbral ratio.
• Nonetheless there are hundreds of statistical studies
which suggest a correlation with temperature and other
weather parameters that is far stronger than the
measured changes in luminosity imply. Is this just
statistics fooling us or is there some unknown amplifier?
• Some studies find that these parameters correlate with
cloud cover which would provide such an amplifier. But
convincing physical arguments have not been made.
Cosmic Ray Flux Variations
• Dozens of recent papers
relate(statistically) variations on cosmic
ray fluxes to global climate
• These studies show a positive correlation
between cosmic ray fluxes and cloud
cover(ie. contributing to warming)
• The argument is that high cosmic ray
fluxes generate ions which can then serve
as cloud condensation nuclei(CCN).
• The problem is, CCN are large(greater than 0.1
micrometer), soluble particles
• Ions, are several orders of magnitude smaller in
size and are not soluble so they do not activate
cloud droplets at real cloud supersaturations. To
become CCN they must coalesce with solvable
aerosols and have sulfates condense on them
which is not all that probable
• Moreover, cloud cover is mainly controlled by
dynamics(ascent and adiabatic cooling) and not
by concentrations of CCN and certainly not total
aerosol concentrations!
(19) The variations in sun activity reflect temperature events: Dalton minimum (Dm),
Maunder minimum (Mm), Spörer minimum (Sm), Wolf minimum (Wm), Oort minimum (Om),
and Medieval Maximum (MM).
Changes in orbital parameters
• The earth undergoes natural oscillations in
orbital parameters such as the eccentricity
of the orbit, the axial tilt, and the precession
of the equinoxes. The theory of climate
change related to variations in these
parameters is called the Milankovitch
theory and it predicts the earth will be
gradually moving into an ice age in the next
5000 years.
The Milankovitch theory
• Predictability of orbital-induced changes is
high but for solar variability in general is
low unless the statistical studies are totally
missleading
Changes in surface parameters
• The net albedo of Earth is determined by percent
cover of oceans vs. land, glacial coverage, landsurface vegetation vs. deserts, etc. In addition, the
latter land-surface parameters influence surface
temperatures through changes in sensible vs. latent
heat transfer.
• Human activity alters the land-surface parameters
through deforestation, agriculture, and urbanization.
• IPCC estimates these contribute to -0.2[-0.4 to 0.0]
W m-2 forcing but this does not include changes in
sensible and latent heat fluxes
• Prediction of land-surface changes
depends on population forecasts as well
as the global spatial distribution of
population--moderate
Differential temporal responses to
external forcing by the atmosphere
and oceans.
• The atmosphere and the deep oceans have
grossly different responses to changes in
external forcing.
• The atmosphere can respond on time scales of
days to months with lingering affects of about 1
year
• The ocean responds on time scales of 10’s of
years to even 1000 years
• This leads to a large natural variability of the
climate system and GCMs are unable to
represent or predict this variability well
• Predictability of deep ocean/atmosphere
remains quite small as ENSO, NAO, and
variability of thermohaline circulations
remains low
Natural variations in aerosols and dust
• Volcanoes are a major contributor to upper
tropospheric and lower stratospheric aerosols.
These particles block sunlight contributing to
surface cooling and can reside from a single
volcano for several years and have even longer
influences through cooling of the oceans.
• The period of warming during the 1930’s has
been attributed to a period of low volcanic
activity.
• There is no predictability of volcanic activity on
10 to 50 year time scales particularly long
clusters of volcanic activity!
Natural variations in dust
• Deserts and Sahalian zones in particular are large
sources of dust. These particles absorb solar radiation
and thereby warm the air layer they reside in and cool
the surface. Warming the air layer stabilizes the layer
reducing convection. Dust also alters cloud properties
appreciably. Human activity contributes to dust as well.
Not predicted well!
• If greenhouse warming contributes to desertification,
increases in surface wind strength, then additional dust
formation counters the warming.
• Meteor collisions with earth also contribute to dust and
have been blamed for the demise of dinosaurs. No
predictability!
Anthropogenic aerosols
• Air pollution aerosols contribute to cooling of the earth’s
surface by either reflecting solar radiation or directly
absorbing solar radiation which stabilizes the air layer
and cools the surface(called the direct aerosol effect)
• They also modify cloud properties (called indirect effect)
so that polluted clouds reflect more radiation (cooling
effect).
• They also modify the precipitation forming process(called
second indirect effect) which is treated in GCMs as
enhancing cloud albedo. But modeling and observations
suggest that there are many non-linear cloud dynamical
responses to aerosol which can reduce cloud coverage,
shift from solid stratus to open cellular convection,
reduce cloud liquid water paths.
• Aerosol variability, especially through altering the
hydrological cycle and precipitation, is a major source of
uncertainty in predicting climate.
Natural Variability
• How much of observed climate change in
the 20th century is due to greenhouse
forcing as opposed to natural forcing?
• How significant, compared to past natural
fluctuations are the changes we now
observe and expect in the future?
(5), The hockey stick according to Mann, M.E., R.S. Bradley and M.K. Hughes (1999) (8) Blue,
Black: reconstructions from tree rings, corals, ice cores, etc. Red: direct measurements from
temperature stations as from 1860.
McIntyre and McKitrick(2003)
• They criticize the Mann et al
reconstructions for:
• Deficiencies in the data used
• Irregularities in the data
• Methodology of analysis
(6), the hockey stick and the corrected temperature curve (green line) by
McIntyre between 1400 and 1980. The green curve is not intended to
indicate the true temperature, but to show the result of a correct use of
data.
• The thing that immediately struck me was the absence of
a strong Midieval Warm Period(800-1200AD) or Little Ice
Age( 1500-1850AD) in Mann’s analysis!
• They argue these were regional not global phenomena
• But other studies have found the MWP in Europe(Lamb,
1965; Shindell et al., 2001), Greenland(Dahl-Jensen et
al,1998), Africa(deMenocal et al, 2000; Holmgren et al,
2001), North America(Campbell et al,1998; Li et al,2000;
Petersen,1994; Shabalova and Weber,1999), South
America(Irionda et al,1993; Villabala,1994) and
Asia(Hong et al, 2000; Liu et al, 1998)
Juckes et al(2007) reconstructions
• They used other proxies other than just
tree rings
• There results seem to confirm the Mann et
al analysis
Problems with reconstructions:
• Proxie data such as tree rings deminish with
time: 22 extend back to AD 1400, 12 extend to
AD 1000(7 in N Hemisphere)
• Cook et al(2004) conclude reconstructions
bases largely on tree-rings should be treated
with caution earlier than AD 1200.
• Proxies are affected by factors other than
temperature which are not fully understood(ie,
Excessive Bristlecone pine growth in 20th
century could be due to CO2 fertilization or??)
• Can we say then that 20th century warming
is unprecedented compared to previous
natural periods like the Medieval Warm
Period with any confidence?
Loehl(2004)
• He fit time series data for “inferred” temperature
from Sargasso Sea SST estimates and from
stalagmites in a cave in South Africa to a simple
periodic set of models
• He fit these periodic models to 3000-year
temperature time series with minimal dating
error.
• Tree ring data were not used because of dating
uncertainties
• None of the models used 20th or 21st century
data
• The results clearly show the Medieval
warm period and the Little Ice Age
• 6 out of 7 of the fit models show a
warming trend over the 20th century similar
in timing and magnitude to the N
Hemisphere instrumental time series.
• One of the models passes right through
the 20th century data
• The results suggest that the 20th century
warming trends are a continuation of past
climatic cyclical patterns.
• Results are not precise enough to partition
20th century warming into natural vs manmade causes
• Nonetheless a major portion of the
warming could be a result of natural
causes
Conclusion
• As far as I am concerned the jury is still
out as to whether recent climate trends are
due to human activity or due to natural
variability associated with other forcing
parameters or internal variability of the
atmosphere/ocean/cryosphere.
There is evidence that the climate
is cooling in the 21st century
Ocean Heat Content:
• This is a better measure of climate
variability
• But records are of limited duration
Note flattening
2004-2008
Model hindcasts of climate trends
Using NCAR coupled model Warren
Washington Argues that Natural
Variations do not Explain Observed
Climatic Change
• Climate models with
natural forcing
(including volcanic
and solar) do not
reproduce warming
• When increase in
greenhouse gases is
included, models do
reproduce warming
• Addition of increase in
aerosols (cooling)
improves agreement
Quote for Jerry Meehl:
These simulations started from a pre-industrial control simulation that
was hundreds of years long. During this control run, none of the
forcings change, so the atmosphere and ocean come into balance
with each other and the drifts are minimal, though the model is left
with systematic errors compared to observations. Moreover cloud
parameterizations are “tweaked” in order to bring the TOA radiation
in balance. The 20th century runs branch from different time periods
in the control run and the forcings then change over the course of
the 20th century. Thus, the model results are anomalies from the
model state, compared to the observations that are anomalies from
the observed state. This is done to assess the relative importance
of different forcings on the time evolution of 20th century global
temperature anomalies.
ECMWF 10-year Hindcasts
• ECMWF(2009) is testing their ocean
coupled model for decade long forecasts
• They do not use techniques like anomaly
initialization, nudging or flux corrections to
avoid the coupled system from drifting
from the observed state
• It includes greenhouse gases and
sulphate aerosols
• The ECMWF simulations use an initialized
climate state (initialized with observations)
based on a 4DDA procedure. Thus, the
model systematic errors cause the model
to drift away from the initialized observed
state towards its own state.
Conclusions
• The model develops a 2-meter temperature bias
of ~ 1C over the first 2-5 years
• The tropical and subtropical oceans exhibit
strong cooling
• A substantial warm bias occurs over the northern
hemisphere extra-tropical continents
• In decadal forecasts, the forecast signals are
much smaller than model biases.
Initial-value vs Boundary-value
problem
• It is often claimed that climate is predictable
because it is a boundary value problem(that is,
only changes in external forcing is needed).
• But, we noted that deep ocean variability occurs
on time scales of 100’s of years
• Thus initialization of deep ocean circulations is
needed for forecasts on decadal time scales.
• This means that decadal climate prediction is
both an initial value problem and boundary value
problem
Is climate really predictable on 10
to 50 year time scales?
• Considering the stochastic external forcing
parameters(eg. Volcanoes), uncertainties
of solar variability forcing, and the
tendency for strong model biases on time
scales of 2-5 years let alone 10 to 50
years, I see no evidence that climate is
predictable on these time-scales nor will it
be for dacades to come(a forecast!).