Transcript Uncertainty in climate change

UNCERTAINTY IN
CLIMATE CHANGE
Andrew Levan
For fans of probability, confidence intervals and margins of error, climate change is a
dream come true. For everyone else, the fact that uncertainty (inherent in any complex
area of science) has gradually become one of climate change's defining features is a
constant headache. Because uncertainty – real or manufactured – is a well-rehearsed
reason for inaction.
The Guardian – 31 Jan 2014
Uncertainty in measurement
Uncertainty in prediction
Uncertainty in action
Uncertainty in measurement
It is extremely likely [95 percent confidence] more than half of the observed
increase in global average surface temperature from 1951 to 2010 was caused by
the anthropogenic increase in greenhouse gas concentrations and other
anthropogenic forcings together. – IPCC 2013
Uncertainty in prediction
It is more likely than not that the mean global mean surface air temperature for the
period 2016–2035 will be more than 1°C above the mean for 1850–1900, and very
unlikely that it will be more than 1.5°C above the 1850–1900 mean (medium
confidence). - IPCC 2013
Uncertainty in action
to preserve a livable planet, scientists tell us we must reduce the amount of CO2
in the atmosphere from its current level of 400 parts per million to below 350
ppm. – 350.org
Uncertainty in temperature measurements
A simple example:
I just took my own temperature using 3 different thermometers
(all good and the type you might find in hospital).
Oral : 36.6
Forehead : 36.3
Right ear: 36.9
Left ear 36.5
1) Different calibration between different types
of thermometer
2) Different locations on the body measured
(not all the same temperature)
3) Actually, temperature at any of these
location is a proxy for core temperature,
which can be a sign of serious illness. The
important thing is the illness (and its
consequence), not the temperature.
Why are these different? Does it matter?
Measuring global mean temperature is attempt to describe
a complex multi-dimensional problem with a single number.
It may be a useful number, but it doesn’t encapsulate the
full complexity of the problem.
Creating a temperature history
• Use direct measurements of temperature (thermometers) going
back as far as possible (~200 years).
• Chose proxies for global climate going back further
(dendrochronology has proved especially popular for the
hockey stick).
• Remove systematic offsets between different methods to bring
them to a common temperature scale.
• “Average” them together across the planet.
• Plot them
Error bars - exercise
• Measure the size of the LEGO block and the piece of
string.
• Write down the length and an error
Likelihood distributions
Probability extends well beyond the “error bars” of a measurement. Standard
deviation (sigma)
Likelihood distributions
That uncertainty is “normal” (alpha = 2 in this plot) a common assumption, but is
often untested. Fat tails mean extremes are much more likely than we expect.
Likelihood distributions
Small earthquakes are exceptionally common, but almost all the damage is caused
by large events. In regions prone to earthquakes buildings must be designed for
worst case scenarios.
Are these the natural considerations in climate related disasters?
Worry about the extremes (but be
careful of false predictions)
Correlation and causation
http://www.tylervigen.com/spurious-correlations
Number of trials
H H H H H H H H H H
H H H H H H H H
H
H
H
H
H
H H H H H H H H H H
H H H H H H H H
H
H
H
H
H
H H H H H H H H H H
H H H H H H H H
H
H
T
T
T
H H H H H H H H H H
H T T T T T T T
T
T
T
T
T
H H H H H H H H H H
T T T T T T T T
T
T
T
T
T
H H H H T T T T T T
T T T T T T T T
T T T T T T T T T T
T T T T T T T T
T T T T T T T T T T
T T T T T T T T T T
T T T T T T T T T T
Probability of 8 heads in a row = 1/28 = 1/256
H
H
H
H
T
T
T
T
T
T
T
T
H
H
H
T
H
H
T
H H
Trials in climate
There is increasing concern that most current published research findings are false.
The probability that a research claim is true may depend on study power and bias,
the number of other studies on the same question, and, importantly, the ratio of true
to no relationships among the relationships probed in each scientific field. In this
framework, a research finding is less likely to be true when the studies conducted in
a field are smaller; when effect sizes are smaller; when there is a greater number
and lesser preselection of tested relationships; where there is greater flexibility in
designs, definitions, outcomes, and analytical modes; when there is greater financial
and other interest and prejudice; and when more teams are involved in a scientific
field in chase of statistical significance. Simulations show that for most study
designs and settings, it is more likely for a research claim to be false than true.
Moreover, for many current scientific fields, claimed research findings may often be
simply accurate measures of the prevailing bias. In this essay, I discuss the
implications of these problems for the conduct and interpretation of research.
Cited 3104 times
When is something true?
• Many studies in medicine, economics and humanities
take a significance level of 95% (2-sigma) to mean
something is true.
• Many physicists will use 99.7% (3-sigma) as their
threshold.
• Teams using the Large Hadron Collider announced the
discovery of the Higgs Boson when it was detected at the
5-sigma level (99.9999426697%).
• New gravitational wave telescopes are using a 7-sigma
threshold for detection (99.9999999997440%).
When is something true?
• Would you put a loved one on a plane if there was a 95%
chance of it arriving safely?
• If I told you there was a 95% chance you would fail your
degree unless you took some action, would you take it?
So why should we believe this?
Answer: Physical basis, direct measurements, predictive power, number of trials
The physical basis for warming
• https://www.youtube.com/watch?v=kwtt51gvaJQ
95% confidence in warming
• IPCC statement
is 95%
confidence that
at least half of
the observed
warming arises
from
anthropogenic
influence.
Out of equilibrium
Stratospheric cooling, its not all about surface
temperature
IPCC approach
Confidence
IPCC approach
Likelihood
Discussion point
• What is the role of expert opinion?
Conditional probability
Conditional probability
CO2 is a known greenhouse gas. It traps IR radiation. Arguing that you don’t expect a
temperature increase from higher greenhouse gas concentrations is a bit not changing
your risk model for lightning strikes while standing in a thunderstorm
The null hypothesis
• The null hypothesis refers to a thing that you adopt as
your baseline and then try to disprove.
• Write down your null hypothesis for climate change (or in
particular global warming).
What are you really saying?
• “Sunday, May 11, 2014. Europe’s biggest clean-energy
market reached almost 75% renewable power market
share noon on that day.” – cleantechnia.com
• But electricity is roughly half of primary energy consumption in
Germany, less if you consider e.g. imports.
• “In the next 20 years we will cut our emissions by 30%
(relative to a business as usual approach)” – generic
statement
• But if business as usual is actually a 3% per year increase then this
is 80% over 20 years, so a 30% cut relative to business as usual is
asking to increase your emissions by ~25%.
Future scenarios
IPCC 2007
When to act
Costs and benefits
• Costs to acting are real. e.g. NHS is ~4% of UK GDP, c.f.
some estimates of cost of climate change mitigation.
• Indeed, NHS costs would be much lower if we mitigated
against many health risks rather than adapting, but most
of us don’t.
• How do we value money today vs money tomorrow –
discounting?
Principles of dealing with uncertainty
• It is a huge and complex area, and nobody gets it right all the
time.
• Think carefully about what you are being told.
• Have you been told everything you need to know?
• Could the data have been cherry picked somehow?
• What is the actual impact of a given result if correct?
• Be aware of your biases
• In climate change a multitude of independent studies offer
support for a a similar picture. Combined with a physical
expectation of this result it offers high confidence in
anthropogenic changes.
• Trying to express that confidence as some numerical value is
fraught with difficulty.