Decadal Warning Systems - Reducing Disaster: Early Warning Systems for Climate Change - page 298

Box 16.1 (continued )
these cool the planet by refl ecting some of the Sun's radiation back to space.
Examples of human external forcing factors include changes in greenhouse
gases which warm the planet by trapping some of the radiation emitted by the
Earth's surface, changes to aerosol concentrations from factory emissions
(especially of sulphur dioxide) and land use changes which change the amount
of solar radiation refl ected back to space.
Natural internal variability
Climate variations that are not caused by external factors are referred to as “nat-
ural internal variability”. Examples include day to day weather, the El Niño
Southern Oscillation (ENSO) which causes the tropical Pacifi c to warm and
cool every few years and the Atlantic Multi-decadal Oscillation (AMO) or
Atlantic Multi-decadal Variability (AMV) in which changes to the strength of
Atlantic Ocean currents modulate sea surface temperatures in the North Atlantic.
The Pacifi c Ocean may also vary naturally on decadal timescales (known as the
Pacifi c Decadal Oscillation, PDO, or the Inter-decadal Pacifi c Oscillation, IPO),
but the extent to which this is independent from ENSO and AMV is unclear.
Climate change projections
Standard climate change projections simulate the climate response to changes
in external forcing factors. This enables the detection and attribution of past
climate change, as well as projections of future changes driven by scenarios
describing changing concentrations of greenhouse gases, anthropogenic aero-
sols and ozone and changes in land use (IPCC 2007 ). Although climate
change projections simulate natural internal variability, it will not in general
be in phase with reality. This is not important for long-term projections if the
projected climate change signal is much larger than natural internal variability
but is a major source of uncertainty on decadal timescales.
Decadal climate prediction
Decadal climate predictions attempt to provide the most accurate information
possible about the coming decade or two by considering natural internal
variability as well as external forcing factors. This is achieved by initializing
climate models with the observed state of the climate system, especially the
oceans. Initialization is necessary to predict internal variability but may also
improve predictions relative to standard climate projections by correcting the
model response to previous external forcing factors. There are strong analo-
gies between weather forecasting for a few days ahead and longer-term
decadal forecasts for the next few years. When we make a weather forecast
we are predicting what the likely evolution of the atmospheric circulation will
be over the next few days; when we make a decadal forecast we are essentially
predicting ocean “weather” (i.e. how the oceanic circulation will evolve over
the next few years) and its subsequent impact on the probabilities of different
weather patterns within the atmosphere.