Climate Science Looks Back to Predict the Future—And It’s Not Pretty
Simulating
past climates gives science an uncomfortably accurate picture of what
the future holds: The last time it was only a little warmer than now,
seas were 9 meters higher.
In July, the separation of a giant iceberg from the Larsen-C ice shelf in Antarctica reminded us yet again that we can expect higher sea levels in the future—a scenario that has long been predicted by experts. In fact, higher sea levels
are of such interest that climate scientists study warm periods of the
past—when sea-levels were higher than today—to predict how our climate
will change throughout the 21st century and beyond. The Eemian
interglacial period, which occurred around 120,000 years ago, is an
example of one such era.
Anyone who doubts the validity
of future projections made by climate scientists should pay attention to
how well their Earth System Models (ESMs) successfully simulate these
past climates. Not only do ESMs do a good job at simulating the climate
of today, which is a routine task in the validation of any model poised
to make future projections, but past simulations—usually referred to as
paleo simulations—have also become remarkably accurate.
During
my time as a PhD researcher, I used ESMs to look at ocean temperatures
around South Africa in the region where the Agulhas Current flows
southwards along the western coast of the continent. Here, the model I
used was able to simulate sea surface
temperatures during the Last Glacial Maximum, a period around 20,000
years ago at the height of the last ice age, to within a degree of those
estimated by geological data.
For
me, seeing the results with my own eyes made a difference, and
demonstrated the predictive power of these models. Our team also used
another similar climate model to look at important weather-affecting
changes in Atlantic Ocean circulation during the last ice age (between
80,000 and 20,000 years ago). Again, the modelled data—mostly air
temperatures this time—were remarkably accurate when compared to
geological data, and the results were published in a paper in Nature.
Notwithstanding
the reasonable margins of error which are inherent in both the models
and the geological data they are compared with, paleo simulations show
us that when you change a model's atmospheric carbon dioxide along with
other paleo boundary conditions, the computed results are able to match
the past temperatures revealed by geological proxy—and with ever
improving accuracy.
Simulating the changing Earth
I
can hear you wondering how these past simulations are made. Well,
they're made by using the same state-of-the-art ESMs which are routinely
used for future Earth prediction, but by applying paleo background
conditions. ESMs are built around mathematical descriptions of the
physics governing our planet, and have become rather sophisticated in
recent years. The models, often referred to as climate models, are built
in components, each representing some part of the Earth’s eco-system.
Such component parts include the atmosphere, oceans, continental land surface
(as well as the vegetation on it), the ice-sheets, and sea-ice. Each
component interacts with the others, sending and receiving information
about temperature and other parameters which change through time at each
location on the modelled planet.
Using
these models, climate scientists can create simulations of Earth at
different times in the past: from the recent Holocene period, back
through the ice ages to the warmer climates of the deeper past, like the
Miocene, or deeper still, the Triassic or Cretaceous, when dinosaurs
dominated the planet. Super deep time simulations have also been made,
including that of “Snowball Earth” a time (or number of times—nobody's
sure) when the Earth is thought to have been almost completely covered
in ice and snow. The results of paleo simulations are then compared with
geological data from ice or sediment cores which are obtained from
locations as diverse as ice sheets on Greenland and Antarctica, to
African lakes or the ocean floors across the world.
My
own ice-age work involved changing a model's atmospheric carbon dioxide
level (in that case lowering it to 190 parts per million, from around
400 today), increasing the height and volume of the ice sheets across
North America, Europe, and Antarctica, and lowering global sea levels
accordingly. The latter effectively closed the Bering Strait, one of the
only major ocean gateway differences between the ice age and today. All
this is done by coding and reconfiguring the ESM, and by observing the
simulated Earth system with a viewer which allows the user to see the
planet on an interactive map.
Nowadays, a range of climate models, each built independently of each other, can reconstruct past climate variables like sea surface
and air temperature with good accuracy compared to geological data.
Again and again, the models produce remarkable results. Even more
impressive is that some models are now able to go beyond simply
simulating so-called time-slices, i.e., moments in time, say, 20,000
years ago at the height of the last ice age, but are also able to
perform transient time varying simulations, say, from the last ice age
forward to today, year by year.
Deep time
On
global scales, modelled results are generally respectable compared to
geological data. Of course, regional differences are harder to be exact
about. And there are other limitations, too. For example, the farther we
go back in time, the less accurate the models become. Things get a bit
tricky, and with good reason: it's much harder to reconstruct a world
millions of years ago than it is one just thousands of years back.
are of such interest that climate scientists study warm periods of the
past—when sea-levels were higher than today—to predict how our climate
will change throughout the 21st century and beyond. The Eemian
interglacial period, which occurred around 120,000 years ago, is an
example of one such era.
Anyone who doubts the validity
of future projections made by climate scientists should pay attention to
how well their Earth System Models (ESMs) successfully simulate these
past climates. Not only do ESMs do a good job at simulating the climate
of today, which is a routine task in the validation of any model poised
to make future projections, but past simulations—usually referred to as
paleo simulations—have also become remarkably accurate.
During
my time as a PhD researcher, I used ESMs to look at ocean temperatures
around South Africa in the region where the Agulhas Current flows
southwards along the western coast of the continent. Here, the model I
used was able to simulate sea surface
temperatures during the Last Glacial Maximum, a period around 20,000
years ago at the height of the last ice age, to within a degree of those
estimated by geological data.
For
me, seeing the results with my own eyes made a difference, and
demonstrated the predictive power of these models. Our team also used
another similar climate model to look at important weather-affecting
changes in Atlantic Ocean circulation during the last ice age (between
80,000 and 20,000 years ago). Again, the modelled data—mostly air
temperatures this time—were remarkably accurate when compared to
geological data, and the results were published in a paper in Nature.
Notwithstanding
the reasonable margins of error which are inherent in both the models
and the geological data they are compared with, paleo simulations show
us that when you change a model's atmospheric carbon dioxide along with
other paleo boundary conditions, the computed results are able to match
the past temperatures revealed by geological proxy—and with ever
improving accuracy.
Simulating the changing Earth
I
can hear you wondering how these past simulations are made. Well,
they're made by using the same state-of-the-art ESMs which are routinely
used for future Earth prediction, but by applying paleo background
conditions. ESMs are built around mathematical descriptions of the
physics governing our planet, and have become rather sophisticated in
recent years. The models, often referred to as climate models, are built
in components, each representing some part of the Earth’s eco-system.
Such component parts include the atmosphere, oceans, continental land surface
(as well as the vegetation on it), the ice-sheets, and sea-ice. Each
component interacts with the others, sending and receiving information
about temperature and other parameters which change through time at each
location on the modelled planet.
Using
these models, climate scientists can create simulations of Earth at
different times in the past: from the recent Holocene period, back
through the ice ages to the warmer climates of the deeper past, like the
Miocene, or deeper still, the Triassic or Cretaceous, when dinosaurs
dominated the planet. Super deep time simulations have also been made,
including that of “Snowball Earth” a time (or number of times—nobody's
sure) when the Earth is thought to have been almost completely covered
in ice and snow. The results of paleo simulations are then compared with
geological data from ice or sediment cores which are obtained from
locations as diverse as ice sheets on Greenland and Antarctica, to
African lakes or the ocean floors across the world.
My
own ice-age work involved changing a model's atmospheric carbon dioxide
level (in that case lowering it to 190 parts per million, from around
400 today), increasing the height and volume of the ice sheets across
North America, Europe, and Antarctica, and lowering global sea levels
accordingly. The latter effectively closed the Bering Strait, one of the
only major ocean gateway differences between the ice age and today. All
this is done by coding and reconfiguring the ESM, and by observing the
simulated Earth system with a viewer which allows the user to see the
planet on an interactive map.
Nowadays, a range of climate models, each built independently of each other, can reconstruct past climate variables like sea surface
and air temperature with good accuracy compared to geological data.
Again and again, the models produce remarkable results. Even more
impressive is that some models are now able to go beyond simply
simulating so-called time-slices, i.e., moments in time, say, 20,000
years ago at the height of the last ice age, but are also able to
perform transient time varying simulations, say, from the last ice age
forward to today, year by year.
Deep time
On
global scales, modelled results are generally respectable compared to
geological data. Of course, regional differences are harder to be exact
about. And there are other limitations, too. For example, the farther we
go back in time, the less accurate the models become. Things get a bit
tricky, and with good reason: it's much harder to reconstruct a world
millions of years ago than it is one just thousands of years back.
these so-called deep-time scenarios, there are huge uncertainties in a
model's input boundary conditions. Modeling deep time worlds requires
complex reconstruction of such things as the period's continental
configurations (whether Gondwana or Pangea, or what have you), the
associated ocean changes (don't forget oceans have a major control on
global climate) and their vegetation types (dense forests at the poles
during warmer periods, for example), which all in all involve a lot of
unknowns.
It's therefore no surprise that this kind of
deep-time modeling is only in its infancy. Having said that, research in
the field is ongoing and the future is sure to herald in a new era of
discovery.
The Eemian as an analog for the future
Either
way, future projections don't involve these continental-scale tectonic
changes which are associated with deep time, and are therefore much
easier to make. As I've already mentioned, some well constrained past
warm periods, such as the Eemian interglacial, stand as good analogs for
the climate of the near future, and are used to study potential
scenarios.
The Eemian was characterized by temperatures
only slightly warmer than today and accompanied by a significantly
higher global sea level, between six and nine meters higher. At that
time, sea levels had increased to such heights due to the melting ice
sheets on Greenland and Antarctica (back then the ice sheets were
smaller than today), and also because of the thermal expansion of the
ocean caused by higher temperatures. This warmer climate permitted
forests to grow at higher latitudes in the northern hemisphere and
different species of plants and animals to live in places they would
unlikely be found today.
So should we believe that the
diversity and geography of our future ecosystems might resemble the
Eemian warm period? Well, current observational studies show that
significant ecosystem changes are already underway, and modeling
suggests that changes in the behavior of many groups of plant and animal
species are to be expected. Temperatures are already rising to Eemian
levels and the polar ice sheets are indeed melting, which is resulting
in sea level rise. But just how close our future resembles the Eemian
will depend on our current action on the energy issue, and will
ultimately be for future investigators to explore.
Without
a doubt, the largest factor determining the Earth's temperature over
the next few hundred years is rising atmospheric carbon dioxide. As it
stands, our best estimates tell us that by 2100 our global climate will
be, on average, anywhere between 2 and 6 degrees warmer than
today—enough to cause extreme weather patterns including storms and
drought, unknown ocean biodiversity changes, global mass migration, and
political chaos.
Most of all, the change will likely
usher in the Earth's sixth mass extinction, which is already thought to
be underway. This is how humanity will really make its mark on
geological time. Perhaps some future human species, or alien race, will
one day run paleo simulations of our times and wonder how we managed to
cause so much destruction in just a few hundred years.
