The Evolution of Climate Science: A Personal View from Julia Slingo

“Climate is what you expect, weather is what you get.” Therein lies an interesting question: what is the difference between weather and climate? It is a matter of timescale – climate is, in effect, the statistics of the weather averaged over some time period. As this paper will show, the science of weather underpins the science of climate.

Climate science is about understanding how the Earth’s climate works at a global and regional scale. Why climate varies and changes through internal interactions, such as El Niño and the Thermohaline Circulation, and in response to external forcing agents such as solar and volcanic activity. It also studies whether human activities, especially those linked to the emission of greenhouse gases, will fundamentally change how the Earth’s climate behaves. Not surprisingly, in recent years climate science has become synonymous with climate change science.

But climate science is about so much more than climate change. Climate science, as a discipline, has truly emerged during my career as a scientist, since my start as a researcher in the Met Office in 1972 following a degree in physics. But its roots go back much further. It is synonymous with the disciplines of meteorology, oceanography and climatology, and it is rooted in classical physics, mathematics, chemistry and – increasingly – biology. Modern climate science is fundamentally a fusion of theory, observations and computational modelling.

In this paper I will provide a personal perspective on how climate science has developed by highlighting a few key points in history, and by drawing on my own 40 years of experience of how it has transformed in that time through scientific and technological advances. Finally I shall address the issues raised by human-induced climate change and how climate science can help us plan for a safe and sustainable future.

Historical context

Climate science has a long and distinguished history. In 1686 Edmund Halley published his iconic picture of the tropical winds in the Philosophical Transactions of the Royal Society - An Historical Account of the Trade Winds, and Monsoons, Observable in the Seas between and near the Tropicks, with an Attempt to Assign the Phisical Cause of the Said Wind. Halley was curious why the winds invariably blew from the east and argued that it must be due to the daily passage of the Sun, whereby the Sun heated the atmosphere, causing the air to rise and hence pulling air in from the east in the wake of the Sun’s passage.

In 1735 it was George Hadley who postulated that in fact it is the Earth’s rotation that drives the easterly trade winds. In a paper that was largely ignored at the time he wrote: “...that the Air as it moves from the Tropicks towards the Equator, having a less Velocity than the Parts of the Earth it arrives at, will have a relative Motion contrary to that of the diurnal Motion of the Earth in those parts, which being combined with the Motion towards the Equator, a N.E. wind will be produc’d on this Side of the Equator, and a S.E. on the other.” He also realized that the greater heating from the sun over the equator must cause the air to rise and that through continuity there must be an equivalent region of descent and the production of westerly winds away from the tropics. From these ideas was born the Hadley Circulation, a fundamental part of the climate system.

It was not until much later that Hadley’s assertion that the Earth’s rotation is fundamental really came to fruition. In 1835 Gaspard-Gustave de Coriolis introduced his theory of how objects move within a rotating frame of reference and the forces that act upon them. Coriolis did not consider rotating spheres, but his theory was quickly taken up by meteorologists to explain Earth’s wind patterns. Hadley had been right in identifying the Earth’s rotation as fundamental, but he had mistakenly assumed that absolute velocity was conserved rather than absolute angular momentum.

In 1856 William Ferrel provided the first explanation of the global circulation and the westerly winds, or passage winds as they were known then, that characterize mid-latitude climates. So by the end of the nineteenth century, through a combination of observations and theory, the fundamental importance of the Earth’s rotation in defining the mean characteristics of atmospheric circulation and hence the climate system, from the easterly trades to the mid-latitude westerlies, had been demonstrated.

Edmund Halley’s map (1686, Philosophical Transactions of the Royal Society) depicts the major reversal of the trade winds between the winter and summer monsoons of Asia and Australia by using shorter dashes

The role of the Earth’s rotation reached its ultimate recognition in the work of Carl Gustaf Rossby who, from the 1930s onwards, introduced the concept of absolute vorticity and its conservation in adiabatic conditions. He developed the theory of planetary waves – Rossby Waves – within the atmosphere and oceans and essentially laid the foundations of dynamical oceanography and meteorology.

In parallel to the development of our understanding of atmospheric circulation, physicists were trying to understand why the Earth has the temperature that it does – in other words its energy balance. Beginning in the late 1850s, John Tyndall showed that the Earth's atmosphere must have a greenhouse effect to explain its warm surface temperature. He also demonstrated that these gases were emitters as well as absorbers of infrared radiation, which is vital for understanding the surface energy budget.

Svante Arrhenius took this a step further in 1896 by making the first calculations of the influence of carbon dioxide on the Earth’s surface temperature. In his book on Worlds in the Making published in 1908 he states that, “…any doubling of the percentage of carbon dioxide in the air would raise the temperature of the earth's surface by 4….” This early climate change projection, although at the high end of current projections, still lies within them.

It is interesting also to see how Arrhenius viewed human-induced climate change, its cause and its impacts:

The enormous combustion of coal by our industrial establishments suffices to increase the percentage of carbon dioxide in the air to a perceptible degree… By the influence of the increasing percentage of carbonic acid in the atmosphere, we may hope to enjoy ages with more equable and better climates, especially as regards the colder regions of the earth, ages when the earth will bring forth much more abundant crops than at present, for the benefit of rapidly propagating mankind.

Throughout the first half of the twentieth century concerns about climate change were very much focused on the possibility of entering another Ice Age based on what paleoclimate records from all sorts of geological evidence could tell us. In line with Arrhenius’ thinking, global warming was not, as yet, a serious concern.

In terms of the dynamics of the climate system, there is another aspect of climate science which is of profound importance – understanding how and why the climate of a region varies from year to year, and from decade to decade due to the internal variations in the climate system associated with oceanic and atmospheric flows and the interactions between them. If these can be understood then it may be possible to predict variations in regional weather and climate patterns at least a season ahead.

While Tyndall and Ferrel were pondering the global aspects of the climate system, India was of growing importance for the economy of the British Empire. Indian cotton and grain harvests made up nearly one fifth of the British economy and depended critically on the monsoon rains. Henry Blanford arrived in India in 1875 as the first British Director (Imperial Meteorological Reporter) of the Indian Meteorological Department. He found a climate where “Order and regularity are as prominent characteristics of our (India’s) atmospheric phenomena, as are caprice and uncertainty those of their European counterparts.” Blanford believed that he had found in India a perfect laboratory to understand how the weather works.

Rossby’s example of waves in the mid-latitude westerlies from his paper in 1940 on Planetary flow patterns in the atmosphere (Quart. J. Roy. Meteor. Soc.; left panel) and a real example of Rossby waves in the mid-troposphere (500 hPa height field) on 6 January 2014 during an extreme cold event over North America.

But he was soon to be confronted by the great famine of 1876–1878 when the monsoon rains failed dramatically and the British economy was deeply affected. He decided that because of the supposed simplicity of the Indian climate, it must be possible to find causes for these monsoon failures; through climate prediction famine could be anticipated and controlled, and India could be governed more effectively.

So began the growing body of research that sought to find relationships between the variations in climate in one region of the world and those in another, termed teleconnections by the British meteorologist, Sir Gilbert Walker.

In 1904 Sir Gilbert Walker arrived in India as the 3rd British Director (Director General of Observatories) of the Indian Meteorological Department. He began to draw together observations from around the world and pioneered statistical climate forecasting by constructing a “human computer”, with Indian staff performing a mass of statistical correlations using these data. As Walker said, “I think that the relationships of world weather are so complex that our only chance of explaining them is to accumulate the facts empirically.” From his endeavours came identification of the Southern Oscillation (and its association with failures of the Indian monsoon), the North Atlantic Oscillation and the North Pacific Oscillation.

Over the next 50 years, statistical climatology became a very important branch of climate science, through which empirical forecasting systems were developed to predict seasonal variations in the climate such as the Indian monsoon. But the causes of these climate variations were poorly understood; this is when oceanography enters the story.

The intermittent warming and cooling of the equatorial eastern Pacific Ocean – El Niño and La Niña – had been known for a long time, particularly by the Peruvian fisherman who saw their catch fail dramatically in El Niño years. In 1961 Vilhelm Bjerknes made the connection between this phenomenon in the ocean and the Southern Oscillation in the atmosphere, and the symbiotic relationship between the two – ENSO – was recognized. Although Henry Blanford did not know it then, the great Indian famine of 1876–1878 was caused by a very intense El Niño event.

So by the time I started my career in the Met Office in 1972, dynamical meteorology and weather forecasting, statistical climatology, paleoclimatology and oceanography were well established and the transformation of climate science was about to begin.

The 1^st transformation: Earth observation

We now know an immense amount about climate and how it varies and changes, through a vast array of observations, especially from space-borne instruments. In the 1970s, what we knew was based primarily on the network of meteorological observations that were used in weather forecasting. These gave us a very limited view of the general circulation of the atmosphere and very little understanding of the role of the water cycle. At that time the first images from weather satellites were appearing, showing how clouds are organized, and by the early 1980s the first direct measurements of the Earth’s radiation budget were made.

Over the subsequent decades the development of a constellation of geostationary and polar orbiting satellites has provided a rich resource for describing and monitoring the climate system. This is complemented by a myriad of surface and in situ observing systems, including surface weather stations, weather balloons, aircraft, ocean buoys, floats and ships.

We have been able to define the global flow of energy through the climate system with sufficient accuracy to know that the planet has been accumulating energy due to increasing atmospheric concentrations of greenhouse gases; and we know that around 90% of this additional energy is taken up by the oceans. We know that excess heat accumulated in the tropics is transported toward the poles, predominantly by the atmosphere in weather systems, and that phase changes of water – from evaporation at the Earth’s surface to condensation in the atmosphere forming clouds and precipitation – are a fundamental part of Earth’s energy cycle.

In fact the ability of the Earth’s climate to support water in its three phases – solid, liquid and vapour – is one of the unique characteristics of the planet. It means that between the Earth’s surface and the troposphere, heat can be taken out in one location and released far away from its original source.

Today, Earth observation tells us an immense amount about our climate system, but it does not tell us why the climate system works in the way it does, how different components interact and drive the variability we observe, and why the climate might be changing. For that we need to use numerical models of the climate system. 

Flows of energy through the global climate system (Wm-2) from Trenberth (2009). This emphasizes that although the balance at the top of the atmosphere is between the net shortwave radiation from the sun and thermal, infrared radiation from the planet, at the Earth’s surface the balance is much more complex. It involves other fluxes of energy besides radiation, predominantly from turbulent transports of moisture. In the atmosphere the balance is even more complex involving clouds, emission and absorption of thermal radiation by greenhouse gases and latent heat release.

The 2^nd transformation: Climate models

In principle, fundamental physics tells us everything about the motion of the atmosphere and oceans, about the thermodynamics of the water cycle, about the transfer of radiation through the atmosphere, and about how the atmosphere interacts with the underlying surface. In practice we have to solve these physical equations on a computer by dividing the Earth’s atmosphere and oceans into millions of volumes using sophisticated numerical techniques.

The first climate models, known as general circulation models, were developed in the 1950s alongside numerical weather prediction, which was also in its infancy. At that time the models were very simple in their construction and the first simulations considered only the adiabatic flow without any representation of the hydrological cycle. It was shown very quickly that to get anything like a realistic circulation required moist processes, but this raised some very big challenges that we still struggle with today.

The problem is that many of the processes that give rise, for example, to cumulus convection, condensation and the formation of clouds and precipitation occur at scales much finer than those resolved by the grid of the model. Much of the early development of general circulation models therefore focused on finding ways to represent these sub-gridscale processes through parameterization, in which the effect of these processes could be deduced from the resolved, large-scale characteristics of the atmosphere. Over the subsequent decades these parameterizations have developed substantially based on greater theoretical understanding, better observations and the use of detailed laboratory and field experiments.

For a climate scientist, the climate model is our laboratory. We cannot perform experiments on the real system to test hypotheses formed from theory and observations as one might in experimental physics or chemistry. Instead we need to use the model to pick apart feedbacks and interactions within the climate system so that we can understand how the system works and why it varies and changes. This means that we are always testing the validity of our models against theory and observations and always seeking to improve their skill.

Over the last few decades models have enabled us to understand so many aspects of the climate system, from how soil moisture feedbacks affect the West African monsoon, to why what happens in the tropical West Pacific drives the climate over North America, to how the 11-year solar cycle affects the winter weather we experience in the UK. We have learnt how mountains affect the position of storm tracks, whether Himalayan snow cover can really influence the progress of the Indian monsoon and many more important findings. All this has been achieved through careful design of “what if” experiments based on hypotheses drawn from observations of the past and present climate.

The 3^rd transformation: Supercomputing

Ever since their inception, climate models have been very computationally-intensive; therefore, the availability of computing power has dictated the level of sophistication and the type of experiments that can be performed. There are few sciences where progress can be so closely linked to the increases in supercomputing power.

Supercomputing has transformed climate science. It has enabled increases in resolution so that the models capture more faithfully the weather systems that constitute the climate; it has allowed the introduction of more components of the climate system, such as the carbon cycle and atmospheric composition, and their transformation to Earth system models; and it has delivered the capability to perform large numbers of simulations to test for robustness and to capture the plausible range of future states of the weather and climate that might arise naturally from the chaotic nature of the atmosphere and oceans.

Example of the improvements in the description of upper-ocean flows with higher model resolution in the Met Office climate model. This shows the surface currents with the strongest currents in white, and highlights the importance of resolution for capturing ocean eddies and western boundary currents, such as the Gulf Stream.

But complexity comes at a computational cost and so the resolution of climate models was compromised to enable different interactions and feedbacks within the climate system to be represented. It is now increasingly recognized that many of these interactions and feedbacks operate on space and timescales that relate to atmospheric weather and ocean eddies, and recent advances in supercomputing power are allowing this to happen.

Increasing the horizontal (spatial) resolution is particularly challenging; halving the grid length requires 10 times the computing power. Even the latest models used in the Intergovernmental Panel on Climate Change (IPCC) 5th Assessment Report (2013) had resolutions coarser than 100 km. But that is changing rapidly now as more computing is becoming available and there is greater appreciation of the scientific need to resolve atmospheric motions – the weather – and how they transport heat, momentum, moisture and other atmospheric constituents.

The ocean is potentially even more challenging because the scale of eddies is a fraction of those in the atmosphere. The resolutions of order 80 km typically used in climate models require the effect of ocean eddies to be parameterized and compromise important components of the ocean circulation, such as the Gulf Stream. The latest climate models, with significantly higher ocean resolution of order 20 km, begin to capture ocean eddies and are leading to substantial increases in skill, but it is generally regarded that resolutions of order 5 km are necessary to represent ocean eddies. As well as being important for transporting heat around the ocean, regions of eddy activity are also where biological production is high, so they are critical to the uptake of carbon by the ocean.

So the climate scientist is always making choices about how best to deploy available supercomputing resources, whether to trade resolution for complexity or ensemble size. There is never a single answer; it depends on the scientific application and on our level of understanding of what those choices mean for the validity of the simulation or prediction. There is no doubt that the availability of supercomputing power continues to hold back climate science and that a compelling case can be made for even greater investment.

The 4^th transformation: Global warming

In 1958 Charles David Keeling began making measurements of atmospheric concentrations of carbon dioxide (CO2) at Mauna Loa (Hawaii, USA), and soon began to notice that the concentrations were rising systematically year-on-year. So began the huge influence that human-induced climate change would have on climate science. 

The first simulations of the possible implications of increasing CO2 were performed in the 1970s and by the early 1980s were an integral part of climate research in the Met Office and beyond. One landmark study from 1974 on The Effects of Doubling the CO2 Concentration on the Climate of a General Circulation Model by Suki Manabe and Dick Wetherald, predicted a global warming of 2.93 Kelvin (K), close to the middle of the current range. They also predicted several other facets that we have now observed as signatures of greenhouse gas-induced climate change – stratospheric cooling, enhanced warming of the poles and greater warming of the tropical upper troposphere. 

The need to understand the sensitivity of the climate system to greenhouse gases undoubtedly had a big influence on model development. There have been major endeavours involving national and international science partnerships from the introduction of a fully interactive ocean model to address ocean heat uptake to the development of terrestrial vegetation and ocean biogeochemistry models to understand the role of the carbon cycle in amplifying global warming; to complex cloud microphysics to understand cloud feedbacks and interactive sea-ice models to address polar amplification.

By the time the IPCC published its 5th Assessment Report in 2013 the evidence for a warming world was unequivocal. The IPCC further stated that it is “Extremely likely that most of observed increase in global surface temperature since 1951 is caused by human influence.” That statement was based on the use of climate models to investigate what the world’s climate would have been like without human emissions of greenhouse gases and land use change. Without the development of sophisticated climate models in the last 50 years the attribution of global warming to human factors could not have been made.

Attribution of climate change now goes beyond just considering the global mean surface temperature to address other components, more regional aspects of the climate system and even extreme events such as flooding, drought and heatwaves. Year after year the evidence grows that human-induced climate change has made a contribution to the severity of these sorts of events.

Despite all the debates about uncertainties in climate models and in the projections of climate change, arguably one of the most important outcomes from the IPCC 5th Assessment Report was the very simple and fundamental fact that if we continue to accumulate carbon in the atmosphere then the planet will continue to warm, just as Arrhenius hypothesized in 1896. Without doubt, climate change has become a defining problem for the 21st century.

Helping us plan for a safe and sustainable future

In 1990, at the time of the publication of the first IPCC Report, Prime Minister Margaret Thatcher established the Hadley Centre for Climate Prediction and Research in the Met Office. Her words are as relevant today as they were 25 years ago:

We can now say that we have the Surveyor's Report and it shows that there are faults and that the repair work needs to start without delay. …… We would be taking a great risk with future generations if, having received this early warning, we did nothing about it or just took the attitude: "Well! It will see me out! ...The problems do not lie in the future – they are here and now – and it is our children and grandchildren, who are already growing up, who will be affected.

The evolution of climate science means that today it is ready to play a central role in helping us plan for a safe and sustainable future. The predictive power of climate models enables us “to see into the future,” so that we can be better prepared to deal with the risks we face from human interference with the climate system.

It is worth returning to where this paper began – “Climate is what you expect, weather is what you get.” Through the evolution of climate science outlined here, we have begun to appreciate more and more that there is no distinction between the weather and the climate; the same science underpins them both. As we look to the future and a warming world, the greatest impacts of climate change on society will be felt through changes in the weather, especially hazardous weather such as floods, storms and heatwaves. The transformation of climate science to a science that is rooted in local weather is the next big step.