GeoLog

climate modelling

How to forecast the future with climate models

How to forecast the future with climate models

Our climate is constantly changing, and with the help of simulation modelling, scientists are working hard to better understand just how these conditions will change and how it will affect society. Science journalist Conor Paul Purcell has worked on Earth System Models during his time as a PhD student and postdoctoral researcher; today he explains how scientists use these models as tools to forecast the future of our climate.

While we can’t predict everything about our future, climate scientists have a good understanding of how our environment will look and feel like in the coming years. Researchers and climate specialists predict that temperatures will increase dramatically in the 21st century, ranging between 1.5°C and 4°C above pre-industrial levels, depending on your location and the amount of carbon dioxide pumped into the atmosphere in the near future. Forecasts of future drought and flood risk, at both regional and global bases, are also provided by climate experts.

Understanding how such features of Earth’s changing climate may manifest, and ultimately impact on our society, takes considerable international collaboration – a collaboration which is largely based around the results of climate modelling. That’s because climate predictions for the future are made using sophisticated computer models, which are built around mathematical descriptions of the physical and biological processes that govern our planet.

These models have become so complex in recent years that they are now referred to as Earth System Models (ESMs). Using ESMs, climate modellers can create simulations of the planet at different times in the future and the past. ESMs are in fact the only tools we have for simulating the global future in this sense. For instance, if we want to know how our climate may look like one hundred years from now, how ocean acidification levels may change and how this might impact ocean life, or how plants will respond to increasing levels of atmospheric carbon dioxide, ESMs are the only tool available.

The models are built in components, each representing a separate part of the Earth system: the atmosphere, the ocean, the land surface and its vegetation, and the ice-sheets and sea-ice. These are constructed by coding each component with the mathematics that describes the environmental processes at work.

Climate models are systems of differential equations based on the basic laws of physics, fluid motion, chemistry, and biology. Pictured here is a schematic of a global atmospheric model. (Credit: NOAA, via Wikimedia Commons)

For example, the winds in the atmosphere are described by the mathematics of fluid motion. Model developers translate these mathematical equations into code that computers can understand, like giving them a set of instructions to follow. Supercomputers can then interpret the code to simulate how winds, for example, are expected to develop at each global location through time. The results are usually plotted on world maps.

As scientists have learned more about our Earth’s systems over time, the complexity of these individual models has been ramped up dramatically. For example, the land surface and vegetation model components become more sophisticated as plant biologists understand more and more about how plants transfer water and carbon between the land and atmosphere.

And it’s not just one giant solo project either: there are tens of ESMs and hundreds of subcomponent models developed and used at research centres around the globe. Collaboration between these facilities is a necessary part of progress, and information is shared at international conferences ever year, like the American Geophysical Union’s Fall Meeting in the United States and the European Geosciences Union’s General Assembly in Europe.

This means that developments are always been made towards increasing the realism of ESMs. On the horizon such developments will include increasing the resolution of the global models for improving accuracy at regional locations, and also incorporating the results from the latest research in atmospherics, oceanography and ice sheet dynamics. One example is research into plants, specifically how they interact with carbon dioxide and water in the atmosphere. Further understanding of this biological process is expected to increase the realism of models over the coming years and decades. In general, improvements to the accuracy of model simulations can help to help society in the future. For example, models will be able to help predict how climate change may impact, say, water scarcity in South Africa, wildfire risk in the western United States, or crop yields in Asia. Indeed, the ESMs of the future should boast incredibly accurate simulations and prediction capabilities unheard of today.

By Conor Purcell, a Science & Nature Writer with a PhD in Earth Science

Conor Purcell is a science journalist with a PhD in Earth Science. He is also founding-editor of www.wideorbits.com and is on twitter @ConorPPurcell and some of his other articles at cppurcell.tumblr.com.

GeoTalk: Nick Dunstone, an outstanding young scientist

 Nick Dunstone, the winner of a 2014 EGU Division Outstanding Young Scientists Award, who studies the Earth’s climate and atmosphere, including how they are impacted by natural variation and anthropogenic emissions talks to Bárbara Ferreira, the EGU Media and Communications Manager, in this edition of GeoTalk. This interview was first published in our quarterly newsletter, GeoQ.

NickFirst, could you introduce yourself and tell us a bit about what you are working on at the moment?

My name is Dr Nick Dunstone and I am a climate scientist working at the Met Office Hadley Centre in the UK. Here I work within the Monthly to Decadal Climate Prediction group which focuses on developing regional climate prediction capability for all areas of the globe. The monthly to decadal timescale (often referred to as ‘near-term’ prediction) is an emerging and challenging field of climate prediction which attempts to span the void between shorter term weather forecasts (days to weeks) and longer term climate projections (many decades to centuries) using numerical climate models. So, similar to a weather forecast, near-term climate predictions are initialised close to the observed state of the climate and yet, similar to a climate projection; they also include the projected changes in external forcings such as greenhouse gases, anthropogenic aerosols and the solar cycle. Much of my research over the last few years has concerned the amount of predictability in the climate system arising from slowly varying internal processes (for example, slowly varying ocean dynamics) versus how much is driven by external forcings (e.g. anthropogenic emissions).

Earlier this year, you received a Division Outstanding Young Scientists Award for your work on the coupled ocean-atmosphere climate system and its predictability. Could you tell us a bit more about the research you have developed in this area?

Some of my work has considered the role of internal ocean dynamics in driving predictability in the atmosphere. Often we think of the tropical regions as being the engine of the climate system, driving some of the variability in the mid-latitude atmosphere. However, this is not always the case and especially on longer timescales (multi-annual to decadal), the mid-latitudes can drive tropical variability. My colleagues and I illustrated this using a set of idealised climate model experiments that tested the impact of initialising the state of different parts of the world’s oceans. The results showed that it was key to initialise the ocean’s sub-surface temperature and salinity (and so density) in the high latitude North Atlantic to have skill in predicting the multi-annual frequency of model tropical Atlantic hurricanes. This is intimately linked to correctly initialising the model’s Atlantic meridional overturning circulation, and to the question of what sub-surface ocean observations would be needed to do this. I have also worked on how external forcings, such as anthropogenic emissions from industrial pollution, may impact regional climate variability.

A lot of the work you have developed focuses on the anthropogenic impact on the Earth’s atmosphere and climate. What does your research tell us about the extent of the impact of human activities on the Earth’s natural systems?

In the last couple of years we have examined the possible impact of anthropogenic aerosol emissions on multi-decadal changes in climate variability. We found that when the latest generation of climate models include the historical inventory of anthropogenic aerosol emissions, they are capable of better reproducing the phases of observed multi-decadal variability in North Atlantic temperatures. In our Met Office Hadley Centre climate model, we find that this is principally due to the inclusion of aerosol-cloud interactions. When aerosols are present in clouds they can modify the cloud droplet size (known as the 1st aerosol indirect effect), increasing the reflectivity of the clouds and hence decreasing the amount of solar radiation reaching the ocean surface. Variations in aerosol emissions from North America and Europe due to socioeconomic changes (e.g. rapid post-war industrialisation in the 1950s and 1960s and then the introduction of clean-air legislation in the 1970s and 1980s) then drive fluctuations in North Atlantic temperatures in our climate model. Furthermore, we also showed that the frequency of model North Atlantic hurricanes is also driven primarily by anthropogenic aerosol changes and that it is in phase with the observed changes in Atlantic hurricane frequency. Further work needs to be done to understand if this aerosol mechanism is truly operating in the real world. If so, then our work suggests a significant role for humans in unwittingly modulating regional climate variability (especially in the North Atlantic) throughout the 20th century. This also has profound implications for the next few decades, as North America and Europe continue to clean-up their industrial aerosol emissions, whilst the impact of short-term increases in aerosol emissions from developing economies (e.g. China and India) also needs to be studied. Of course, at the same time, the signal of greenhouse gas warming is likely to become more dominant with associated climate impacts.

What is your view on having the Anthropocene accepted as a formal geological epoch? Do you think there are scientific grounds to define the Anthropocene in such a way, or at least in what your research area is concerned?

This is an interesting question but not one that I’ve thought very much about! From a climate scientist perspective, I think it is fairly obvious that we have entered a time when the human fingerprint extends to all (or at least very nearly all) environments on Earth. We see the fingerprint in the concentration of greenhouse gases and water vapour in the atmosphere, land and sea-surface temperatures, deep-ocean warming, ocean sea-level rise, ocean acidification, etc… If physical climate changes alone were the main criterion, then surely there would be no doubt that we have entered a new epoch. Beyond this though, the wider Earth biological system is also being impacted by human activity. For example, previous epochs have also been defined based upon mass species extinction, so there may also be a case here for viewing the Anthropocene as a time when the actions of humanity have led to species extinction. Of course there are then questions about how to define the beginning of this new epoch. Many suggest a geophysical marker such as the 1940s and 1950s when radionuclides from nuclear detonations first became present. Or would it be when the atmospheric CO2 concentration started to rise above pre-industrial levels in the early nineteenth century? Or would it be earlier still, when we started significantly altering the land-surface via large-scale deforestation? Then when would the Anthropocene end? Could we envisage a time in the future when we effectively remove our influence on the climate system, e.g. returning the atmospheric constituents to pre-industrial ratios? Or, rather more grimly, would the Anthropocene only truly be over when our species itself becomes extinct? Whilst these are very interesting ‘dinner-table’ type discussions, from a working climate scientist viewpoint the definition seems largely academic and we’d probably be better off investing our time into researching how we are changing the planet and predicting the associated climate impacts!

On a different topic, according to your page on the Met Office website, you started your career in science as an astrophysicist. Could you tell us a bit about how you made the transition from astrophysics to climate science, highlighting any difficulties you may have had with making such a career change and how you overcame them? What advice would you have for young scientists looking to make a similar move?

To a large extent I think ‘science is science’! Many of the skills are very transferable, especially between physical, computationally based, subjects, where numerical modelling skills are essential. I’ve now met a surprising number of climate scientists who are ex-astronomers, or from some other branch of physics. I think what you need most of all is the drive for learning new things, and making new discoveries, about the physical world in which we live. I found that this is very transferable, applying equally to astrophysics and climate science. I think you settle into a subject slowly and even though I’ve been working in climate science for over 6 years now, I still have lots to learn about our existing understanding of climate system, and that’s exciting. The important thing to realise however, is that you can still make important and useful contributions to a new field quite quickly, especially one as broad as climate science, given the right guidance or supervision.

Finally, could you tell us a bit about your future research plans?

We need to progress both our understanding of natural (internal) variability in climate models and improve the fidelity of important climate teleconnections (processes linking variability in one part of the climate system with climate impacts in a remote region). At the same time we need to progress our understanding of the relative roles of external vs internal forcing in driving variability and extremes in the climate system. On the shorter (seasonal) timescales I am interested in what drives the year-to-year variability in the winter North Atlantic Oscillation, which our latest Met Office seasonal climate prediction systems can now predict with surprisingly good skill. Much of this work I hope to develop during my new post as manager of the Global Climate Dynamics group in the Met Office Hadley centre that I will start in December.

 

Interview conducted by Bárbara Ferreira

EGU Media and Communications Manager and GeoQ Chief Editor

 

GeoTalk: Xavier Fettweis

Today in GeoTalk, we’re talking to Xavier Fettweis, an award-winning climate scientist from the University of Liège. He tells us about his work on the Greenland ice sheet.

First, could you introduce yourself and let us know a bit about your current projects at the University of Liège?

As a child, I was already interested by the meteorology thanks to a small weather station installed at my parents’ house. After graduating in 2000 with a Master’s in Mathematics at University of Liège (ULg, Belgium), I continued my training with a post-graduate diploma in climatology and meteorology at the Université Catholique de Louvain (UCL, Belgium). Captivated by climate modelling and the cryosphere, I started a PhD at UCL in 2001, investigating the Greenland ice sheet (GrIS) surface mass balance (SMB) with the help of the regional climate model MAR (for “Modèle Atmosphérique Régional” in French).

Since my return to ULg in 2007, I have been continuing my work on modelling the climate and SMB of Greenland in the Laboratory of Climatology at ULg as a post-doctoral researcher. In 2010, I had the opportunity to join the team of Professor Michiel van den Broeke at Utrecht University, The Netherlands. Working in such a team was very stimulating and a great experience for me. I haven’t yet a fixed position at ULg but I hope to get it soon for continuing my fascinating research about the GrIS here.

Xavier Fettweis out on the ice.

During the EGU General Assembly, you received an Arne Richter Award for Outstanding Young Scientistsfor “fundamental contributions in understanding and quantifying the current and future surface mass balance of the Greenland ice sheet”. Could you summarise your work in this area?

Since the start of my Phd in 2001, I have improved the regional climate model MAR for especially studying the GrIS surface mass balance (in first approximation, the SMB is equivalent to the snowfall-meltwater run-off) at high resolution (25 km). I’ve had the opportunity to build on my PhD research during my post-doc. To calibrate the MAR model over current climate (1958-2012), I used in situ ice sheet measurements  and satellite derived data (melt extent, albedo etc.). I also coupled model results with a circulation-type classification to show the important role of the general atmospheric circulation in the recent melt records (2007-2012) observed over the GrIS. Over time, the MAR model has become one of the reference regional climate models currently available for simulating the GrIS SMB.

The MAR outputs are now used as forcing or comparison in most of the studies about the recent changes in the GrIS SMB and my SMB future projections have been chosen as reference in the ICE2SEA project for forcing ice sheet models (to produce an output that is needed to run other models). These projections will be used in the next IPCC assessment report as evaluation of the sea level rise coming from changes in SMB of the GrIS.

How has atmospheric circulation influenced the recent melt of the Greenland Ice Sheet – what can we expect for the future?

In a recent paper in The Cryosphere, we have shown that about 70% of the recent (1993-2012) surface melt increase over Greenland is a result of changes in general circulation in summer that can be gauged by the North Atlantic Oscillation (NAO) index. The remaining 30% comes from the global temperature increase. Since the end of the 90s, the NAO index has been decreasing in summer (June-July-August). Such negative NAO phases favour warmer and drier conditions than normal over Greenland and mainly explain the recent melt increase.

The general circulation models (GCMs) from the CMIP5 database (that will be used in the next IPCC assessment report) don’t project NAO changes in summer for future. This suggests that the current summers with a negative NAO index are just a result of natural variability. However, the probability of having the current observed succession of summers with negative NAO indices is extremely low with respect to the observed NAO time series starting in the 1820s. This suggests that the GCMs could be wrong.

The Greenland Ice Sheet. Photo Credit: Christine Zenino (source).

Sensitivity experiments with the MAR model using perturbed oceanic conditions (sea surface temperature (SST) and sea ice cover (SIC)) suggest that the current circulation changes in summer could be induced by SST increases and SIC decreases caused by global warming. We could imagine that this positive feedback could be amplified in future, enhancing the melt increase over Greenland and mitigating summer climate changes over Europe. However, more research needs to be carried out to confirm this likely NAO-circulation-SST feedback.

Climate research is a competitive and highly active field – do you have any advice for young scientists about to get going in this area?

With the next IPCC assessment report to be published in the near future (Autumn 2013), every climatologist has wanted to make future projections to include in this report, but before performing future projections we need to better understand the current climate variability. If a model fails to reproduce the current climate, its projections are lacking in reliability, particularly because the response of the climate to warming is not linear.

Most of the general circulation models (GCMs) fail to simulate the current climate variability because the response of the climate to a greenhouse gas increase is far away to be obvious as expected several years ago. Therefore, a lot of progresses still needs to be done to develop our understanding of current climate variability. With a better training in IT than before, young scientists could significantly improve climate modelling.

What do you hope to work on in the future?

I would like to continue my research on the Greenland ice sheet at ULg. Studying GrIS is enthralling because a lot of things have been changing since the 1990’s. The Greenland community is small, but very much like a family and, after several years in this field, many of them have become friends as well as collaborators.

With the help of my small team at ULg, I plan to also apply the MAR model to the other ice sheets, such as Antarctica, Svalbard, Ellesmere, and to couple MAR with ice sheet models for being able to simulate the total ice sheet mass balance (not only the surface mass balance).

I have an only regret: I have never been to Greenland because modelling does not need to be on the field but I hope to have the chance soon to walk over the ice sheet.

If you’d like to suggest a scientist for an interview, please contact Sara Mynott.

Geosciences Column: Rainfall and Climate – a Dynamic Problem

“Rain is grace; rain is the sky descending to the earth; without rain, there would be no life.” – John Updike

Rain quenches the thirst of soils and vegetation, fuelling ecosystems and much of the world’s agriculture. Whether it ruins a day on the beach or destroys a season’s harvest, it makes humans deeply aware of their vulnerability to the vagaries of the atmosphere. It’s important to understand how rainfall changes in a changing climate. Here, I will describe the issues in understanding precipitation changes and how two recent papers help to solve the puzzle.

Predicting rainfall is difficult. It is a small-scale phenomenon, especially in the towers of convective cloud in the Tropics. Weather forecasting models are just beginning to capture them properly at scales of a kilometre or so, but climate models, which have to be run for decades rather than days, calculate atmospheric conditions on scales of hundreds of kilometres. Rainfall has to be simplified in these models, since we cannot calculate the physical properties of individual clouds. These simplified representations are called parameterisations. A precipitation parameterisation relates the average rainfall over a large area to the average amount of water in the air. Different models do this in different ways and, because it’s a simplification, there is no definitive ‘right’ way. This means there is some disagreement among climate models about how rainfall will change in the future, especially in the Tropics (areas on the figure which are not stippled).

Climate model projections of precipitation change in a future with high greenhouse gas emissions. Left: current generation of models, Right: previous generation of models (around 2005). Top: December-February, Bottom: June-August. Stippling shows areas where models largely agree. White areas show complete disagreement among models (source: Knutti & Sedlacek, 2013).

If we think about precipitation in general theoretical terms, we can find laws which must be followed and use them to make predictions, as Issac Held & Brian Soden did in their study of how the hydrological cycle responds to global warming. Rain is caused by the upward transport of water vapour from the surface into the atmosphere, where it condenses, forms clouds and rains out. The amount of moisture going up must, of course, balance the amount coming back down as rain.

As the climate warms, the amount of water vapour a fixed mass of air can hold increases. This means that, as long as the circulations transporting water upwards remain the same, the total amount of water vapour going upwards must increase – which means the amount of rain coming down must also increase. This is called the ‘rich get richer’ mechanism, because it increases rainfall in regions where there is already a lot of rain driven by upward moisture transport. It’s a fundamental mechanism driven by thermodynamic laws…but that doesn’t mean it’s the only thing going on.

Convective raincloud in tropical Africa (photo credit: Jeff Attaway).

If climate model projections followed the ‘rich get richer’ mechanism, precipitation would increase most in the regions with the most precipitation currently. In fact it is more complicated than that. Robin Chadwick and his colleagues explored the effect of weaker vertical motions in a warmer climate. We can understand this by thinking about what carbon dioxide does to the vertical temperature profile. It warms the mid-troposphere (about 5 km up) more than the surface. To get convective upward motion, the air at the surface must be less dense (i.e. warmer) than the air above. Warming the air aloft suppresses this motion. The Chadwick decomposition calculates the part of the precipitation changes caused by changes in moisture (which goes at about 7% per K) and the part caused by the reduction in upward transport. They find the two tend to roughly cancel each other out, which means the spatial shifts in precipitation are determined by changing patterns of surface temperature (since warm surfaces produce upward motion).

Sandrine Bony and her team decompose precipitation changes into two main components rather than three: one is the ‘dynamical’ component, associated with changing upward motions, and the other is the ‘thermodynamical’ component, including changes in atmospheric moisture content. Unlike the Chadwick method, the thermodynamical component is not designed solely to represent the ‘rich get richer’ mechanism. This means the thermodynamical component isn’t just a 7% per K increase; it includes things like the spatial changes in surface temperature. The dynamical component isolates the change in precipitation caused by changes in upward motion.

Monsoon raincloud over a lake in the Tibetan Plateau (photo credit: Janneke Ijmker).

The ‘rich get richer’ rule of thumb becomes increasingly irrelevant at smaller scales. This is frustrating, because these are the scales we really care about! It’s not particularly useful knowing what will happen in a general sense over the whole Tropical region. Farmers want to know what will happen to the seasonal rains on their small piece of land.

Bony also points out that geoengineering schemes which aim to reduce incoming solar radiation to cool the planet’s surface would leave the dynamical component of precipitation change untouched. This is because the dynamical component is caused by the warming of the mid-troposphere by carbon dioxide, and this remains even if we cool the surface. It is an example of the inexact nature of the cancellation between carbon dioxide increases and geoengineering schemes to decrease the amount of carbon dioxide in the atmosphere, and demonstrates that the only way to stop carbon dioxide-driven climate change properly is to stop emitting carbon dioxide.

Bony and Chadwick’s decompositions show how one can glean a lot more information from climate model projections than one would expect from first glance. We have established some general facts about climate change related to the Earth’s energy budget. In that sense we understand quite well what will happen in a warming climate. However, there is still a lot of diversity between model projections, most of which comes from differences in the dynamical response. Local changes in rainfall are related to changes in circulation, and this is the area in which a lot more work needs to be done.

By Angus Ferraro, PhD student at Reading University

References:

Bony, Sandrine, Gilles Bellon, Daniel Klocke, Steven Sherwood, Solange Fermepin & Sébastien Denvil, 2013: Robust direct effect of carbon dioxide on tropical circulation and regional precipitation, Nat. Geosci., doi:10.1038/ngeo1799

Chadwick, Robin, Ian Boutle & Gill Martin, 2013: Spatial Patterns of Precipitation Change in CMIP5: Why the Rich don’t get Richer in the Tropics. J. Climate, doi: 10.1175/JCLI-D-12-00543.1

Held, Isaac M., Brian J. Soden, 2006: Robust Responses of the Hydrological Cycle to Global Warming. J. Climate, 19, 5686–5699. doi: 10.1175/JCLI3990.1

Knutti, Reto & Jan Sedláček, 2013: Robustness and uncertainties in the new CMIP5 climate model projections, Nat. Clim. Change, doi: 10.1038/nclimate1716